The Creality Sermoon P1 is a handheld, standalone 3D scanner that packs a long list of features in a single enclosure that can comfortably fit in your hands. Blue laser scanning allows for high-resolution captures using either a 22, 7, or single laser mode, and the NIR (near-infrared) mode can capture geometry with no markers in a variety of lighting conditions. The bright 6-inch LCD screen allows users to align, edit, and mesh 3D scans right on the device, with no computer required. At $3,129 for the complete bundle, the Sermoon P1 is more expensive than most consumer 3D scanners, but the impressive results will help justify the cost for professional users.

The Sermoon P1 excelled at creating detailed 3D models during testing and performed impressively when scanning indoors, outdoors, both with and without tracking markers. Complete 3D scans can be captured, edited, and even shared directly from the device, while maintaining high scan quality. For users who want more control, the CrealityScan app allows for precision editing on a Windows or macOS machine.

Specifications: Creality Sermoon P1

The Sermoon P1 offers multiple scanning modes that can be used with the scanner in standalone, wired, or wireless mode. As a result, the specifications can vary depending on the specific combination of settings and configuration selected by the user. The spec sheet below is a general guide, but be sure to read through the full list of specifications on the Creality site if you are looking for the full breakdown.

Included in the Box: Creality Sermoon P1

The Creality Sermoon P1 arrived in retail-ready packaging that would look right at home on the shelf of a Micro Center or Best Buy. The Sermoon P1 is sold as a bundle, which includes a carrying case, a glass calibration board, the 3D scanner, a power supply, a USB-C cable, an extra battery, and a set of reflective marker dots. The carrying case is sturdy and well-made, and has enough room to comfortably fit all the included accessories and everything you need to scan on the go.

Design of the Creality Sermoon P1

The Creality Sermoon P1 is essentially a 3D scanner with a built-in computer, making the LCD more than just a simple source of feedback during the scanning process. Users can create, modify, and even export meshed data directly from the unit, no computer required. The 6-inch LCD touchscreen on the rear of the unit is bright and responsive and behaves similarly to the LCD on a cell phone, with a touch-and-drag menu bringing up typical settings like screen brightness, Wi-Fi networks, etc.

The rubber grips on the side of the unit and included wrist cord highlight the mobile nature of this scanner, which has clearly been designed as a mobile counterpart to the Sermoon S1, which requires a constant connection to a computer to operate. The Sermoon P1 itself has an advertised weight of 830 grams and feels comfortable to hold with both hands. The included wrist strap is a low-cost insurance policy against dropping the Sermoon P1 during a scan, and the scanner has a ¼ thread on the base of the unit that allows it to be used with a tripod or other stand.

The Sermoon P1 includes an extra battery, which is a welcome addition given the mobile-first nature of the scanner. The battery has a USB-C port right on the side, so it can be charged by either inserting it into the scanner and charging or by plugging directly into the included power supply. Creality claims the battery can be charged to 80% in ~50 minutes, and I was able to charge from about 20% to 100% in under an hour for both batteries. For anyone used to shooting video with a battery-powered DSLR camera, running out of power while in the field can be a tough situation, and it’s a smart move by Creality to include the extra hot-swappable battery.

Most consumer scanners like the 3DMakerPro Mole are handheld units which are tethered to a computer during use, limiting the user in where they can effectively create a 3D scan. By integrating the processing capabilities of the Creality scanning software directly into the unit, the Sermoon P1 can be tossed in the included carrying case and brought anywhere in the world to capture and process 3D scans, no computer required.

My only concern with the design of the Sermoon P1 is the enclosure around the sensors (laser, camera, and LEDs). The perimeter around the lens is nearly flush with the glass, meaning the clear protective layer could be scratched or damaged if laid down flat. Most cell phone cases include a small rim that extends past the “camera bump” to prevent this type of damage. Examples of this type of case can be found on sites like Makerworld, such as an enclosure for the Creality Otter 3D scanner designed to protect the lenses.

Getting Started and Calibrating the Creality Sermoon P1

When I turned on the Creality Sermoon P1 for the first time, a video played that showed the full scanning process with some helpful tips on preparing a model for scanning. I was surprised to discover the scanner had an onboard speaker, which played an audio track with narration during the scanning demo. While experienced users will likely skip through this video, it’s a nice addition for first-time users who are new to 3D scanning. During most operations, the Sermoon P1 will also show tips related to scanning while the operation is processing, a nice way to help users pass the loading time.

The Creality Sermoon P1 includes a glass calibration board that is used to calibrate the scanner before use. The calibration process involves laying the glass board down on a flat surface and gradually moving the scanner around in a controlled manner. This process takes about 10 minutes, and can be a bit tricky so moving slowly and gradually is the best move here.

Installing the Software for Creality Sermoon P1

Creality offers a first-party software for the Sermoon P1 called CrealityScan that is compatible with both Windows and macOS. This software offers essentially the same workflow steps as the onboard software on the Sermoon P1: scan data capture, point cloud fusion, meshing, and texture generation. When the Sermoon P1 is connected to this software via Wi-Fi or a USB cable, the LCD on the rear of the scanner is disabled. A nice touch is the “User Guide” section, which contains helpful tutorials and best practices for various skills, as well as general information about 3D scanning and mesh cleanup.

I’m using CrealityScan 4.1.7 on macOS 26.3 for this review, using a 2021 M1 Max MacBook Pro with 64 GB of RAM. With this computer, the scanning frame rate in NIR mode is up to 30 FPS using a wired connection, a substantial increase over the 18 FPS frame rate in standalone mode. This improved performance comes at a price: the Sermoon P1 needs to be tethered to the computer with a USB cable during scanning and the rear screen is disabled during scanning.

The workflow for the desktop CrealityScan app is the same as the standalone scanner: alignment, meshing, mesh editing, and exporting. The operations are laid out in a linear sequence, making it easy to follow as you go through the editing phase. CrealityScan also has a “One-Click Process” which will go through all the steps automatically, without any adjustment or input needed from the user. I found this to work pretty well for simple objects, but for more complex models with lots of geometry or holes it would be better to use the individual operations with more granular control.

Blue Laser Line 3D Scanning with the Creality Sermoon P1

The Creality Sermoon P1 uses blue laser line scanning for detailed geometry capture, and offers three different blue laser scanning modes: 22 crossed laser lines, 7 parallel laser lines, and 1 single laser line. The 22 crossed laser line mode is intended for fast scanning on large objects, the 7 laser line mode is best for capturing fine details, and the single laser line is intended for scanning deep holes or recesses that would be difficult to capture with a NIR or photogrammetry approach.

Automotive applications are a common use of high-resolution 3D scanners: products like aftermarket covers, clips, and other accessories require a quality model to build from, something that automotive OEMs typically don’t publish. An example is this carbon fiber window switch panel on a Mazda CX-3: it looks simple but the compound curve may be difficult to intuitively 3D model without a scan or model to reference. Using the included reflective markers, I placed about a dozen of them at random intervals and switched the Sermoon P1 to laser mode.

For my first pass, I used the 22 crossed laser line mode to capture the bulk of the geometry. This mode is much faster than I was expecting, with the scanner operating at 53 FPS and the geometry filling in almost immediately. After scanning the entire area, I did a second pass using the 7 laser line mode to capture some of the fine detail around the transition between the door panel and the switch cover.

In the center of the panel is a recess where the rocker switch for the window is located. Neither the 22 or 7 laser line mode was able to penetrate all the way to the bottom of this recess, so I switched to the single laser mode which is intended specifically for this type of application. Using the single laser, I worked slowly and carefully and was able to capture the complete cavity. The Sermoon P1 has a “remove tracking markers” function, which will remove the markers before creating the final mesh.

Scanning this switch panel was an eye-opening experience, with the Sermoon P1 surpassing my expectations at every turn. The 22 crossed laser line mode captured a large amount of geometry quickly, and switching between this and the other laser line modes was a seamless experience to complete the scan. From here, bringing the model into a traditional CAD or surfacing program would be the next step to create a parametric solid model to build from.

I’ve tried to scan the badges on the Orange amplifier a few times, but I’ve never had much success. After placing a few tracking dots on the front of the grille of the amp, I used the 7 laser line mode to capture both the bottom and top badges. The Sermoon P1 averaged around 41 FPS during the scan and despite the repeating pattern of the grille in the background, the scanner didn’t lose tracking or get confused when moving it manually.

The bottom badge looked sharp and detailed, with only a small amount of distortion around the crisp edges. The top badge had the majority of the detail present, but the sub-millimeter text on the bottom of the badge didn’t quite resolve. For a quick pass without any scanning spray or other preparation, I was happy with the result and felt confident the results could be improved further with some time spent adjusting the scan parameters.

The laser line scanning mode also works when tracking markers are placed on a turntable, with a marker-free object placed in the middle. I scanned a Tom Sachs Nike GPS sneaker by rotating the turntable as well as moving the Sermoon P1 around the model during the scan. My goal was to capture the text on the bottom of the outsole, as well as aligning multiple scans to create a full model of the sneaker. I used the 7 laser mode and scanned the sneaker three times: once with the outsole facing down, and then twice more to capture each side.

When rotating a model manually (flipping it over, etc.) during the scanning process, the resulting scans need to be aligned in space before the point clouds can be fused. This is typically a difficult process, with most auto-alignment processes picking the wrong features to use as landmarks or just failing to align properly. I used the Automatic alignment on the Sermoon P1, and was impressed to see it perfectly aligned three scans in under a minute.

Scanned with the 7 laser line mode, the text on the bottom of the shoe is clearly legible, and the wear on certain parts of the outsole can be clearly identified. The fabric also showed a good amount of detail, with stitching on the laces visible in the 3D model. I wasn’t expecting such a fast alignment process, and it’s even more impressive that it is completed on the device.

3D Scanning Outdoors with the Creality Sermoon P1

The standalone mode of the Creality Sermoon P1 allows users to bring the scanner to places that would be difficult or impossible if they also needed a computer, internet connection, and power supply. Scanners like this typically work best on an overcast day where the amount of directional light is minimal and the overall lighting on objects outside is uniform. I brought the Sermoon out with me on a day like this and captured a few objects out in the wild to test it out.

The Sermoon P1 has features that are designed explicitly for this type of scan, and I was impressed with how detailed the captured model was. Using the “Large (500 - 2000 mm)” mode with the tracking set to “Geometry”, I scanned a large brick sculpture using the NIR structured light mode. When scanning such a large object, the frame rate dropped to around 3 FPS, which meant the tracking would occasionally drop when I walked around a corner or zoomed in to capture a small detail.

The geometry of the scan is impressive: accurate, sharp, and even very small tool marks on the edges of the bricks are picked up and represented in the meshed model. This high resolution comes at a price: this model in a meshed form takes up about 3 GB of data on the device, and the exported model was 416.7 MB after cleanup. The Sermoon P1 ships with 256 GB of storage, which is enough to capture dozens of large objects like this without running out of space. If storage runs low, models can be exported to the CrealityScan desktop app to free up space on the device.

Unfortunately, the color texture wasn’t as sharp as the geometry from this scan and had several major issues. There is a clear line around the middle of the sculpture where the brick texture seems to split and diverge. Without an intuitive way to remap the texture on the device and no controls to adjust it, the only option for users here would be to export the model and manually edit the UV mapped texture. There are a few areas where shadows seem to start and stop abruptly, which is caused by the color texture not blending between scans (more on that in the next section).

Scanning sculptures made from bronze can be a challenge, especially when they are located outdoors. There is a wide range between the darkest areas (typically indentations or recesses) and the lightest areas (surfaces reflecting direct sunlight), and the geometry can be hard to capture because of this variance. This is typically a job for photogrammetry, but I wanted to test the Sermoon P1’s ability to successfully scan a bronze sculpture. I attempted to scan the head of a sculpture located in an urban plaza, and was impressed with the results.

Using the default scanning resolution for NIR standalone mode, I saw an average of between 14 to 18 FPS during the capture and captured 1,410 frames and 60,762 points. The scanner handled the bright spots around the sculpture incredibly well, and the tracking didn’t drop even when light was reflected directly into the lens. Because I was only interested in capturing the head of the sculpture, I worked my way around in a circle to capture under the chin, behind the ears, etc. and was able to capture everything I needed in only a few minutes.

I used the integrated point selection tools to isolate the area that I wanted to keep and deleted the rest, which only took a few seconds of processing time. From the point cloud, I was able to fuse, mesh, and add a color texture all from the standalone Sermoon P1 unit, no computer required. The entire post-processing workflow took about 20 minutes; a truly impressive feat for a handheld scanner. The color texture was a little blotchy due to the reflective nature of the material, but the geometry was crisp and accurate to the model that I scanned.

3D Scanning Color Textures with the Creality Sermoon P1

In 3D modeling, “color texture” typically refers to a UV mapped image that is a separate file from the geometry of the model. For instance: the scan of the toad above is composed of two primary pieces of data: a 3D model containing all the geometry in the model, and an image that contains information on how it is UV mapped (or wrapped).

Most users of the Creality Sermoon P1 will likely be using it for reverse engineering, geometry capture, or other applications where color texture isn’t a concern. For anyone interested in also capturing color data, Creality offers a “Texture” operation that allows NIR scans to have a color texture applied to the scan data.

The Sermoon P1 uses an onboard Qualcomm ISP chip, RGB camera, and integrated lighting to capture color textures on a model during scanning. These textures can be baked on a model after the meshing step, but CrealityScan offers limited adjustable parameters and instead offers a one-click texture function. Using this texture function will allow a user to export an .obj file with an associated color texture.

The result is underwhelming, and the color textures have a patchwork appearance where the seams form between scans. Most photogrammetry software will typically blend textures by finding edges or transitions and gently interpolating or averaging between them to prevent hard edges or seams from forming. The Sermoon P1 does not have any adjustable parameters for texture generation, so if the result isn’t acceptable, the user doesn’t have many options for fixing or improving it.

Despite the poor color texture, the geometry from the NIR scan is still impressive. There were only a few small holes around some of the deeper recesses, and the software was easily able to identify and fill them without disrupting the surface of the mesh. Exporting the .stl from CrealityScan resulted in a manifold (watertight) file that was 50 MB in size and was immediately ready for 3D printing.

Printed on a Bambu Lab X1-Carbon, this toad model looks and feels identical to the original. All of the geometry (bumps, ridges, grooves, etc.) of the original sculpture are present, and the scaling appears to be very accurate to the original model.

Bottom Line

The Creality Sermoon P1 is a standalone 3D scanner capable of capturing small objects, large objects, and everything in between without requiring extensive preparation work or CAD editing. The onboard software can isolate specific areas of a scan, align multiple scans, and generate mesh files thanks to the 8-core 3.36 GHz CPU and 24 GB of RAM. The three blue laser modes all have unique use cases, and can be combined to quickly scan a large surface and later go back to capture a deep hole or fine geometric texture.

The Sermoon P1 is available as a bundle for $3,129 direct from Creality, which includes the scanner as well as all the accessories required to bring it out into the field. The price of this bundle will keep it out of the range of most hobbyists and users looking to start exploring 3D scanning, especially with less-expensive units like the 3DMakerPro Mole available for under $400. However, for professionals with an application-specific use case who are willing to pay for results, the Sermoon P1 is a compelling choice.

The Sermoon P1 created accurate scans in both laser and NIR modes, and the tracking was impressive in a variety of controlled and uncontrolled environments. The mesh editing (smoothing, hole filling, etc.) worked well both on-device as well as using the CrealityScan macOS app, and I didn’t see a major difference between using the two. The color textures were disappointing given the price point, and that shortcoming is the only major critical point I found when using the Sermoon P1.

ASML, maker of the world's most advanced chipmaking tools, and the Van Gogh museum have collaborated on the design of a 3D scanner to capture and create digital twins of the masterpieces of the Netherlands’ most famous artist. Vincent van Gogh’s artistic brushwork was characteristic for its ‘Impasto’ style, where paint can be applied in a very thick and textured style. Thus, the ‘Impasto Project’ was born, with ASML and the museum creators developing a 3D scanner that can capture exceptional levels of artistic detail in approximately 100 gigabytes of data per painting.

The curators of the Van Gogh Museum had a dream project, where they could create something akin to the ‘Google Maps for paintings.’ They thought it was worth striking up a conversation with local tech titan ASML and got lucky, as the leading chip toolmaker develops technologies that are perhaps surprisingly applicable to the detailed 3D scanning of paintings.

When ASML was initially approached in 2017, the partnership might have sounded unusual. However, both van Gogh and ASML are Dutch icons, with their roots in the province of North Brabant. Moreover, “at ASML we don't just make chip machines,” explained (machine translation) Head of Research Metrology at ASML, Maarten Voncken. “We also make equipment to measure how accurate those machines are.” This technology turned out to be surprisingly applicable for 3D scanning paintings.

As mentioned above, the pilot program began in 2017 and, after a promising start, the collaboration was extended in 2019. Work has continued apace, and the partners will now collaborate on the Impasto Project until at least 2028.

How the scanner works

The Impasto Project hardware is essentially a small, shoebox-sized scanner guided by a framed system with rails. This imaging process takes photos of dozens of segments of each painting. Each specific area photo is taken four times with the camera at slightly different angles.

This varied data representing the same image segment allows the custom software to calculate the varied thickness of the paint from shadows that are cast. As we mentioned in the intro, a finished 3D image scan of a painting is approximately 100 GB.

ASML measuring equipment is accurate to the nanometer level, so this 3D scanning tool operates at a scale approximately 100,000 times larger. Nevertheless, significant customization of the hardware and software was required to meet the requirements of the museum's curators.

In addition to the digital twin plans, the scanner can also be used to automate the process of inspecting paintings before and after they are lent to other institutions. Thanks to this new tool, curators can check the effects of handling, temperature, and moisture changes when paintings are loaned. Observations will likely be used to improve the handling and transport of these treasured works.

Giving back

Interestingly, ASML is nonchalant about the investment in non-core areas, such as the Impasto Project. It shrugs off the cost of the project at ‘a few million a year.’ It also revealed that around 10 to 15 ASML staffers work on this project. We aren’t familiar with Dutch business practices, but this project seems to be part of a wider commitment to ‘give back’ to the region. Currently, ASML invests around $100 million in local employment projects as part of its ‘social budget.’ ASML also sponsors a gallery at the museum.

With the ASML collaboration established and 3D scanning becoming routine, visitors to the Van Gogh Museum can now visit a new science center within the museum. Here, they will find the 3D scanning equipment on display, among other technical art curation-related things.

Outside interest

The 3DMakerPro Mole offers 0.05mm high accuracy 3D scanning at 10 frames per second in a form factor that fits nicely in the palm of your hand. Offered in multiple packages starting at just $519 and intended for marker-free scanning applications, the Mole is capable of creating scans of models on a turntable or dynamically moving around a subject. The JMStudio software can align multiple 3D scans before processing to allow users to scan models in several stages, or scan highly detailed areas and combine them with larger, lower-resolution areas.

The Mole worked generally well during testing, although the software experienced several crashes and some models were surprisingly difficult to capture completely. JMStudio features a linear workflow for most processing, and as a result there are fewer levers to adjust. This simplified workflow works well for the less-expensive 3DMakerPro Lynx, but power users may feel limited by the lack of available settings and professionals may require more granular control to unlock the benefits of the higher accuracy Mole. 

Specifications of 3DMakerPro Mole

Included in the Box: 3DMakerPro Mole

The 3DMakerPro Mole premium combo includes: the Mole scanner, the data transfer cable, a power supply with several global adapters included, a tripod, a carbon fiber handle, a turntable with USB cable, a plate which extends the diameter of the turntable, a user manual, and a carrying case that fits everything neatly inside. Considering the size and shape of the Mole 3D scanner, it’s reasonable to assume that this scanner will appeal to users interested in taking the scanner with them to job sites or scanning in the field, making the carrying case a welcome addition. 

The Mole itself weighs 390 grams, similar to the weight of a 12 oz. can of soda. The scanner has a rubber overmolded grip that wraps around the sides and rear, which makes it comfortable to hold as well as giving it a professional appearance. The camera / LED array is located on the front of the scanner, and the general shape makes it easy to maneuver and control during scanning.

Package Variations of 3DMakerPro Mole

Like other 3DMakerPro 3D scanners, the Mole is available in three combo packages: standard, premium, and luxury. The standard combo includes the basics needed to complete a 3D scan, including the Mole 3D scanner, a power cord, and a data cable. For an additional cost, the premium combo also includes a turntable, tripod, and carbon fiber handle while the luxury combo includes a “color kit.”

The premium combo hits a sweet spot between price and functionality, and the modest increase in price over the base package will be interesting to anyone without a tripod or turntable. Meanwhile, the complexity and lack of reviews or examples of the color textures captured by the luxury package will make that a less appealing purchase.

Setup of 3DMakerPro Mole

Unlike the 3DMakerPro Lynx, which uses a surprisingly complex wiring harness for connecting the scanner to a computer and other accessories, the Mole uses only a single input terminal to provide power and data transfer. This terminal plugs into the Mole, and the other ends connect to the power adapter as well as the USB plug on a computer. The cable has a large warning sticker applied to it and requires that two red dots (one on the cable, one on the Mole) be aligned when plugging in the scanner. The cable has a quick-release sleeve with a tether attached to it which makes removing the cable a fast process, but can be tricky to remove while the Mole is attached to the ball swivel mount.

In addition to the ball swivel mount, the Mole includes a tripod and a lightweight carbon fiber handle which can be combined or separated depending on the application. I found the body of the Mole was easy to handle for handheld dynamic scanning, but the included carbon fiber handle is a nice addition for anyone with smaller hands or who is used to a more traditional grip.

The Mole has a vent engraved on the top of the scanner in the shape of the 3DMakerPro logo, which allows for airflow and keeps the scanner from overheating during use. Considering the scanner is almost always held with the top facing away from the operator during desktop use, this is a nice touch that keeps the scanner cool and also doesn’t vent the hot air in the direction of the user.

Installing the Software for 3DMakerPro Mole

The companion software for the 3DMakerPro Mole is called JMStudio, and is the same software used by 3DMakerPro Lynx, the larger-format version of the Mole. Available for both Mac and PC, JMStudio combines data capture with mesh editing, offering full functionality for users who can’t or don’t use additional mesh post-processing software like Blender, Rhino, or others. This software allows users to capture multiple scans, align them to create a full model, remove duplicate and isolated (noise) points, and even repair holes before exporting a mesh for printing or further processing.

Using the latest build available for Mac (v2.2.8), the program crashed surprisingly often during the Remove Noise step of post-processing, and I found myself going back and manually removing noise points from the scans before attempting alignment. While this wasn’t a major issue and I never lost any work other than the work in progress, it’s still worth noting and could be a potential headache for users processing complex scans on a tight deadline.

3D Scanning with the 3DMakerPro Mole

JMStudio offers two ways to initialize a 3D scan: Easy Scan and Table Scan. Easy Scan is the quickest and easiest way to create a scan; simply point the scanner at your subject and start moving around to capture geometry. Table Scan is a more automated process that uses the included turntable to rotate the model a full revolution while the scan is being captured. This mode will capture a total of 320 frames before stopping automatically and allowing the user to continue to processing or appending with an additional scan.

Table Scan first initializes the scene by detecting the surface of the turntable, which will later be automatically removed from the scan data. Once the turntable turns red in the interface, this initialization step is complete and the scan data can be acquired. Because the turntable only rotates in a single axis, there will usually be geometry that is not captured by scan. It’s helpful to visualize “painting” a model with a flashlight; anywhere the light strikes the model will be captured but overhangs or sharp curves typically require rotating the model on the turntable to capture.

To fully capture a model like this toad, multiple scans are required to fully capture the geometry of the model with no holes. I scanned the model a total of three times in three different configurations to capture as much of the geometry as I could for later reconstruction. Each scan is only a partial representation of the model, but they can be combined to create the finished model.

Because the model was rotated into a different position, the scans overlap and collide in 3D space. In order to process the scan correctly, JMStudio can automatically (or manually) align a model by combining the scans together so the various features are all in alignment. After running an automatic alignment, the toad model is correctly positioned and the relative position of the turntable surface geometry is adjusted and shown in red on the aligned scan.

After the alignment is completed, the model is processed in a linear workflow through the steps selected in the Process window. After Fusion and Remove Noises is complete, the model can be repaired and any holes or gaps are automatically filled. If processed automatically, this workflow will sometimes take outlier noise points and connect them to the rest of the mesh, resulting in a small protrusion in the model. JMStudio also offers basic mesh editing capabilities, so the misplaced triangles can be selected, deleted, and the resulting hole can be repaired.

The final product is a mesh 3D model that has been made from the three individual point cloud scans, after merging the clouds, removing isolated points, and filling holes. For very large or very complicated models, the JMStudio app also provides a Simplify operation, which will reduce the number of triangles in the model to reduce the overall file size. This operation worked OK in testing, but using a decimate modifier is also a popular step in post-processing software like Blender or my own browser-based decimation tool lowpoly3d.xyz.

3D Scanning Dark Objects with 3DMakerPro Mole

3D scanners can struggle to scan dark objects due to the way the light is projected and interpreted by the camera. 3DMakerPro advertises the Mole as being able to clearly scan dark models, a feature that would give the Mole a competitive advantage over others in the market. I was curious about this claim, as the similarly-featured 3DMakerPro Lynx struggled to capture black surfaces.

To test this feature, I scanned a black bicycle saddle that has a dull, satin finish and a pattern of shiny black dots. This saddle has several compound curves that make it ideal for 3D scanning, and the dull finish doesn’t require any pre-processing or spraying. The Mole scans at 10 FPS, which is fast enough to capture surfaces while allowing you to reposition the scanner manually during the scan without losing tracking on most wide, flat surfaces.

However, the Mole struggled when scanning thin parts on Table Scan mode and would lose tracking when the narrowest part of the scanned object was directly facing the scanner. I used Easy Scan mode on Fine Detail to capture as much of the surface as clearly as I could. Even in a well-lit room, I had to adjust the brightness and sensitivity of the scanner to the maximum settings to capture the dark saddle against a white background.

The saddle was scanned a total of four times: two scans of the top surface and sides and two scans of the bottom surface and sides. The top surface appears nearly flawless in the 3D scan, but the glossy appearance of the rendered model in JMStudio makes it difficult to see if the repeating dot texture on the surface was captured as well. The bottom scan looks generally good, but even with a large amount of overlap I was not able to automatically align the scans in the software and the manual alignment wasn’t much better.

The appearance of the surface is undeniably impressive for such a dark model that took only a few minutes to capture and a few more minutes to process. The repeating dot texture became visible when viewing the 3D model in Blender and shows an overall impressive level of accuracy to the original model. The overall quality and consistency of the scan would be enough for most reverse-engineering tasks or for understanding model geometry for creating accessories or mounts.

The base model 3DMakerPro Mole is compatible with macOS and PC for capturing and processing scans, while use with a mobile phone requires the optional Connect accessory. This accessory is currently offered at $89 for Android and $159 for iOS and still requires a PC or Mac for processing scans, which is worth noting if you plan on using the Mole with a mobile phone. 

Using the Connect involves transferring scans between the scanner and a computer using a private Wi-Fi network with firewall disabled. The official 3DMakerPro instructional video on YouTube has multiple negative comments from frustrated users, indicating this process may be more complex than expected.

3D Scanning Detailed Models with 3DMakerPro Mole

The 3DMakerPro Mole is advertised as featuring an accuracy of 0.05mm, or 50 microns. This sub-human-hair-width accuracy should lend itself to scanning highly detailed models, but I found the Mole struggled to capture fine feature details on models as well as fine details on surfaces. Using a light colored Gundam model that had been sprayed with a matte clear coat, I tested the Mole on Fine Quality mode to see how much detail could be captured.

Unfortunately, the Mole failed to pick up much of the detail on the model, and the dull gold components (hands, elbows, knees, etc.) were skipped entirely. Even when scanning one piece of the model at a time to try and merge them together, I was unable to capture any usable amount of detail. This was a disappointing test considering the advertised resolution and intricate models shown as examples of products the Mole is able to scan. I was able to capture the rough form of the model, which may be the only thing some users are interested in. For anyone interested in reverse engineering small parts like this, the Mole will likely struggle to provide a quality output.

Scanning a highly detailed metal model of an old Jeep led to a similar outcome. This model has a dull green finish with a few components being painted in silver, red, or black. Scanned in Fine Quality, this model was captured a total of 4 times in different positions and then automatically aligned and processed.

The Mole captured the general shape of the model, but still required extensive manual cleaning to get to a semi-complete finished product. The blank area where the windshield would be was filled with floating points, which had to be selected and deleted from each scan before merging to prevent the area from being filled in. The hood, tires, and bumper look good, but some of the darker areas in the backseat were not fully captured.

The power of software like JMStudio is the ability to capture a model that has been propped up at an angle and then manually delete the object propping it up. For example, this lizard model has a lot of surface detail on the top of the model, but it may not be important to capture the detail on the back of the model. I scanned this model twice, once with it lying flat on the turntable, and once with it propped up at an angle so I could get additional detail on the very top surface of the model with text.

In JMStudio, I’m able to delete all of the geometry related to the bracket I used to prop up the model, and only merge the surface scan with the scan of the full model. After fusing and cleaning up the point cloud, the mesh model shows a high level of detail, including the excess paint in the top “B” of “Barcelona” that partially fills the recess. 

3D Scanning Textures with the 3DMakerPro Mole

In 3D modeling, “texture” typically refers to a color image of a 3D model, which is projected onto a 3D model using a process called UV mapping. This is different from “geometric texture”, which is a 3D texture modeled directly on the surface of a model (think diamond plate, brick wall, etc.). Color textures are useful because detailed images can be projected onto a simple surface, like a colorful image of a poster mapped to a square composed of two triangles. 

The 3DMakerPro Mole is capable of capturing texture during a scan, but the version I was sent can only capture textures in grayscale, meaning the silver beetle with green wings that I scanned will be captured in shades of gray and not in full color.

When scanning in Texture Mode, the LED array fires with each frame at full intensity and can be blinding if faced directly. Considering the 3DMakerPro site shows a Mole scanning inches away from the face of a child, this is something to carefully consider if you are planning to scan faces with texture using the Mole.

When creating a scan in JMStudio, toggling from “Geometry” to “Texture” will enable texture capture, something that can be toggled on and off in the viewer. The grayscale texture is applied to the model in the “TextureMapping” (sic) step and is exported as a UV mapped 2D image when saving the model as an .obj file. 

Importing the textured model into Blender allows for a clear view of the texture, which shows a decent amount of detail despite the grayscale color palette. Capturing a grayscale texture on a 3D model feels like a half-measure, and I’m not entirely sure what use the texture would have to most users. Digital sculptors could use the grayscale texture to perform a displacement offset and add additional details, and it’s possible the UV map could have color added back in an additional post-processing step using software like Photoshop. For most users, the greyscale texture likely won’t add much value and will simply add additional processing time.

On the 3DMakerPro site, the Mole is advertised as featuring “Ultra-Clear Texture Mode” when using the optional color kit included with the luxury package. The version shipped to me did not include this package, so I’m not able to test the quality of the color textures.

Bottom Line

The 3DMakerPro Mole is a powerful handheld 3D scanner that is capable of making 3D models from real-life objects with virtually no set-up, calibration, or additional post-processing required. I was impressed with the ease-of-use of the JMStudio software, although I did experience several crashes and wished I were able to control some parameters (Fusion, Repair, etc.) with more precision. The overall detail in scanned models looked good due to the 10 FPS scanning and 0.05 accuracy of the scanner, but some fine features failed to resolve and the manual alignment wasn’t as useful as I would have liked.

The standard combo package of the Mole is currently available for $519, and the more capable premium combo package isn’t much more at $599. Unless you already have a tripod, turntable, and grip that’s compatible with the Mole, it’s hard to recommend saving the $80 and going with the standard package when the slightly-more-expensive premium package contains everything you’ll need to make stationary as well as dynamic scans. 

However, the optional $89 (Android) or $159 (iOS) mobile phone Connect module and $679.00 color luxury package combo are harder to recommend given the lack of positive reviews or complete examples published by 3DMakerPro.

I’m generally impressed with the quality of mesh 3D models made with the Mole, and believe that it’s a useful tool for makers, hobbyists, and anyone else interested in creating 3D models. The JMStudio software acts as a simplified one-stop interface, at the expense of not being able to control some of the processes with the level of precision some users may require. As with other scanners, the 3DMakerPro store page for the Mole shows images of cast metal parts, people, and even colorful shoes being scanned, all of which are unlikely use cases for a first-time user.

Shanghai Micro Electronics Equipment Group (SMEE), a state-backed firm, is on track to reveal its first scanner capable of producing chips on a 28nm process technology by the end of 2023, state-controlled the Global Times reports. The machine could help China reduce its dependency on foreign wafer fab equipment, Bloomberg noted.

SMEE is aiming to introduce the first domestically produced SSA/800-10W lithography machine to the market by the end of 2023, which would be a major breakthrough for the company, which currently only produces scanners good enough for 90nm nodes and older. The litho scanner is expected to use locally developed and produced components, which is why the stock price of China-based optical components manufacturers Mloptic, Kingsemi, and Castech were up on Wednesday.

Meanwhile, there is still uncertainty about whether the Shanghai-based firm can produce these machines at scale. Details about this technological advancement were not readily available on Wednesday, and the sources of information are the state-owned Global Times and Xinhua News Agency. 

China's major chipmakers, such as Semiconductor Manufacturing International Corp. (SMIC) and Hua Hong Semiconductor, all use equipment made abroad by companies like ASML, Canon, Nikon, and Tokyo Electron. However, the U.S., Japan, and the Netherlands recently imposed restrictions on Chinese companies when it comes to sales of advanced wafer fab equipment. These restrictions essentially blocked access to 14/16nm-capable tools for China-based foundries. 

For now, SMIC and Hua Hong can purchase 28nm-capable tools from abroad, but a local scanner would be more cost-effective. Furthermore, its emergence would mark SMEE's ability to leapfrog several generations of scanners and go straight from a 90nm-capable to a 28nm-capable machine. SMEE's reported breakthrough could be a significant step in developing self-reliant chip manufacturing, assuming that the company designed the device entirely in-house.

Founded in 2002, SMEE is China's leading lithography machine maker and is considered China's only potential competitor to ASML.

The 3DMakerPro Lynx 3D scanner starts at $369 and brings high-resolution 3D scanning to a low price point. The Lynx is capable of being used in both handheld as well as turntable mode, both of which worked well during testing. 3D scanning can sometimes be difficult for beginners due to the complexity of the process involved, but 3DMakerPro has created a linear workflow using the JMStudio software app that will help first-time users successfully create 3D models.

During testing, the Lynx had issues scanning facial hair and hair, which may be a problem for users interested in using the scanner for scanning people. The JMStudio software app worked well for capturing point cloud data, aligning and cleaning, and making watertight mesh models that can be used for 3D printing. While many 3D scanners in this price range suffer from difficult-to-use software that is poorly documented and performs inconsistently, 3DMakerPro has put time and effort into the documentation of the JMStudio app which will help to reduce the learning curve of the Lynx for first time users. 

Specifications of 3DMakerPro Lynx

Included in the Box: 3DMakerPro Lynx

The 3DMakerPro Lynx premium package ships in a box that includes everything you need to start scanning immediately, with no additional accessories required. Since the Lynx uses visual tracking instead of markers like other scanners, there are no sticker sheets, textured mats, or any other temporary alignment aids included in the package.

The premium package includes the Lynx 3D scanner, a tripod that doubles as a handle, a data cable, a power adapter, a turntable, an extended surface for the turntable which increases the diameter, a cable to connect the turntable to the scanner, a quick start guide, and a carrying case for everything. I was impressed with the case and the overall presentation, something that shows thought was put into the customer experience with the scanner.

The scanner itself weighs about 500 grams, and fits easily in a hand. The cables plug into the rear of the scanner, and it can be held without using the handle/tripod. The enclosure of the scanner is lightly textured, and it feels like a high-quality product. There is a port on the side of the scanner where the power cord for the turntable plugs in.

Package Variations of 3DMakerPro Lynx

The Lynx is available in three packages: standard, premium, and luxury. The standard package includes only the scanner, power adapter, and data transfer cable, which is technically all you need to create a scan. The premium bundle also includes a turntable, tripod, and a carrying case. The luxury bundle includes a “color kit”, which will allow you to create color textures for your model with the addition of a DSLR camera that takes pictures during the scan.

3DMakerPro shipped me the premium version of the Lynx, which is the package that will provide the most value for the majority of users. The rotating turntable is perfect for scanning small objects, and the tripod doubles as a handle which is useful for scanning larger objects and people. These are both additions that most users will likely need at some point, while the color texture kit included with the luxury version will likely not be as useful for users focused on reverse engineering or mesh generation.

Setup of 3DMakerPro Lynx

The 3DMakerPro Lynx requires multiple connections for power, data, and turntable control and includes a wiring diagram that shows the proper connections for all the components. It’s a surprisingly complex setup for a consumer scanner, and the non-standard proprietary cables are something you’ll want to keep in mind if using this scanner in an environment where they could potentially be cut or damaged.

The primary device cable is composed of two joined cables; one provides power to the Lynx scanner provided by the external power supply, and the other transfers data to a USB 3.0 connection. In addition to these cables, there is a separate microUSB cable that connects the scanner to the turntable. The turntable itself has no physical controls, and there’s no way to increase/decrease speed or turn it on and off without unplugging it.

The included handle doubles as a tripod and it feels sturdy and robust. The design is clever, and the handle can be twisted to allow the tripod legs to fold out. The legs all have a rubber tip on the end, so the tripod grips the surface it sits on when fully extended. There is a collar on the base of the tripod that twists and locks the legs into place after extending them.

After extending the tripod, it can be connected to the scanner using a standard threaded camera tripod mount. There is a thin layer of cork on the top of the tripod which provides grip and creates an interference fit when fully screwed in to prevent the scanner from wiggling or being jiggled if bumped. When plugged in, a fan inside the scanner turns on and stays on while the scanner is in operation. It’s not particularly loud, but it’s not something you’ll want to leave on unless you’re using it.

Installing the Software for 3DMakerPro Lynx

3DMakerPro offers a software app called JMStudio for capturing scan geometry, editing point clouds, and exporting a 3D mesh. This app is available for PC and Mac, and offers advanced functionality that should satisfy the needs of the majority of users. In addition to simply capturing scanned surfaces as point clouds, JMStudio can also automatically merge surfaces, delete isolated points, and simplify an .STL mesh before exporting. I’m using the Mac version of JMStudio for this review (v2.2.6) on an 64GB M1 Max MacBook Pro.

Turntable Scanning on 3DMakerPro Lynx

The JMStudio app has a dedicated turntable mode which can be used to make scans with an automated turntable-based process. This setup is ideal for scanning models that are between 6 and 8 inches cubed, and allows for multiple scans on a turntable that are merged together to create a full mesh. This mode has three steps you must complete before creating a mesh: preview, initialization, and scanning.

The first part of the scanning process is the initial preview, which you can use to make sure the model to scan is fully in frame while rotating. If you plan on flipping the object or moving it between scans, this is your chance to make sure that all of these orientations will fully fit within the scan volume. Once you’ve confirmed that the model will fit in the scan volume, this step is complete.

Next comes the initialization step. Remove the object from the turntable and the JMStudio software will automatically detect the top of the turntable and highlight it in red. This step allows the software to remove the turntable from the scan geometry automatically without tedious point cloud selection. Once the turntable is highlighted in red, this step is complete.

Finally, the scanning step begins the geometry acquisition process. The software will capture exactly 320 frames before automatically stopping, which represents a full revolution of the turntable. The turntable is still visible in this step during the scan, but it will be automatically edited out later. This step should be repeated until all surfaces of the model to capture have been captured and are visible in the scan preview.

JMStudio can automatically align scans, a feature that I found to work well without any manual intervention required. In the first photo, you can see that the turntables are all aligned, while the scanned surfaces all overlap. After automatic alignment, the scans are combined and the turntable geometry shifts to show how the models were positioned.

Finally, the processing step brings everything together to make a 3D model. This step fuses the point clouds, removes any disconnected points (or noise), repairs any small holes, and can also reduce the polygon count or apply a color texture. Since I don’t have the color kit, I wasn’t able to apply a color texture but I tested all the other features. From here, the model can be exported as an .STL to share, continue processing, or 3D print.

Scanning Shiny Objects with 3DMakerPro Lynx

Like most 3D scanners, the 3DMakerPro Lynx can sometimes struggle when scanning shiny, glossy, or reflective objects. This is compounded when scanning glossy black objects, which sharply transition from very dark to very bright and can be difficult to accurately capture. For example: this model of a koala has  a matte granite-like texture, but the nose, eyes, and ears all have glossy black paint applied. Using turntable mode, I scanned the model three times, aligned the scans, and merged them into a mesh.

The shiny nose of the koala is almost completely missing, and the eyes and ears appear misshapen and lack detail.The surface of the model is a speckled granite-like finish, and the scanner had no difficulty capturing the geometry in areas like the arms and stomach of the model. While this model could likely be repaired or fixed in post-processing, the best way to get a quality scan is to use a matting spray like AESUB Blue or a dry shampoo to reduce the reflectivity of the model and make it easier to process. 

To prep the model, I used Batiste Dry Shampoo, which creates a very thin layer of white powder on the model that can easily be washed off after spraying. The difference is immediately apparent, and the geometry of the nose is easier to see, even in a photograph. This thin layer of dry shampoo is all that is needed to prepare the model for 3D scanning, and it leaves no permanent mark or deformation.

I scanned the model again using the same orientations and processing as the first, and the difference is dramatic. The nose was captured perfectly, and the mesh is manifold and even. If you pay close attention to the subtle reliefs in the ear, you can see that they are consistent with the original statue. The first scan shows the reliefs as wobbly and rough, while the second scan looks much closer to the actual model. The matting spray made a large impact, and it’s a powerful tool to use when scanning shiny black models.

Mobile Scanning on 3DMakerPro Lynx

3DMakerPro also offers an optional accessory called the Connect for Lynx which allows users to plug the Lynx scanner directly into a mobile phone to use with the JM Studio for Mobile app. The app currently has 2 ratings on the iOS App Store; 1 is 5 stars and includes an enthusiastic blurb while the other is 1 star and claims the app does not work their scanner. Given the computational power typically required by scanning apps, I would imagine most users would be better off using a computer for creating a mesh instead of using a mobile app.

Reverse Engineering on 3DMakerPro Lynx

A common application for 3D scanning is reverse engineering, or capturing geometry so it can be reproduced or improved upon. To test out this particular application, I scanned a vintage side table that is missing a piece of wood on one side. The piece is present on the other side of the table, so my plan was to scan this area, isolate the missing piece and create a printable file, and print it out to replace the missing piece. 

The side table has a satin finish, so there was no processing or prep required before scanning aside from brushing off dust and generally cleaning it up. I took three scans total; approximately capturing the piece to replicate from the front, top, and rear. The pieces didn’t align perfectly using the automatic merge, so I used the manual alignment and picked points to snap them together. Once the three scans had been aligned, I processed them with the point cloud fusion but skipped the repair step as I didn’t need a watertight mesh at this point. The fused point cloud was then converted to a mesh and exported as an .STL file.

Once the .STL file had been generated, I imported it into Blender to isolate and mesh the missing piece. This process involves selecting the pieces of the mesh I don’t need (legs, table surface, etc.) and deleting them. Once the piece had been isolated, I repaired the holes and made sure the file was a single watertight mesh with a flat surface that could be printed on an FDM 3D printer.

I printed the model on an Elegoo Neptune 3 Pro, one of the best FDM 3D printers available for desktop 3D printing. The piece fit perfectly on the build platform, and I printed the model at a .2mm layer height using Printed Solid Green Slime Glitter PLA filament.

The printed model fit perfectly in the gap left by the original part, and the flat bottom was perfect for adhering to the side table. The engravings on the wood match the look of the original, and the general texture of the part has an uneven, organic look to it. I thought about painting the part, but I like the contrast between the digitally created duplicate and the carved organic original.

The Lynx made scanning the part a fast and easy process, and the resulting mesh file was perfectly to scale and could be processed without much work. It’s easy to see the value of 3D scanning in reverse engineering, and the Lynx performed well in this particular test.

Scanning People with the 3DMakerPro Lynx

A common application for 3D scanning is taking 3D scans of people, which is a surprisingly difficult process. People tend to move slightly during a scan, which leads to minor distortions in scanned meshes. Hair is difficult to capture accurately, and things like facial hair, glasses, and earrings can also be hard to faithfully reproduce. 3DMakerPro includes a video on the Lynx page captioned “Faithful Object Reproduction”, which features a person being scanned and a very blocky sculpt of hair added to the final mesh. When the company that makes the scanner can’t show a quality scan of a person, that’s typically a sign that you won’t be able to either.  

I tried multiple times to get a scan of my face, but was not able to get a full scan regardless of what adjustments to the brightness and sensitivity I made. My facial hair was too dark to capture, and as a result the scanner simply skipped my facial hair and hair which left me with a model that looked a little spooky. While there are several potential solutions that could be attempted (shaving my face, shaving my head, wearing a hat, spraying my entire face with dry shampoo, etc.), none of them are fast or easy to implement. 

In their advertising material, 3DMakerPro shows a person with a beard, long hair, and glasses being scanned, something that would be difficult under the best of circumstances. Between this image and the mesh showing the sculpt of hair added, it seems that scanning faces is something that will require significant trial and error on the part of the user. Given that this feature is being advertised as “more user-friendly”, I think that it would be reasonable to expect a “Face Mode” with some presets that would make it easier to capture a scan of a person.

Bottom Line

Like many 3D scanners in this price range, the 3DMakerPro Lynx is advertised with images showing the scanner capturing people wearing glasses, shiny machined aluminum, and other objects that will set unrealistic goals for most first-time users. In practice, the Lynx works well when scanning objects on a turntable as well as larger models like the side table to create 3D geometry. The clean-up process in JMStudio is simple enough for beginners or users who are inexperienced with editing point cloud data, but the software is powerful enough to handle alignment and mesh clean-up.

With an intro price of $369 for the standard package, the 3DMaker Pro Lynx will likely attract a large number of users interested in learning more about 3D scanning and what the technology can accomplish. However, the jump in price to the premium package (which includes a turntable as well as the handle/tripod) is easy to justify as these accessories are tremendously useful when making 3D scans. It’s unclear what is required to take advantage of the full color texture mapping feature included in the luxury package, but the rendering on the site shows a separate digital camera which may be an additional cost for some users.

Overally, the Lynx feels like a step in the right direction in the entry-level scanning market. Inexpensive desktop scanners typically have confusing, buggy, and difficult to use software, like the Creality CR-Scan Lizard when it was reviewed. The turntable mode in JMStudio was intuitive and simple, which should result in beginners experiencing early success instead of frustration trying to align scans and merging point clouds.

Huawei has filed a patent application covering an extreme ultraviolet (EUV) lithography scanner, according to UDN. If the company builds such a scanner and achieves decent productivity, uptime, and yields, Chinese chipmakers could produce chips on sub-7nm-class technologies. The only question is when. 

In mid-November, Huawei filed an application with the State Intellectual Property Office for a patent covering an EUV scanner and its key components. The patent application number is 202110524685X, reports MyDrivers. 

The patent application appears to cover all crucial components of an EUV scanner, including a 13.5 nm EUV light generator (light source), a set of reflecting mirrors, the lithography system, and 'control management technologies' (we would speculate that this is how they call metrology), according to descriptions published by various media sources.  

Filing a patent is not equal to being able to build an EUV scanner, which is a highly complex machine featuring numerous state-of-the-art components that have to work perfectly in concert and for prolonged amounts of time. Furthermore, even with an EUV tool at hand, chipmakers still have to figure out the right pellicles for masks, resists, and many other things necessary for high-volume production. 

An EUV scanner with a 0.33 numerical aperture is the pinnacle of today's semiconductor production tools. Numerous companies attempted to develop such a tool, but only ASML succeeded after over ten years of development and with financial support from Intel, Samsung, and TSMC. Today, Samsung, SK Hynix, and TSMC actively use EUV tools from ASML, but Intel has yet to start high-volume chip production using these tools.  

For now, only Intel, Micron, Samsung, SK Hynix, and TSMC either use or plan to use EUV scanners. Moreover, only these five companies have developed (or plan to develop) process technologies sophisticated enough to take advantage of EUV scanners. Meanwhile, China-based SMIC could not get an already procured EUV tool to develop its own EUV-based fabrication process due to the Wassenaar arrangement. Therefore, it is evident that demand for EUV scanners potentially exists in China, and apparently, Huawei was eager to address it. 

As a world-class high-tech conglomerate with some $100 billion in annual revenue, Huawei pursues different goals and develops many technologies. The company's semiconductor production ambitions are well known, and they are not limited to chip production but also to building wafer fab equipment. Huawei's WFE efforts are advancing pretty well as the company applied for a patent covering an EUV scanner.

A Russian institute is developing its own lithography scanner that could produce chips using 7nm-class fabrication technologies. The machine is under development, with the plan to build it by 2028. When it is ready, it should be more efficient than ASML's Twinscan NXT:2000i tool, whose development took over a decade. 

After Russia unleashed its bloody war against Ukraine on February 24, Taiwan was quick to ban shipments of advanced chips to the nation. The U.S., the U.K., and the E.U. then followed up with sanctions that effectively prohibit virtually all contract chipmakers with advanced fabs from working with Russian entities. In addition, companies like Arm cannot license their technologies to Russia-based chip designers. As a result, the Russian government rolled out a national program to develop the country's own 28nm-class fabrication technology by 2030, reverse engineer as many foreign chips as possible, and educate local talent to work on domestic chips. 

However, there is a problem with a 28nm-class production node by 2030. Russia's most advanced fab can produce chips using a 65nm fabrication technology. Meanwhile, American and European makers of fab tools cannot supply their equipment to Russia due to sanctions, so the country has to design and build domestic wafer production equipment if it wants to adopt a 28nm node. Essentially, what has taken companies like ASML and Applied Materials decades to develop and iterate has to be done in about eight years.  

Apparently, the Russian Institute of Applied Physics of the Russian Academy of Sciences intends to beat all expectations and produce a 7nm-capable lithography scanner by 2028, according to its plans published on Nizhy Novgorod Strategy Development website (via CNews). 

A modern lithography scanner capable of processing wafers using a 7nm-class process technology is a highly complex apparatus that involves a high-performance light source, sophisticated optics, and precise metrology, just to name a few critical parts. However, as a leading applied physics university in Russia, IAP believes that it can develop such a tool in a relatively short amount of time.  

The tool will be somewhat different from scanners produced by companies like ASML or Nikon. For example, IAP plans to use a >600W light source (total power, not intermediate focus power) with an 11.3nm exposure wavelength (EUV wavelength is 13.5nm), which will require considerably more sophisticated optics than exists today. Because the light source of the device will be relatively low power, it will make the tool more compact and easier to build. Yet, it also means that its production of the scanner will be considerably lower than that of modern deep ultraviolet (DUV) tools. That might not be a problem, according to IAP. 

When it comes to timing, IAP may be slightly too optimistic. For everything below 32nm, chipmakers use the so-called immersion lithography (which is essentially a booster to DUV tools). ASML introduced its first immersion lithography system — the Twinscan XT:1250i —  in late 2003 with a plan to deliver one in Q3 2004 to produce 65nm logic chips and 70nm half-pitch DRAMs. It took the company about five years and another generation of tools to announce its 32nm-capable Twinscan NXT:1950i in late 2008, with customer deliveries starting in 2009. 

Then, it took the market leader some nine years to deliver its 7nm and 5nm-capable Twinscan NXT:2000i DUV tool in 2018. TSMC used less advanced tools with multi-patterning for its first-generation N7 fabrication technology, but the timing of ASML’s introductions demonstrate how hard it is to transition from 65nm to 7nm. It took ASML 14 years to go from 65nm to 7nm. Now, IAP, which does not have any experience in chip production or ties with chipmakers, intends to build a 7nm-capable machine for volume production from scratch in about six years. While the plan does not sound feasible, it looks like IAP is full of enthusiasm.

"ASML, the global lithography leader, has been developing its EUV lithography system for almost 20 years and the technology has turned out to be incredibly complex," said Nikolai Chkhalo, Deputy Director of the Institute of Physics of Microstructures of the Russian Academy of Sciences for scientific and technological development. "The main objective of ASML in this case was to maintain the extremely high productivity that is needed only at the world's largest factories. In Russia, no one needs such high productivity. In our work, we start from the needs and tasks faced by domestic microelectronics — and this is not so much about quantity, but about quality. First of all, we need to transit to our own fabrication processes, develop our own design standards, our own tools, engineering, materials, so our own path is inevitable here. In fact, we need to balance between simplicity and performance." 

IAP plans to build a fully functional alpha scanner by 2024. This one will not have to offer high productivity or maximum resolution but will have to work and be attractive to potential investors. IAP intends to build a beta version of the scanner with higher productivity and resolution by 2026. This machine should be mass production ready, but its productivity is not expected to be at its maximum. The final iteration of the litho scanner is said to emerge in 2028. It should get a high-performance light source (hence better productivity), better metrology and overall capabilities. There is no word how many of such machines IAP and/or its production partners will be able to produce by 2028.

It should be noted that fab equipment is not limited to lithography scanners. There are other types of machines performing etching, deposition, resist removal, metrology, and inspection operations that are not made in Russia. Furthermore, there is somewhat less advanced machinery like ultrapure air and water generators which also are not produced in Russia. Even if IAP RAS manages to build a lithography tool, Russia will still be a few hundred tools short of building a modern fab. Also, fabs need ultrapure raw materials produced in countries that will not supply to Russia.

Thingiverse has some serious competition. There’s a new library of 3D printable files in town, except it’s not new at all. It’s Printables.com, formerly known as PrusaPrinters.org. 

Joseph Prusa and his Prusa Research team wanted to create a place to house high-quality 3D models and make it open for people using any brand of 3D printer. Although .stl files can be reproduced by any 3D printer, some makers weren’t sure if they would be welcome on a site that also sells the popular Prusa MK3S+, which is one of the best 3D printers.

By renaming their file library, Prusa Research is making it clear that Printables.com is a community hub for everyone who enjoys 3D printing.

The fast-loading site remains ad free. Despite being run by a leading manufacturer of 3D printers, there are no ads for their printers, filament and resin. Only a small item on the menu labeled “Prusa Eshop” reminds users that, they too, could own a Prusa.

Prusa Research gradually rolled out changes to its website, starting in December 2021 when it announced a point system to reward frequent visitors. People could rack up points by uploading designs, posting photos of their makes and creating collections of models. Designers are further rewarded when other makers download their designs.

These points are called Prusameters and once you “fill a roll” with 350 points, you get a voucher for a free roll of Prusament filament. Points can also be exchanged for Prusa-branded mugs, hats, shirts and hoodies.

Printables.com users are also rewarded with virtual Badges that they can display on their profile pages. The badges are awarded like game levels, with levels for Designers and Makers. There are also badges for frequent visitors, contest winners and more.

The new website also features themed contests, with Prusa printers and Prusameters given out as rewards. The contests not only create excitement for the community, but they also encourage designers of all levels to fill the Printables.com library with quality new files.

Artec Europe, makers of precision 3D scanners, is claiming Creality 3D Technology copied code from Artec Studio to make its own CR Studio 3D scanning software. It requested a temporary restraining order to prevent Kickstarter from distributing funds pledged to Creality for the CR -can Lizard. Artec is seeking a jury trial against the popular 3D printer maker for alleged patent and copyright infringement.   

Creality did not respond to our request for comment but has reassured backers that it will honor their Kickstarter pledges. One backer in St. Louis received a message from Creality saying:

"Dear backers,

We have received your messages, and we are dealing with it now. Please give us some time, and rest assured that we will definitely send the product to you. Any questions please contact scanner@creality.com.

Creality team”

A little more than $4 million US dollars was raised by 9,472 backers interested in Creality’s new 3D scanner. According to court documents, Artec feared that once the backer’s money is transferred to China, it would be “virtually impossible for Artec to recover any portion of these funds.”

Kickstarter’s lawyers have stated the company is a neutral party to the copyright dispute and, according to its terms of service, is obligated to transfer funds within 14 days of the conclusion of a successful campaign. However, the Kickstarter page for the CR-Scan Lizard has been replaced by a message about the intellectual property dispute.

Creality is best known as the manufacturer of affordable 3D printers, such as the popular Ender 3 Pro and Ender 2 Pro, which are among the best 3D printers overall. It partnered with Shenzhen Jimuyida Technology, a 3D digital application provider, to develop its new scanner. 

Artec is known for the Artec Leo, a professional wireless 3D scanner priced at $29,800.

The copyright dispute revolves around the software behind the CR-Scan Lizard and the CR Scan 01. Tom’s Hardware was given the opportunity to independently review a prototype CR-Scan Lizard. We found the beta software to be somewhat lacking and in need of improvement, but we liked the hardware.

In its filing, Artec claims several former employees went to work for Jimuyida and brought with them improper copies of Artec’s code. Litigation over this intellectual property rewarded Artec exclusive rights to software it had developed.

Creality’s lawyer asserts that Jimuyida independently developed the code without the work of former Artec employees.

Kickstarter is an American company headquartered in Brooklyn, New York. Creality is based in Shenzhen, China. Artec headquarters are in Luxembourg City, Luxembourg with offices in both Santa Clara, California and Shanghai, China. The complaint was filed in the U.S. District Court Eastern District of New York.

Court documents are available to the public at CourtListener.com under Artec Europe S.A.R.L. v. Schenzen Creality 3D Technology Co., LTD.

When the tool you need is out of reach, sometimes the best solution is to create one yourself. Single-board computers like the Raspberry Pi can really help speed up the process. When Eric, host of the That Project channel on YouTube, needed to read and generate QR codes, he built hardware to do it using a Pi running Open CV.

This particular project is built around a Raspberry Pi Zero 2 W, which is responsible for both scanning and processing QR codes. The project is aided by his ESP32 QR code generator which can be used to make custom codes on the spot. If you’re interested in other options for reading codes with a Raspberry Pi, check out our review of the Barcode Reader HAT.

Instead of using a HAT like we did, Eric is using a USB webcam to capture images. The Pi Zero 2 W then scans the image for possible QR codes. Using Open CV, it determines whether or not a QR has been detected and if so, what information it contains. The image processing can be viewed in real-time as Eric has included a 7-inch display. Possible QR codes can be seen with bounding boxes around them as Open CV tries to interpret the image.

The ESP32-based QR code generator runs in tandem. Eric goes into detail breaking down the code for this device, which runs primarily on Arduino-based .ino files. The source code for the QR code maker is available at GitHub but unfortunately, there isn’t much information available yet breaking it down.

If you want to recreate this Raspberry Pi project yourself, Eric provides enough information to help get you started. It doesn’t take much hardware, but you will need some experience with Open CV to bring everything together. (Check out our tutorial on how to use Open CV with a Raspberry Pi for help getting started.) 

Intel posted a new Linux driver yesterday that surfaced an interesting new feature coming to Sapphire Rapids server processors. The feature takes silicon integrity and reliability checks to the next level and is dubbed In-Field Scan. Phoronix spotted the new Linux driver and highlighted the importance of defect discovery in server processors prior to deployment or before working on critical tasks.

Servers already boast parity or ECC checks as standard features for RAM, and there are similar checks on data going to and from storage, across networks, and more. However, the new In-Field Scan feature specifically targets the processor. To get the new In-Field Scan system to work, Intel Sapphire Rapids Xeon CPUs will have integrated checking features, and the In-Field Scan kernel driver will provide a user interface for the checks.

After delving deeper into the software, the source publication says that the Intel In-Field Scan driver can instigate full CPU tests. There is also a more granular level of control available, allowing testing for each separate CPU core. These scan results are then saved in log files.

When In-Field Scan becomes available, system builders or administrators might run it before commissioning a server, before specific critical tasks are run on the server, or simply on a given schedule. The Linux driver is out now, but the CPU-specific test files that reveal more about the checks and CPU characteristics aren't currently available.

Why Do We Need In-Field Scan Now?

With the growth of massive data centers thanks to the internet economy, streaming video/gaming and cloud computing, manageability is increasingly important to the smooth running of servers. For example, if specific CPUs or cores can self-report errors, they can be easily found and swapped.

Another movement in technology that might make In-Field Scan an important tool is the race to the Angstrom era. As chip features get smaller in the search for higher density, performance and efficiency, they become more susceptible to known errors and unexplained errors – sometimes called soft errors.

Soft errors might occur more often in our most state-of-the-art chips due to pushing physics and the physical size of transistors to their extremes.

Some think that the errors may not emanate from the new smaller chip processes but simply the susceptibility of these tiny structures to the nature of the universe (i.e., cosmic rays).

A year ago, news hit the wires that the cutting-edge NASA space exploration vehicle,  Mars Perseverance, was running the same single-core Power PC CPU that powered the Bondi Blue iMac back in 1998. In brief, the reason for this apparent tech mismatch was that the RAD 750 (based on the PowerPC 750) was hardened to withstand up to 1,000,000 Rads and extreme temperatures. Moreover, its larger process logic gates were far less susceptible to cosmic ray interference than modern CPUs. Of course, our atmosphere reduces cosmic radiation on earth, but it is still there.

Interestingly, Intel interpolates the error susceptibility of its chips and ICs concerning cosmic ray interference using a particle accelerator at the Los Alamos Neutron Science Center.  

SVS Embedded has created a new Raspberry Pi Pico project capable of checking in students, employees or whoever else you need to keep track of in the form of a smart attendance system. Although this project is somewhat complex, the modular nature of the functions are simple enough for the RP2040-powered board to handle. 

The team at SVS Embedded regularly share microelectronics projects intended mainly for students and educators. Users have the option to purchase kits for many of the projects that provide the hardware you need as well as detailed construction notes. 

The project can be configured with additional components or less depending how complex you’re willing to make the system. In the demonstration, a Pico is used to monitor for input from a finger print scanner and RFID chip reader. After the person is verified or denied, a text message notification is sent via SMS.

To recreate the project, users will need a Raspberry Pi Pico microcontroller. SVS Embedded is using an R307 finger print sensor and an EM18 RFID reader to scan RFID chips. To get SMS notification support, a SIM800L GSM module is used. The project also features a 16 x 2 LCD display used to provide instructions to the user as they scan into the system.

When the power adapter is connected, the Pico loads a welcome screen on the LCD inviting students to use the finger print scanner. When the print is registered, the system then prompts for a chip to be read using the RFID scanner. When the ID has been assessed, the Pico then sends a command to the GSM module to initiate the notification text.

This project could easily be adapted or modified to be more or less secure by simply removing components or adding even more levels of identity verification like password confirmation. If you want to recreate it yourself, check out the project page and take a look at their project kit at SVS Kits to get everything you need in one go.

If you’re looking for more cool creations from the maker community, check out our list of best Raspberry Pi projects and see what they’ve been up to.

Even if you have one of the best 3D printers, you’re faced with a dilemma: what do I print? If you’re a talented artist, you can design your own models and, if not you can free ones on sites such as a Thingiverse. You can even convert a 2D logo into a 3D print. However, if you want to take a real-world object and make it printable, you’ll need a 3D scanner.

This month Creality launched a Kickstarter for a new 3D printing accessory: the CR-Scan Lizard. Named after the fantastic eyesight of the lizard (they didn’t say which one) this scanner promises to open doors for makers who don’t want to design models from scratch.

The CR-Scan Lizard has interesting possibilities, and with better software could be a must have tool in any maker’s workshop. But what we were given to review is limited by a poor user interface, buggy software and modest scan quality. The software had functions that were not explained and often crashed while attempting to process data. Marketing for the scanner is overly optimistic and shows a happy family doing things I was not able to replicate, like face scans or taking it more than 6 feet from a power outlet.

Retail pricing starts at $599, but Kickstarter backers will receive an Early Bird discount, currently 45% off. Units are promised to ship in April 2022.

The CR-Scan Lizard copies lightly colored objects quite well and does a decent job with dark items. It doesn’t need marking points, but when removed from its rotating baseplate the scanner easily gets lost. It works best as a tabletop unit, especially since it requires three different wires with 7 connection points, making your desk a bit messy. If you need a portable 3D scanner, this is not the one for you.

Kickstarter

I want to emphasize that the CR-Scan Lizard is currently a Kickstarter. It smashed the pledge goal of $12,819 within minutes of launch and currently has nearly 6000 backers pledging over 2 million dollars. Creality is also a well-known leader in the 3D printing space. It’s a safe bet that backers will receive their product.

However, pre-launch review models – like the one sent to Tom’s Hardware – came with buggy Beta software and vague directions. As other reviewers unboxed their scanners online, I noticed at least two models of baseplates and different colored cords were shipped out. The scanner I’m reviewing for you today may very well be different from the one you would get if you buy it.

Specifications: Creality CR-Scan Lizard

Included in the Box: CR-Scan Lizard

I received the CR-Scan Lizard premium package for review, which comes with a USB powered turntable, tripod and all the required cords tucked into a padded carrying case. The review scanner came with a USB stick with a “demo” copy of CR Studio and a PDF of the quick start guide. I also received a small plaster bust of Michelangelo’s David for practice. The bust is not mentioned in the Kickstarter.

The scanner has a sleek design and at 370 grams feels quite solid. The data cable connects to the scanner with a sturdy aviation grade plug that won’t fall out when attempting hand held scans.

It has a noticeably loud cooling fan and the LED lights strobe when it scans dark objects. It’s not a stealthy machine.

PC or Mac

The CR-Scan Lizard is compatible with both Windows and Mac and requires a minimum of 8GB of ram to process scans.

Software Compatibility of CR-Scan Lizard

The CR-Scan Lizard comes with CR Studio, a one-stop shop for scanning and editing the resulting data. Under good conditions, this is all you need to take an object from scan to a sliceable STL file for 3D printing.

CR Studio saves files as proprietary project files, STL and OBJ. STL and OBJ files can be imported to the modeling software of your choice for further touch up or transformation.

Assembling the CR-Scan Lizard

The CR-Scan Lizard is shipped fully assembled. You only need to plug the various cords into the scanner, turntable, a computer and power source.

Three cords are needed for Table Top Mode. There is a Data Transmission cable that connects the scanner to your computer via a USB 2.0 port, with a Y split that connects to a separate power cable. The power cable is about 5 feet long.

The turntable is USB powered, meaning you have another cable to connect to a USB power source. The table spins at one speed and does not have a power switch. It is not controlled by the software, it just spins until unplugged.

The 10-inch tall tabletop tripod is well-designed and feels sturdy. It has a short telescoping neck and folds down into a handheld monopod. The screw mount is a standard size and can be swapped out with any commercial tripod.

Installing the Software for CR-Scan Lizard

The CR-Scan Lizard requires CR Studio to operate and has no buttons or monitor of its own. Installing the review copy of CR Studio was far from simple – hopefully the final version will have a better user interface.

Directions for how to use CR Studio were vague and many options had to be discovered by trial and error. Creality seems to have great faith in its automatic scan processing, so when I had a bad scan I was on my own to sort it out. It took several frustrating sessions with the software to find a good routine.

Table or Hand Scan Modes

After CR Studio is launched, you’ll be asked if you want to do a Table Mode or Hand Scan.

Table Mode is fully automatic and the easiest method. This technique uses the turntable and tripod.

Handheld removes the need for the turntable and its alignment aids. You will need to carefully and slowly pass the scanner over an object, covering all sides. I did not have the patience to make this method work, as the scanner kept losing its place and crashing.

A third technique uses Hand Scan mode and the turntable together. This eliminates the need to walk around the object you’re trying to scan. It’s useful for items that fit on the turntable but are too large for the scanner to capture while mounted on the tripod. This works better than a pure handheld as you can concentrate on holding the scanner steady while slowly moving it up the object.

How to Scan with CR-Scan Lizard

The CR-Scan Lizard works best in Table Mode while scanning objects with the smallest dimension being 15 mm to 200 mm for the longest side.

The scanner body lacks a power switch, so it immediately turns on when plugged in.

Lighting does not matter to the CR-Scan Lizard as it scans with an infrared light beam. I scanned in normal indoor lighting without trouble. The scanner also works fine under studio lights or outside if you have a power source. Here are the steps you need to take to scan an object.

1. Mount the scanner on the tripod and set it 300 mm (about one foot) from the turntable. The scanner only “sees” 400 mm from the lens, so make sure the immediate area behind the turntable is free of clutter. Place your item on the center of the turntable. Adjust the tripod so that scanner is pointing at the center of the turntable and most of your object is within view.

It’s ok if you can’t see the entire object, as the CR-Scan Lizard is able to combine multiple scanning passes.

2. Select the “Slam Mode” (Simultaneous Localization and Mapping) to Geometric for objects with three-dimensional details or Texture for smooth objects with painted details. I used Geometric mode for all my test scans.

Note: Texture mode appears to be for a full color option that Creality will release at a later date. I saw no benefits when testing the texture mode on the several items I scanned.

3. Select Size (Scan Object). On the Demo version of the app I was given, you need to select the size of the item. The PDF directions show screenshots where this information is either missing or replaced with a Scan Quality of Normal/Fine. I don’t know which version you will get on the final release. Small objects are under 500 mmm, so I selected “small”.

4. Set the Brightness, which is the scanner’s exposure setting. This is a scale from 1 to 4, with 1 being used for white objects, 2 and 3 for medium shades and 4 for black. We’ll use the 1 setting for this bust of David.

5. Select Frame Rate. Creality gives no advice on how to set this, so I used a High frame rate.

6. Click the blue Preview button. The screen will show what the scanner sees and you can make adjustments to the tripod’s angle and the brightness levels. Preview mode continues until you press Stop.

7. Conduct initial scan for the baseplate. After pressing Stop, the button now reads “Initial”. Remove the object and click the button. It will now scan the turntable to assist with aligning the scanned images. Creality didn’t say how long you should do this step, I waited at least 15 seconds. Press “stop.”

8. Scan. Place the object back in the middle of the turntable approximately where you set it up during the preview and scan. The CR-Scan Lizard will now scan a full rotation of the object. Most scans take about 320-340 images, which you can see on the “scanning count” box. Wait for the scanner to stop on its own.

Most objects need at least two passes to capture all the angles. You can see that the David bust is missing the top of it’s head. Reposition your object on it’s side or back for a second pass.

9. Click Append to add another scan, then click Scan when you have it in place. Wait for the CR-Scan Lizard to make another pass.

10. Process. Two scans are enough for this model and now we can click the Process button in the lower right corner. CR Studio will analyze the scans and try to fit them together. It will ask you “Whether the effect is reasonable” which means, “did I line this up right?” If all the images are facing the same direction, click “Qualified” and CR-Studio will process the scans automatically.

If images are facing different directions you’ll need to manually align the scans.

Creality’s directions on manual alignment are vague and honestly, this is the buggiest part of the software I was given for review. It only works if you have two scans – I was never able to align three or more scans using their method. Instead of painstakingly matching three points of each hard to read scan, let me share a hack with you.

11. Delete the bases. I don’t know why, but if you merely delete the base of each scan – highlighted in red – it works. To delete parts of the scan you need to switch to Edit Mode which is the symbol with the cube and circle. Anything that is red can be deleted. Click delete.

12. Select any stray pixels surrounding the scan with the polygon sector and delete these too. Now press Process and your scans should pop together correctly.

One thing the CR-Scan Lizard does right is allow you to take as many scans as you want for one project. You can reposition the tripod, select a new light level or even frame rate. All the scans are saved in the project folder and you can examine them closer later.

Bad scans can be cleaned up in editing mode or discarded entirely. You can process the scan as many times as you like.

Once you’re happy with your scan, export it as an .OBJ or .STL. You can take these files directly into Cura or PrusaSlicer and print them.

Here’s my print of the David bust. You can see that it lacks details around the eyes, lips and hair, but is a fair representation of the model. This scan was only edited using CR Studio, brought into Cura 1.13 and printed with a .12 layer height on my Ender 3 Pro. I used Printerior White PLA.

How to Scan Difficult Surfaces

Shiny, clear or black items are incredibly difficult for any hobby level scanner to capture with current technology. Though Creality has made great strides in capturing darker objects, it is still hard to scan items with both dark and light portions, like this Baby Yoda figure. The plastic toy is green, gray and tan with shiny black eyes. It looked pretty bad, even after printing it in Protopasta Recycled Still Colorful 10 PLA.

For $45 you can buy a can of professional “fixing” spray – a substance that coats your object in a mat finish that disappears after several hours.

Instead, I found that cheap dry shampoo works as a fixer. Dry shampoo is basically alcohol, starch and fragrance. Since it's made to spray on a human head, it’s pretty safe to use on your scannable objects. I used a can of Suave I picked up for $6 at Walmart. It coats your item with a fine, light tan powder that wipes right off and is easy to read for the scanner.

After spraying Baby Yoda with dry shampoo I was able to capture a good scan. It’s soft around the cuffs and necklace, but you can see how well the ears and eyes are replicated. To make sure no detail was lost in printing, I sent this to the Phrozen Sonic Mini 8K and used Siraya Tech Fast Navy Gray resin.

How Accurate is the CR-Scan Lizard?

Creality lists the scanning accuracy of the CR-Scan Lizard as .05 mm. Items scanned do have decent dimensional accuracy but are not perfect copies. Fine details and texture are lost. No one is going to be counterfeiting Baby Yodas with this scanner.

However, the scans are close enough to serve as a template in CAD software. If you wanted to scan a tool so that you could make a cover or holster for it, this scanner would do great.

As you can see, the simple lines of this wooden horse scanned very well and required only minimal clean up in CR Studio. This horse was printed at .2 line height on the Ender 3 Pro, using Printerior Purple PLA.

Bottom Line

The Creality CR-Scan Lizard is an interesting foray into the world of 3D scanning. The potential for this device is amazing – imagine being able to scan your kid’s actions figures and put their own faces on them? Or scan an object to make a perfect replacement part? There’s a lot of fun things we could do with a 3D scanner.

I believe the scanner that Creality sent for review works – but its software currently does not work well a all. The demo copy often crashed and has many features without explanations. The scans I’ve shared on this review took several hours of trial and error in scanning and clean up efforts.

The scanner’s resolution is overhyped, with results – even printed on the best resin printer currently on the market – soft and unfocused.

The need to tether the scanner to a robust laptop or your desktop PC really limits how you can use this device. I tried the unit on three computers – two of them custom-built gaming PCs with good video cards – and still the CR-Scan Lizard would falter on darker colors.

Creality can release updated software, and they have until April to fix the bugs before paying customers receive their units. If you have money to gamble on a fun experiment, go ahead and buy the CR-Scan Lizard while it's on Kickstarter for a deep discount. Otherwise, wait for more final software to come out before making your decision. We’ll update this review if and when a new and less-buggy version of the app comes out.

In an enthusiastic update posted to Reddit, OpenScan creator Thomas Megel has announced support for the latest high res Arducam autofocus camera with a demo video showing a notably detailed 3D scan of a safety pin—all powered by a Raspberry Pi 3B+.

Megel began his work on the OpenScan project back in 2018 but prior to the 3D scanning development, he worked as a high school mathematics teacher. These days, he dedicates more time to the project which now has an official website for updates, downloads, news and anything related to OpenScan.

OpenScan works using a scanning process known as photogrammetry and relies entirely on a Raspberry Pi to power the operation. The photogrammetry process involves taking many pictures of an object from various angles and using those images to stitch together an accurate 3D rendering of that object. Megel provides not only the tools to process these images but also 3D-printable rigs for scanning objects with a Raspberry Pi.

These scanning rigs are designed to house the Raspberry Pi, a camera module of your choice and provide a small platform to place the object to be scanned. According to Megel, the biggest issue with the official Pi Cam modules involves the focus adjustment settings. This latest update provides support for Arducam’s IMX519-based high res camera which features autofocusing capability, making it an ideal candidate for this type of project.

Megel explains that several open-source options are available for processing the images into a 3D object including applications like VisualSFM, Meshroom, and Colmap. However, users can also opt to use Megel’s cloud-based platform, OpenScanCloud, to process the scans, freeing up valuable processing power on your personal PC.

OpenScan is, as the name suggests, an open-source project with code available for anyone to explore. More information about OpenScanCloud is available on Megel’s OpenScanEu GitHub profile. There users can also find more information about the latest firmware update for the high-res Arducam module.

Visit the official announcement over at Reddit for more details and the opportunity to ask Thomas Megel questions in the comments. If this project is exciting to you, check out our list of best Raspberry Pi projects for more inspiring creations from the maker community.

Intel is clearly behind its rivals TSMC and Samsung with the first generation of extreme ultraviolet (EUV) lithography technology, but it certainly wants to be the first to adopt the next-generation EUV tools featuring a 0.55 NA (or high-NA) that provides higher resolution and productivity. This week Intel announced that it had ordered its second experimental High-NA tool from ASML.  

Intel announced plans to adopt ASML's High-NA Twinscan EXE scanners for high-volume manufacturing (HVM) starting in 2025, which is when the company intends to start using its 18A (~1.8 nm) fabrication technology. To do so, Intel has experimented with High-NA tools since 2018 when it obtained ASML's Twinscan EXE:5000, the industry's first EUV scanner with a 0.55 numerical aperture. This week the company ordered ASML's next-generation High-NA tool, the Twinscan EXE:5200. 

"Compared to the current EUV systems, our innovative extended EUV roadmap delivers continued lithographic improvements at reduced complexity, cost, cycle time and energy that the chip industry needs to drive affordable scaling well into the next decade,” said ASML President and CTO Martin van den Brink. 

High-NA EUV tools are crucial for higher resolution (<8 nm vs ~13 nm for 0.33 NA EUV) that enables smaller transistors and higher transistor density. In addition to a completely different optics design, High-NA scanners promise to offer significantly faster reticle and wafer stages as well as higher productivity. For example, the productivity of the ASML Twinscan EXE:5200 is over 200 wafers per hour (WPH). In contrast, ASML's top-of-the-range 0.33 NA EUV machine, the Twinscan NXE:3600D, produces around 160 WPH with a 13.5nm light wavelength.

"Intel's focus is to stay at the forefront of semiconductor lithography technology and we’ve been building our EUV expertise and capacity over the last year," said Dr. Ann Kelleher, executive vice president and general manager of Technology Development at Intel. "Working closely with ASML, we will harness High-NA EUV's high-resolution patterning as one of the ways we continue Moore's Law and maintain our strong history of progression down to the smallest of geometries." 

Just like the adoption of 0.33 NA EUV did not eliminate the usage of deep ultraviolet (DUV) lithography, 0.55 NA EUV will not replace the existing DUV and EUV tools used in modern fabs. In fact, ASML plans to continue developing more advanced DUV and 0.33 NA EUV scanners for years to come. Meanwhile, High-NA EUV lithography will be a key technology to reduce transistor sizes and increase transistor density.

Orico has introduced its new inexpensive portable SSDs with biometric security in the form of an embedded fingerprint sensor. The new drives feature the performance of SATA SSDs and are compatible with all PCs and some tablets released since 2009.  ]

The key selling point of these new drives is, of course, their fingerprint scanner that works in tandem with special encryption software (which uses AES-256 technology) to prevent unauthorized access to the drive's content. The application is compatible with Windows (7 or later), MacOS, Linux, and Android, so you can use the SSD with all PCs and tablets (or even smartphones) released since 2009.

Set to be available in 256GB, 512GB, and 1TB capacities, Orico's Fen 300-series portable SSDs feature a USB 3.2 Gen 2 (10 Gbps) interface as well as a maximum sequential read transfer rate of 520 MB/s, reports TechPowerUp. 

Orico's external SSDs with a fingerprint reader come in an aluminum enclosure that not only protects the mSATA drive inside but also serves as a cooling system for the guts of the SSD. The drive measures 90×55×9.7 mm.

Nowadays there are a lot of USB flash drives with biometric or other kinds of protection. But when it comes to external SSDs (which tend to feature higher capacities and therefore may store more confidential information), there are few models with advanced security capabilities. In many cases, drive that store private data are transported without any protection, which means additional risks. Orico's portable SSDs partly solve this problem, even though some will prefer drives that don't require a program that has to be installed before using an SSD.

Traditionally, storage devices with sophisticated security come at a hefty premium, but this is not the case with Orico's SSDs — the 256GB version is $79.99, the 512GB variant should cost $99.99, and the 1TB model is priced at $159.99.

Ever wonder how those cool special effects come together in movies? With the help of a few hundred Raspberry Pi of course! We spotted a tweet by Arturo182 which gave us a behind the scenes look at creating the visual effects for Marvel's recently released Black Widow. In the image we get a glimpse of actress Olga Kurylenko decked out head-to-toe in some cool combat gear while surrounded by Raspberry Pis all controlling Nikon DSLR cameras.

The best Raspberry Pi projects use at least one Pi but this incredible rig is constructed with over 200 individual Raspberry Pis. Looking at the port layout we are confident that these are Raspberry Pi 3 / 3B+, which means that this rig was built in the past few years. Other Marvel titles to use the same technology include Venom and Avenger’s End Game.  We even spotted a glimpse of Tom Hardy being scanned for Venom. This exciting system wasn’t developed by Marvel but rather a specialist team, The Scan Truck who offer scanning services for movies, music videos and commercials. Plenty of other well-known names have worked with The Scan Truck, including Nintendo for their Super Nintendo World title and Rockstar Game’s GTA V Online.

What some might be inclined to call a 3D scanner is actually a technology known as photogrammetry. Each Raspberry Pi is attached to a DSLR and mounted all around a custom frame. The cameras are oriented to point toward the same center point. Whatever you want to scan is placed in the center while the Pis capture the subject from every angle. All of the images are sent to processed by a dedicated machine, also inside the truck, which then stitches the images into a full 3D scan of the subject. The use of exceptionally bright lights prevents shadows forming in the scan, giving the best results possible.

This isn’t the first Raspberry Pi photogrammetry project we’ve covered but it is the first one we’ve found that uses hundreds of Pis packed into a mobile studio. The Scan Truck has plenty of samples on its website showcasing the end results of their hard work but not much is known about the software side.

To get a closer look at Hollywood’s finest Raspberry Pi photogrammetry rig, check out the official website for The Scan Truck and see what they’ve been up to.

The Nvidia GeForce RTX 3060 Ti brings the entry price for Ampere GPUs down to $400, because apparently the GeForce RTX 3070, GeForce RTX 3080, and GeForce RTX 3090 weren't selling out fast enough. Maybe that's not fair, but round four of the Nvidia Ampere architecture seems destined to repeat recent history and sell out within minutes of launch. The fact that this is now the best price to performance ratio in GPU land makes it one of the best graphics cards, and it currently occupies seventh place on our GPU Benchmarks hierarchy, not counting any Titan cards. Maybe if you've been really good (or you're really lucky), you'll be able to procure one this side of 2021. [Update: Yup, the RTX 3060 Ti sold out fast. Here's our advice on how to buy an RTX 3060 Ti (or other new Nvidia GPU).]

Not a whole lot has changed relative to the RTX 3070. Both the RTX 3060 Ti and 3070 use the same GA104 GPU, but the 3060 Ti has fewer functional units enabled. You get the same 8GB of GDDR6 memory, clocked at the same 14Gbps. Meanwhile, the GPU core has 38 SMs enabled, versus 46 SMs on the 3070, making the 3060 Ti theoretically around 17 percent slower. Clock speeds also factor in a bit, as the boost clock on the 3060 Ti is 3.5 percent lower, so in general, the largest performance gap will be about 20 percent — and as we'll see, it's often quite a bit less than that.

The specs tell the same story relative to the previous generation Nvidia Turing GPUs. The 3060 Ti has quite a bit more theoretical compute performance than the RTX 2080 Super, and more than doubles the compute performance relative to the RTX 2060 Super. We know GPU prices are all kinds of screwed up right now, but it's worth remembering that the 2060 Super had the same official $399 price as the 3060 Ti. Power requirements are slightly higher than the 2060 Super, but lower than the 2080 Super.

Of course, theoretical specs don't always jive with real-world performance. One of the big changes with Ampere is that the INT32 cores can now do FP32 calculations as well. However, there's a decent amount of INT32 work in most games, so a good chunk of the potential FP32 performance is used up elsewhere. Still, with the other architectural changes, we expect the 3060 Ti to perform roughly on par with the 2080 Super, which means it will also be significantly faster than the 2060 Super.

Meet the GeForce RTX 3060 Ti Founders Edition 

There are no surprises in regards to the design and aesthetics of the GeForce RTX 3060 Ti Founders Edition: It's identical to the RTX 3070 FE, except for the small text that now says RTX 3060 Ti. It's a reasonably compact card that will fit in most cases, measuring 9.5 inches long. If you're doing a mini-ITX build, plan accordingly, but most ATX and micro-ATX cases should be fine.

With a reduced TDP of 200W, and considering what we saw from the 3070, cooling and noise shouldn't pose any problem. We'll get to the hard numbers later, but let's just say that the 1665 MHz boost clock is very conservative at stock, and there's a decent bit of overclocking headroom as well.

Connectivity consists of the same triple DisplayPort 1.4a outputs and single HDMI 2.1 port that we've seen on other Ampere Founders Edition models. You can use up to four monitors, and there's some ventilation on the rest of the IO bracket.

Somewhat interesting is that even the RTX 3060 Ti Founder Edition continues to use Nvidia's 12-pin power connector. Again, we're not quite sure what the point is, particularly since it just takes a single 8-pin PEG connector on one end and passes that through to the 12-pin connection. Even better is that Nvidia had concept renderings of the 3070 with an 8-pin connector. It doesn't really matter, as the adapter cable is included in the box, but it's also completely unnecessary. The 3080/3090 at least changed the PCB and have the 12-pin connector mounted in such a way that it uses a bit less area. Not surprisingly, none of the AIB partners seem interested in pushing the 12-pin agenda, so if you really don't like it, there will be plenty of other options.

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

I didn't have good luck with the MSI Afterburner frequency scanner this round — it kept failing after 10-15 minutes or so of testing. That put us back to manually overclocking, which in practice usually means similar results, just with a bit more fiddling. We start by looping the Unigine Heaven 4.0 benchmark in a window at 1440p extreme, then bump up RAM clocks in 250 MHz increments until the GPU crashes or visual anomalies appear. Then we go back to the previous 'safe' level and repeat the same process for GPU clocks, except with 25 MHz increments.

We were nearly able to max out the GDDR6 overclock at 16.5-17Gbps, but the EDR feature (Error Detection and Retry) meant performance was actually slower than a lesser RAM OC. Eventually, we stuck with a 1000 MHz overclock on the RAM (the best-case maximum would be in the 1000-1250 MHz range, based on our testing), giving an effective data rate of 16Gbps. For the GPU, 150 MHz was nearly stable but crashed in two tests, so we dropped down to a 140 MHz overclock that completed the full test suite.

Besides changing the memory and GPU clocks, we also increased the power limit to the maximum 110 percent and adjusted the fan curve. That last one is critical for maintaining a stable overclock, as sticking with a default fan curve tends to lead to periodic crashes. Basically, we set the fan to a 50 percent minimum and gave it a steady ramp from 50 percent at 40C up to 100 percent at 80C. We could certainly do more tuning, but that was sufficient to keep temperatures in check and still wasn't overly loud.

Overclocked results are present in our standard benchmark charts, but not in the bonus extended tests or DirectX Raytracing (DXR) suites. We have several AIB 3060 Ti cards we'll be looking at in the future that have factory overclocks, so if you want something a bit more potent than the reference design, stay tuned.

Nvidia GeForce RTX 3060 Ti — Test Setup 

As with the recent Radeon RX 6800 XT and RX 6800 review, we have multiple test suites we'll be running. The baseline suite consists of the same games we've been using throughout 2020: nine games, none of which were released this year, which means drivers should be good and optimized by now. We ran the stock and overclocked GPU through this suite.

Our second extended test suite includes the nine legacy games and then adds two ray tracing enabled games (Dirt 5 and Watch Dogs Legion) — that's a heavily AMD-biased game, with an Nvidia-influence game, just to keep things a bit equal. Worth noting is that the DXR support in Dirt 5 is currently in beta/preview, so take it with an extra scoop of salt. Two more games from 2020 (Assassin's Creed Valhalla and Horizon Zero Dawn) complete the suite, with both being AMD promoted games. It doesn't seem to matter much on HZD, but Valhalla very much favors AMD GPUs right now. This extended test suite is only run on the latest generation GPUs (RTX 30-series and RX 6000 series), but we also run it on a Core i9-10900K PC and a Ryzen 9 5900X PC to show CPU scaling. However, there are differences in motherboard and RAM, which appear to at least partially influence the final results.

If the extended suite seems a bit AMD slanted, we make up for that in the DXR suite. Here we've benchmarked 10 games and tests, most of which are Nvidia promoted. That's largely because prior to the RX 6800 launch last month, AMD didn't have any ray tracing capable GPUs, but there's certainly potential for the games to be more heavily optimized for Nvidia's RTX hardware. The DXR tests are run on our standard i9-9900K PC.

We've run all of the tests at 1080p, 1440p, and 4K ultra. We've also run the GeForce RTX 3060 Ti with DLSS enabled where possible (in Quality mode if it's an option) on the DXR suite, to show how that affects things. Spoiler: It helps performance a lot. Given this is a $400 graphics card, 1080p and 1440p are the primary targets, but 4K isn't out of the question, so we'll go through all three resolutions. If you want medium quality results, we'll be updating our best graphics card article shortly with the full standard suite of results at all six settings/resolution combinations. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

9 Game Average

Starting with the 1080p combined results, which are the average of the nine individual games, the RTX 3060 Ti looks quite nice. At stock, it's slightly faster than the RTX 2080 Super and a solid 34 percent bump in performance relative to the 2060 Super. It's also just 7 percent slower than the RTX 3070 while costing 20 percent less. Overclocked, it comes pretty darn close to matching the 3070, but then the 3070 can also be overclocked.

We won't bother commenting on the individual gaming results and just let the charts speak for themselves. There are a few games that are clearly running into CPU limits, and some of the games tend to favor AMD more while others lean toward Nvidia. In general, though, there's not a huge difference in standings. For 1080p in particular, this is probably about as high up the GPU ladder as we'd recommend going — unless you're playing a ray tracing game, of course.

Borderlands 3

Division 2

Far Cry 5

Final Fantasy XIV

Forza Horizon 4

Metro Exodus

Red Dead Redemption 2

Shadow of the Tomb Raider

Strange Brigade

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

Stepping up to 1440p, there's not a huge change in rankings, but the difference between the various GPUs becomes a bit more pronounced. That's because CPU bottlenecks are largely gone at 1440p, except perhaps for the absolute fastest GPUs (like the RTX 3090). 

9 Game Average

Nvidia claims the RTX 3060 Ti is faster than the outgoing RTX 2080 Super, and that continues to be the case. The aggregate score puts it 5 percent ahead of the 2080 Super, and 40 percent ahead of the 2060 Super — not bad for a single generation upgrade. Overclocking improves performance another 8 percent, putting it just behind the stock 3070. It's going to be interesting seeing what AMD does with lower-tier RDNA2 offerings, as right now the RX 6800 is just 18 percent faster than the 3060 Ti. The rumored RX 6700 XT will supposedly have 40 CUs and 12GB of memory, which would be a pretty significant cut relative to the 6800. 

Borderlands 3

Division 2

Far Cry 5

Final Fantasy XIV

Forza Horizon 4

Metro Exodus

Red Dead Redemption 2

Shadow of the Tomb Raider

Strange Brigade

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

4K is more of an academic question for most gamers buying a $400 GPU, but the 3060 Ti still does as well as the 2080 Super. If you're willing to turn down a few options, you can still hit 60 fps in most games. Just don't plan on playing maxed-out settings with ray tracing at 4K and 60 fps. 

9 Game Average

The margin of victory at 4K doesn't change much with the 3060 Ti. It's still 5 percent faster than the 2080 Super, and 44 percent faster than the 2060 Super. Overclocked, it's only 2 percent behind the stock RTX 3070. It's also 19 percent slower than the RX 6800, but that's fine considering it costs over 30 percent less. Of the nine games we tested, the 3060 Ti manages more than 60 fps at 4K in four games. We'll look at a slightly wider selection below. 

Borderlands 3

Division 2

Far Cry 5

Final Fantasy XIV

Forza Horizon 4

Metro Exodus

Red Dead Redemption 2

Shadow of the Tomb Raider 4

Strange Brigade

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

As with the RX 6800 XT and RX 6800 review, our second suite of gaming benchmarks includes four newer games from 2020, and it's run on two additional CPUs, the Core i9-10900K and the Ryzen 9 5900X. These are pretty much the three fastest gaming CPUs currently available, and other options (e.g., the Ryzen 9 5950X or Core i9-10850K) wouldn't change the overall picture much. Actually, outside of a few games, the CPU doesn't really matter too much. Core i9-9900K remains plenty powerful, so if your primary focus is gaming and you already have a fast Intel 8th or 9th gen Core i7 or better — or an AMD Ryzen 5 3600 or better — you won't miss much.

We're going to group all three resolutions into each gallery this time to keep things from becoming too bloated. Just swipe through the galleries to check out the 1440p and 4K results, as well as the fps percentile line charts. Note once more that the RAM configuration isn't identical on the three platforms, so there's a bit of variation in how each CPU performs. We'll provide individual game commentary this time, as there's a lot more going on.

We're only looking at the latest generation Ampere and RDNA2 (aka Big Navi) GPUs, so it's no surprise that the RTX 3060 Ti falls to the bottom of these charts. That's not because it's slow so much as it's slower than the other options. Hopefully, the more mainstream pricing means that Nvidia will have a lot more GPUs to go around, though we'll be surprised if inventory lasts more than a few minutes. After Black Friday and Cyber Monday, where even previous generation cards were selling for more than 25 percent over the launch pricing, it's clear that demand for gaming GPUs is at an all-time high.

Looking at the overall results, the 3060 Ti ends up trailing the RTX 3070 by anywhere from 6 to 11 percent, depending on the CPU and resolution. Most of the individual results are about what you'd expect, but there are a few anomalies that we'll point out below. Something else to mention is that the RTX 3060 Ti breaks 60 fps at 1440p in all but two of the 13 games tested — and those two exceptions had ray tracing enabled.

We're going in alphabetical order, so Assassin's Creed Valhalla is the first in our extended test suite. This is a new game, promoted by AMD… and damn if AMD doesn't crush Nvidia here, at least when we're looking at the 6800 series. We expect things will improve over the coming weeks, and this is a DirectX 11 game, so drivers can affect performance more than in DX12 or Vulkan games. It's also quite the demanding game for the 3060 Ti, just barely managing to break 60 fps at 1440p. Of course, we don't have a direct AMD competitor for the 3060 Ti yet, so we'll have to wait and see how this changes over time. 

Borderlands 3 is another AMD promotional game, and this time we're using the DirectX 12 (DX12) API. There's not much to say about the 3060 Ti here, though it's interesting that AMD's new RX 6800 GPUs (and the RTX 3090) do slightly better on our Intel CPUs than on the 5900X, while the other GPUs don't seem to care much about the processor. 

Huge caveat here: Dirt 5 is an AMD promotional game and we're using a beta / early access DXR patch. This is one of just two games that are promoted by AMD and feature ray tracing effects, the other being Godfall. But Godfall doesn't support DXR on Nvidia GPUs yet, so this is our only real look at how an AMD-optimized DXR game might behave. Again, big scoops of salt, but AMD GPUs come out ahead in most instances relative to their Nvidia counterparts. The RTX 3060 Ti also only breaks 60 fps at 1080p on the 9900K, and not by much.

Speaking of which, something else that's interesting is how much better the Nvidia cards do on the Core i9-9900K compared to the other two platforms. Chalk it up to beta wonkiness, and let's also note that the ray traced shadows don't particularly change the appearance of the game much (swipe through the gallery to see the comparison screenshots). It looks more like a contrast filter has been applied with DXR enabled, and the performance hit is lower than many other DXR games as well.

We're on a bit of an AMD promotional roll, and The Division 2 is another game with AMD branding. We test using DX12, and this is an old enough game that both AMD and Nvidia GPUs appear to have received plenty of optimization work. 1080p is a bit more CPU limited this time, and the 5900X comes in ahead of the 9900K, with the 10900K mostly coming in on top, at least for the Nvidia GPUs. For AMD GPUs, there's a bit more variation at 1080p, with the 6800 XT doing best on the 5900X. That's probably just the margin of error in the testing, though. 

Far Cry 5 is one of the oldest games in our test suite, and while it's yet another AMD promoted game, it's DX11 only, and Nvidia GPUs tend to have a slight advantage. It's also very much CPU limited at 1080p and even 1440p on the faster cards, with only 4K ending up with the expected GPU groupings. AMD's new GPUs also top out at a lower average fps, suggesting there are some driver inefficiencies at play. 

Another relatively old game, Final Fantasy XIV, is our first Nvidia promoted game — or at least, it uses a few of the GameWorks libraries for graphics effects. This time, AMD's 5900X CPU clearly comes out on top for most of the GPUs, particularly at 1080p. Minimum fps is also much lower than in other tests, mostly due to the way the standalone benchmark works (lots of scene changes that can cause fps dips). Anyway, even at maximum detail, the 3060 Ti and every other card easily break 60 fps, making this one of the lighter games in our test suite. 

This is technically a GPU-agnostic game that's not specifically promoted by AMD or Nvidia, but Forza Horizon 4 definitely favors AMD GPUs. They have higher maximum fps and better minimum fps as well, potentially thanks to the DX12 code. The game also tends to be pretty CPU-limited on the higher GPUs at 1080p and 1440p, with the 5900X and 10900K outperforming the 9900K. 

Horizon Zero Dawn came out on PS4 back in 2017 but only arrived for PC this past August. It's based on the same game engine as Death Stranding and uses the DX12 API. Given its console heritage, you might expect this one to favor AMD GPUs a bit, and it does. Or at least, AMD has more consistent fps and fewer oddities. 

For example, the 3060 Ti with the 10900K outperforms the 3070 with 5900X and 10900K. We may need to retest (again), but we've run out of time for the initial launch review so we'll just mark this down as another 'questionable' result. (That's why some of these games are expanded coverage and not part of our traditional test suite, which we're working on updating.) 

Metro Exodus was one of the first games with ray tracing support, but we're just using the default ultra preset here. It's an Nvidia promoted game, thanks mostly to the RTX enhancements, though the 3080 does come out ahead of the 6800 XT here. The 3060 Ti does fine up to 1440p, but 4K proves to be a bit much. 

Red Dead Redemption 2 is our first Vulkan API result — it supports DX12 as well, but our previous testing showed that Vulkan performs best in most cases. The GeForce RTX 3060 Ti again does well up to 1440p, but starts to struggle at 4K. The three CPUs all perform similarly, and the lengthy benchmark sequence tends to vary a bit more than in other games, so we wouldn't worry too much about the 1-2 fps difference between runs. 

Shadow of the Tomb Raider is another DXR-enabled game that we're testing without ray tracing (but we'll include the RT results below), and it's an Nvidia promoted game thanks to the DXR and DLSS support. There's quite a bit of noise in the charts at 1080p, where the game is mostly CPU limited and the 9900K at least holds back a few of the GPUs. Things settle down at 1440p and 4K, with the 3060 Ti once more coming up a bit short of 60 fps. 

Strange Brigade is our other Vulkan API game, and this time it favors Nvidia's GPUs (even though this is an AMD promotional game). Of course, it's mostly academic as all of the cards hit 100 fps or more even at 4K, and everything is in the 200+ fps range at 1080p. 

Last but not least, Watch Dogs Legion is our second DXR enabled game for the expanded test suite. This time, we're looking at an Nvidia promoted game, and patches and driver updates have fixed AMD's earlier rendering problems. (At launch, the RX 6800 cards weren't rendering all of the ray tracing effects properly and performed about 40 percent better than the new results shown here.) 

With the full DXR rendering now in place, Big Navi doesn't look so hot. In fact, the RTX 3060 Ti basically ties the RX 6800 XT at 1080p and 1440p, and we won't worry much about the 4K results considering nothing is even remotely playable — at least not without DLSS enabled. How does the 3060 Ti with DLSS perform? We'll get to that next. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

Given what we've seen so far, it's a safe bet that the RTX 3060 Ti will be about as good as the RTX 2080 Super when it comes to ray tracing performance. We've tested most DXR-capable GPUs from AMD and Nvidia, omitting only the Titan RTX and the non-Super 20-series GPUs.

We've also tested the RTX 3060 Ti with DLSS enabled on the games that support DLSS. That's seven of the ten DXR games, which makes sense as DXR can prove to be a bit much for most of the GPUs running at native resolution. Note that the composite score for the 3060 Ti DLSS does not include results from the other three tests at all, so it's not fully comparable, but it's an interesting data point regardless.

We'll go with the same format as the extended test suite, so 1080p, 1440p, and 4K results are lumped together in each gallery. Swipe to view the additional charts, and as mentioned earlier, all of these tests were run on the Core i9-9900K PC. Again, since Nvidia was the only ray tracing option until last month, most of the games are Nvidia promoted titles. They're also inherently optimized more for Nvidia's hardware.

Overall, ray tracing continues to be brutally demanding, particularly when using native rendering. On average, the 3060 Ti just barely breaks 60 fps across the test suite at 1080p. It also beats the RX 6800, illustrating quite clearly how much faster Nvidia's ray tracing hardware is compared to Nvidia's Ray Accelerators. At 1440p, even with DLSS enabled, the 3060 Ti can't average 60 fps, though it's still quite a bit faster than the PS5 and Xbox Series X when it comes to ray tracing workloads. And 4K … well, the less said, the better. With DLSS, 4K averages just over 30 fps — though it does boost performance by over 50 percent. 

 Our first DXR 'game' is the 3DMark Port Royal test, which definitely isn't a game. It does use multiple ray tracing techniques, including shadows and reflections, and performance is generally a bit lower than our composite score. Interestingly, AMD's RX 6800 does better in 3DMark than in most other DXR games, though that could be due to UL taking a more agnostic approach to specific GPU optimizations. 

Boundary isn't an actual game, at least not yet, but it does use a bunch of ray tracing effects, and it's extremely demanding. Without DLSS, only the 3080 and 3090 can break 60 fps, even at 1080p. With DLSS Quality mode enabled, the RTX 3060 Ti joins those cards. It also basically ties the RX 6800 XT. Based on what we're seeing right now, the actual game will probably need to have DLSS Balanced or DLSS Performance mode enabled to get smooth framerates. 

Call of Duty: Black Ops Cold War uses ray tracing effects for shadows, with a higher fidelity (and more demanding) result in the previous Modern Warfare (2019) reboot. The 3060 Ti breaks 60 fps at 1080p native, and DLSS can further improve its performance. At 1440p, DLSS Quality mode still manages a decent 67 fps, and the 3060 Ti continues to lead AMD's options. The 6800 XT does squeak ahead at 4K, probably thanks to its 16GB of memory, but we doubt many people are going to want to play multiplayer Call of Duty at 30 fps. 

Control remains our favorite example of what ray tracing can add to a game, thanks to the office environment with lots of vertical mirrored windows. Ray tracing is also used for diffuse lighting and ambient occlusion, and the result is once again the 3060 Ti leading AMD's 6800 XT. DLSS Quality meanwhile improves performance by more than 50 percent, so even 1440p is very playable. 

Crysis Remastered uses some funky hybrid of DirectX 11 code with some hacks to get ray tracing working. Our initial look at the ray tracing setting didn't impress us too much (most of the scenes hardly change when disabling the RT setting). Still, it's extremely demanding, particularly at the 'Can It Run Crysis?' setting. There's no DLSS support, but the 3060 Ti does break 60 fps at 1080p, while 1440p is moderately playable, and 4K generally proves to be too much for today's hardware. 

We already looked at Dirt 5 with DXR above, but this time we have a few additional GPUs. The game is clearly optimized more for AMD's hardware right now, and the RT effects aren't particularly noteworthy. But then, we could say the same about many of the RT effects. Most people will likely prefer running with higher fps and fewer stutters more than the enhanced visuals. 

We didn't even max out the ray tracing options in Fortnite, as it's already brutal — we're basically using medium ray tracing. At 1080p with DLSS Quality, the 3060 can break 60 fps, as can the 3090 (without DLSS). And that's about it. DLSS does nearly double the fps for the 3060 Ti at 1080p (and actually does double it at 1440p and 4K), so the 3080 and 3090 might be able to manage a playable result at 1440p with DLSS enabled. Unless you're just kicking around some creative mode worlds, though, you'll probably want to skip the visuals and go with more competitive high fps settings. 

Metro Exodus is pretty old now, and it only supports DLSS 1.0. There are some weird rules about what GPUs can enable DLSS at various resolutions, and you can see that at 1080p – DLSS isn't enabled. It provides a minor bump in performance at 1440p and a larger improvement at 4K, but if you're hoping to run Metro at maxed-out settings and 4K with ray tracing enabled, you pretty much need the 3080 or 3090. Interestingly, the RX 6800 comes out ahead of the 3060 Ti here, perhaps because RT is only used for global illumination. 

Shadow of the Tomb Raider is another DLSS 1.0 game, and you can't use DLSS at 1080p on the 3060 Ti. DLSS also results in lower minimum fps at 1440p since it adds to the processing time of the scenes and can create a bit of stutter, but it does better at 4K. With only one RT effect in use (shadows), AMD's GPUs again rank higher than in many of the other more complex ray tracing games. 

We looked at Watch Dogs Legion in our CPU scaling suite, and here you can see additional GPUs as well as DLSS  on the 3060 Ti. The newcomer basically matches AMD's 6800 XT, with DLSS pushing it into the lead. DLSS is also the only way to get above 60 fps with the 3060 Ti. Like Control, there are quite a few vertical reflective surfaces that actually help make the ray tracing effects more noticeable, but the performance hit is still quite steep. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

The RTX 3060 Ti has a relatively tame power rating of just 200W. Even overclocked with the power limit set at the maximum 110 percent isn't going to be too bad, as that's only around 220W. For these tests, we run Metro Exodus at 1440p ultra, without ray tracing or DLSS. We also use FurMark running at 1600x900 in a window using the stress test mode. We'll look at power, GPU clocks, temperatures, and fan speeds.

We also measured the noise levels of the RTX 3060 Ti from a distance of 15cm. At idle, from that distance, our PC's noise floor is 46.0 dB (or 34 dB from where my head is relative to the case on the floor). After running Metro for 15 minutes, the 3060 Ti FE noise levels were barely above that — 46.9 dB. For reference, that's basically the same result as the 3070, which you'd expect considering the two share the same core design.

Clock speeds, temperatures, fan speeds, and power are all interrelated. Drop the clocks, and you reduce the power and temperature. Raise the fan speed, and you reduce the temperature and maybe even power use, potentially allowing for higher clocks. It's a four-way balancing act, and utilities like EVGA Precision X1 and MSI Afterburner allow you to customize your particular card (within limits).

As we've seen with several other Nvidia GPUs, power use ends up being slightly higher than the official TDP. I don't think anyone will cry foul about a card using 205W instead of 200W, but it's interesting that the 3070 so far is the only Ampere GPU that came in below its official TDP (while gaming). Overclocked, the 3060 Ti actually exceeded the 3070 in power use. 

Nvidia tends to be quite conservative with its advertised boost clocks, and the RTX 3060 Ti Founders Edition proves that yet again. The official boost clock is only 1665 MHz, but during our Metro Exodus test, the GPU averaged 1900 MHz. Overclocking pushed that up to 2070 MHz. FurMark does drop the GPU clock just a bit below the rated boost clock (1623 MHz), but it's higher than the base clock of 1410 MHz. Much as you'd expect, our 150 MHz overclock also translates into a 155 MHz average clock speed increase for FurMark. 

Temperatures for the RTX 3060 Ti FE are good as well. Most GPUs can happily run at 70C, and that's basically where the card ends up at stock. Because of the higher fan speeds we applied during overclocking, that actually drops the temperature (but increases the noise levels). Both Metro and FurMark top out at around 69-70C.

Interestingly, the 3060 Ti's fan speeds are slightly higher than the 3070. That could just be variance on the cards and fans, or perhaps more likely, the GA104 silicon in the 3060 Ti was binned and determined to be a better fit for a lower-tier card, with the higher-quality silicon going into the 3070 cards. At stock, the 3060 Ti fans ran at around 1380 RPM in Metro and 1440 RPM in FurMark. Our aggressive fan speed curve for overclocking bumped the RPM up to around 2350. We could certainly tune things better, but this was sufficient for our testing purposes. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

Now that we've seen the RTX 3070 along with AMD's RX 6800, we have a better feel for the GPU market we're likely to see during the coming year. AMD doesn't have a viable alternative to the GeForce RTX 3060 Ti at the $400 price point yet, so for now, this is the undisputed upper mainstream champion. Some might say $400 isn't really a mainstream GPU, but given the sales of previous-gen GTX 1070 and RTX 2060 / 2060 Super cards, there are clearly plenty of people willing to spend that much money on a graphics card. And to be clear, this is a great card for that price.

The RTX 3060 Ti is about 35-40 percent faster than the previous-gen 2060 Super and about 30 percent faster than AMD's RX 5700 XT — and that's without even getting into ray tracing or DLSS support. In ray tracing games, the 3060 Ti lead grows to 40-45 percent over the 2060 Super, and AMD's RX 5000 series doesn't even have the option to try and run DXR games. Even against AMD's new RX 6800, though, the RTX 3060 Ti delivers better overall DXR performance — and can improve by around 30 percent on average for games that support DLSS, using DLSS Quality mode. Plus, Cyberpunk 2077 will launch in about ten days [not that I'm counting, #fanboy] with plenty of ray tracing effects and DLSS 2.0. The RTX 3070 is supposed to be good for 1440p ultra with ray tracing, or 4K ultra without ray tracing, which means the 3060 Ti should be able to handle those settings as well. And thankfully, unlike graphics cards, digital downloads of a game aren't going to sell out.

Which brings us to the problem of supply. We'll be shocked if RTX 3060 Ti doesn't sell out just as fast as the 3070, 3080, 3090, 6800, and 6800 XT. Hell, we've seen people buying used previous-gen cards off eBay for more than their original launch price, which is sad. It's not quite as bad as the GPU shortages during the great cryptocurrency mining craze of 2017, but it's not much better. And hey, Bitcoin is flirting with $20,000 per BTC again, so maybe there's still hope. (/sarcasm, if that's not clear.)

Anyway, Nvidia's CEO said Ampere supply likely wouldn't keep up with demand until 2021. That actually makes plenty of sense, as it takes about 5-6 months from placing an order for a bunch of chips (silicon wafers) to getting shipping graphics cards using those chips. Six months ago, we were right in the midst of the first COVID-19 surge, and there was a lot of uncertainty about the future. In retrospect, Nvidia could have doubled or even tripled its orders and probably been just fine, but we can't imagine what the reaction would have been had someone suggested that back in April! Hindsight, such a lovely thing.

The one potential bright side is that Nvidia should have more RTX 3060 Ti cards than all previous RTX 30-series parts combined. That's just our guess, but when you look at GPU sales, $400 cards sell way units more than $500 and $700 cards. Plus, we'd imagine there are plenty of GA104 chips that can't quite work as a 3070, so hopefully there will be enough to last more than a couple of minutes. (Side note: We've also received more RTX 3060 Ti AIB cards before the launch has even happened than any of the other RTX 30-series cards, which is potentially a good sign.)

If you already have an RTX 2080 Super or even a 2060 Super, there's not much need to upgrade to an RTX 3060 Ti. But if you purchased a GTX 1070 back in 2016 and you're still sitting on it today, you can now double your GPU performance for roughly the same price that the GTX 1070 launched at. Again, assuming you can find one in stock. As for AMD, an RX 6700 XT could be a formidable opponent, particularly in games that don't use ray tracing or DLSS. How long will we have to wait for those cards to arrive, and will AMD ship as many GPUs as Nvidia? After the RX 6800 launch, we're guessing the answer to the second part of that question is no, but we'll probably find out more in early 2021.

Right now, Nvidia just gave a nice boost in performance and features to the $400 market. The GeForce RTX 3060 Ti is a relatively small step down from the RTX 3070, with a larger step down in price. That hits the sweet spot in both price and performance — in fact, out of the current and previous-gen GPUs, it's the best overall card in price to performance ratio (fps per dollar). If you're hoping to upgrade to a new graphics card, the RTX 3060 Ti definitely belongs on your shortlist.

MORE: Best Graphics Cards

MORE: GPU Benchmarks

MORE: All Graphics Content

The Nvidia GeForce RTX 3070 Founders Edition is here — or at least it will be on October 29, and we can provide a full review today. Following in the footsteps of the GeForce RTX 3080 and GeForce RTX 3090, this is the third of the new Nvidia Ampere architecture GPUs. In terms of performance, price, and features, these new GPUs rank at the top of our GPU benchmarks hierarchy and are some of the best graphics cards right now. Except, they're still almost impossible to find in stock. Will the RTX 3070 suffer the same fate at launch, and how does it stack up against the other current and previous-gen GPUs? That's what we're here to find out.

Fundamentally, the GeForce RTX 3070 will be similar in a lot of ways to the RTX 3080 and 3090 … except, not really, because changes to the core counts, memory capacity, and memory speed can all have a pretty significant impact on performance. Nvidia claimed during the Ampere reveal in September that the RTX 3070 was "faster than the RTX 2080 Ti," which would be an impressive result if it's true. It has more cores and higher TFLOPS ratings, but less memory and bandwidth. Here's the block diagram for GA104 along with the full spec sheet for the Ampere GPUs, along with the RTX 2070 / 2070 Super that the newcomer replaces.

The RTX 3070 is both a big step up from the previous-gen cards it replaces, as well as a relatively large step down from the RTX 3080. On paper, it has about 32 percent less compute than the RTX 3080, and 41 percent less bandwidth. For comparison, the 2070 had 25 percent less compute than the 2080 but had the same bandwidth. Compared to the RTX 2070, however, the 3070 should be a significant jump in performance. It has over 150 percent more FP32 compute and 67 percent more RT TFLOPS, though it still only has the same 448 GBps of bandwidth. Going into the review, we were very curious to see how the 3070 would stack up against the previous-gen cards in real-world benchmarks, and Nvidia's claims of beating the RTX 2080 Ti seemed unlikely.

It's also interesting to note the massive difference in power requirements compared to the RTX 3080. There's a 100W gap in TDP, and we know from our testing that the RTX 3080 and RTX 3090 can use every bit of their power budgets. This is the first Ampere GPU where we have a basically direct TDP comparison point with the previous-gen Turing GPUs. Both the RTX 2070 Super and RTX 2080 have 215W TDPs, so this will finally be Nvidia's chance to prove that, yes, Ampere actually can be significantly more power efficient than Turing.

Meet the GeForce RTX 3070 Founders Edition 

The design and aesthetics of the GeForce RTX 3070 Founders Edition are similar in many ways to the RTX 3080 and 3090 Founders Editions — it's just nowhere near as large. You get the finned radiator look on the main body, with dual fans that include a built-in rim to improve static pressure, all in a dark gray color with silver highlights. There are several noteworthy differences between the 3070 FE and the 3080/3090 FE, however.

First, both fans are on the same side of the cooler (the side that normally faces down in a typical ATX case). There's also no 'X' silver cross on the side that faces up. Instead, a large grooved rectangular piece with a cutout at the back of the card allows the air to still flow through the radiator fins. Nvidia talked a lot about redesigning the 3080/3090 to handle the higher TDP. Obviously, the 3070 isn't in the same category of power and heat, and being a smaller card, it perhaps wasn't as beneficial. The card still runs cool and quiet, as we'll see later.

The other big change in terms of aesthetics is that there's no lighting at all, RGB or otherwise. The GeForce RTX logo on the top of the card is pretty tame compared to some graphics cards, though for some people, this would be a positive change. There's also only a small RTX 3070 logo on the side of the card, which is a bit weird to me as there's a ton of room for a larger logo. All of Nvidia's previous-gen 20-series Founders Edition models at least had a green GeForce RTX logo on the top edge, and the 3080 and 3090 have RGB logos (though the tool to configure the RGB lighting still hasn't been released). The RTX 3070 eschews all such nonessentials and just provides the basics as far as looks go.

There's not a lot to say about the rest of the card. It has the same triple DisplayPort 1.4a outputs and a single HDMI 2.1 port and can drive four monitors without trouble. The rest of the IO bracket is for the front fan's exhaust, but the bracket is silver instead of the matte black finish used on the 3080 and 3090 FE cards.

The RTX 3070 Founder Edition also uses Nvidia's new 12-pin power connector, but this time, instead of a y-combiner that takes dual 8-pin PEG connectors, it's just a single 8-pin port that's converted to a 12-pin port. Was this really necessary? I mean, I sort of like the extension cable because it's far easier to leave it connected and just detach the 8-pin connector. But then again, I swap GPUs pretty much daily, so my usage is not at all like a typical user. Also, the 12-pin connector is basically the same length as an 8-pin connector and is only a few millimeters narrower, but the orientation is such that putting on an 8-pin connector instead wouldn't have been a problem.

We haven't had time to do a full teardown of the 3070 Founders Edition yet, as we wanted to finish benchmarking before taking it apart. We'll update this section with additional images and commentary in the near future.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

I thought I'd try something new with the RTX 3070 FE overclocking. Rather than manually tweaking settings to try to find something that appeared stable, I used the OC scanner functionality built into MSI Afterburner (4.6.3 beta 2). You can access this by pressing Ctrl+F. I increased the power limit to 109 percent, the maximum allowed, and increased the GDDR6 clocks by 500MHz (15 Gbps effective). Increasing the core voltage didn't appear to help at all, as the GPU appears to run into power limits, so I left that alone. I also tweaked the fan curves to be a bit more aggressive, so the fan will run at 100 percent if the temperature reaches 80C — but don't worry, it didn't come anywhere near that mark.

With those settings in place, the OC scanner generated the above voltage-frequency curve. What you'll immediately notice, hopefully, is that the gains at the far right of the voltage range are very limited, while gains at lower voltages are quite a bit higher. Basically, anything above 1075mV ends up with the same 'stable' result. Overall, the RTX 3070 was only about 3-5 percent faster when overclocked, and a lot of that is thanks to the power limit increase and higher memory clocks. This is similar to what I've seen on the other 30-series Ampere GPUs so far.

Overclocked results for the 3070 are present in all of the charts, but not in the bonus gaming tests. Other RTX 3070 cards will likely hit higher clocks than the Founders Edition, and this is a pretty conservative overclock. However, it's also an overclock that the card can probably maintain for many years without difficulty.

Nvidia GeForce RTX 3070 — 1080p Gaming Benchmarks

The GeForce RTX 3070 is in a bit of an interesting situation as far as resolutions and performance go. On the one hand, it's supposed to be as fast as the 2080 Ti, and we generally don't think people with that sort of GPU are going to play games at 1080p. At the same time, it's a $500 instead of a $1,200 card, and some people prefer higher refresh rate 1080p displays. What I'm getting at is that, even though this still isn't really a card built for 1080p gaming, we're going to go full monty and show all the benchmark results this time. Yeah, that includes RTX 3090 and 3080 at 1080p as well.

So, disclaimers out of the way, let's hit the benchmarks. Our conclusion will focus more on the 1440p and 4K performance than 1080p, and all of the bonus games will only be tested at 4K due to time constraints and intended usage. Still, if you're wondering how the 3070 and other cards handle 1080p, we've got you covered.

Our test PC remains the same as in recent GPU reviews. We'll probably be looking to change things up when Zen 3 launches, based on the early rumors, but the Core i9-9900K remains plenty fast for now. We did do CPU scaling tests on the RTX 3080 and found that the Core i9-10900K was a bit faster in a few instances, but nothing particularly noteworthy.

9 Game Average

We'll skip commentary on the individual games at 1080p, but the overall results are telling. At 1080p ultra, the 3070 FE comes in just behind the RTX 2080 Ti, but it's still 15 percent faster than the 2080 Super, 24 percent faster than the 2070 Super, and 38 percent faster than the 2070 FE that launched two years ago.

1080p medium narrows things down even more, and the RTX 3070 does squeak past the 2080 Ti, thanks to slightly better minimum fps. However, many of the games run into CPU bottlenecks, and the 3080 even comes out ahead of the 3090 at 1080p. Not that you should use either one for 1080p unless it's with ray tracing or maybe VR, or in a game where you could push 360 fps on a 360Hz monitor, maybe.

Borderlands 3

The Division 2

Far Cry 5

Final Fantasy XIV

Forza Horizon 4

Metro Exodus

Red Dead Redemption 2

Shadow of the Tomb Raider

Strange Brigade

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Okay, let's move on to the more meaningful tests for the RTX 3070. While the card can certainly still handle 4K at playable framerates, like the RTX 2080 Ti, there are going to be plenty of situations where you'll have to tune and tweak the settings to get a smooth 60 fps. 1440p gaming, on the other hand, shouldn't pose any problem to the card. 

9 Game Average

Nvidia says the 3070 is faster than the 2080 Ti, but that's only in select games. At 1440p ultra, it's more like a tie — there was only one game in our test suite where the 3070 won by more than a hair. Still, even matching the 2080 Ti is no mean feat, especially considering the price difference. Of course, now 2080 Ti cards will likely fall to much lower prices on places like eBay, but historically speaking, Nvidia's new architectures tend to improve with time, so the 3070 could end up being a bit faster than the 2080 Ti down the road.

That's interesting to see, considering the rather large memory bandwidth deficit. Overclocking doesn't make much of a difference either, so perhaps the 3070 is mostly held back by the GPU and not the GDDR6 memory. Regardless, this is a good showing for the Ampere architecture. It's as fast as the previous generation 260W TDP card while using 40W less power. And it costs $500, which means plenty of gamers can actually afford it, whereas the $1000-$1200 pricing of the 2080 Ti was out of reach for most people.

Borderlands 3

There's not much to say about the individual games. Borderlands 3 reaches very playable framerates of 85 fps at 1440p ultra and 153 fps at 1440p medium. Compared to the RTX 2070, it runs 50-65 percent faster — or 35-45 percent faster than the 2070 Super. Not that you'd really need to upgrade from either of those GPUs just yet. However, if you're sitting two generations back, the 3070 is now more than twice as fast as the 1070 and basically double the performance of the Vega 56. 

The Division 2

The Division 2 puts the 2080 Ti ahead of the 3070 at all settings, but never by much. At 1440p, 2080 Ti is about 3-5 percent faster than 3070. Meanwhile, the 3070 is 30-35 percent ahead of the 2070 Super, 40-55 percent ahead of the 2070, and 110-135 percent ahead of the GTX 1070. Vega 56 definitely widened the gap relative to the 1070 over time, so the 3070 is 'only' 60-90 percent ahead of Vega 56. 

Far Cry 5

Far Cry 5 is getting a bit old, and it definitely hits some CPU limits at 1080p. That starts to fade away at 1440p ultra, but 1440p medium showed relatively weak minimum fps on the overclocked 3070 for some reason. Anyway, we're looking forward to Far Cry 6 next year, so we won't spend any more time analyzing this 2.5 years old game. 

Final Fantasy XIV

This is another older game that hits pretty high framerates, and it's one of the few instances where even the stock 3070 leads the 2080 Ti. By less than half of a percent, of course, so it's basically just a tie. This is a less demanding game, so the gaps between the various GPUs aren't quite as large here. We'll be switching to something else (Horizon Zero Dawn, probably) in the near future. 

Forza Horizon 4

Forza Horizon 4 is the biggest lead we saw for the 3070 over the 2080 Ti, to the tune of up to four percent at 1440p. It's not much, but it's something. It's another game we're looking to retire from our GPU tests, mainly because of how much of a pain the Microsoft Store is to deal with. Project CARS 3 is the likely replacement, which we'll hit in our bonus games below. 

Metro Exodus

We didn't enable ray tracing or DLSS (1.0) for Metro Exodus, but it's still a very demanding game — we'll check ray tracing and DLSS performance in the bonus round. The 3070 still runs very well at 1440p ultra and enabling RT and DLSS should still result in 60 fps gameplay, but the 2080 Ti ends up being 3-5 percent faster yet again. Elsewhere, the 3070 again leads the 2070 Super by around 30 percent, and is up to twice as fast as the GTX 1070. 

Red Dead Redemption 2

Red Dead Redemption 2 runs into CPU bottlenecks with the fastest GPUs, even at 1440p medium, but pushing up to ultra settings spreads things out. While the 3070 leads at 1440p medium, the 2080 Ti ends up five percent faster at 1440p ultra. The 3070 also leads the 2070 by 50 percent, and is close to 2.5 times as fast as the GTX 1070 here. Note that we test using the Vulkan API, which tends to perform best for both AMD and Nvidia GPUs. 

Shadow of the Tomb Raider

Like Metro, we've tested without RT or DLSS enabled for our primary benchmarks. We'll have RT + DLSS results below in the bonus gaming section. At the highest preset, the 3070 cranks out 115 fps, just a bit behind the 2080 Ti yet again. Do keep in mind that the 2080 Ti FE has a 90 MHz factory overclock, but then so did most other 2080 Ti cards — not many people were willing to pay that much money only to get a stock-clocked card, it seems. 

Strange Brigade

Last, we have Strange Brigade, another game where we use the Vulkan API. It's pretty light compared to some of the other games in our suite, and the 3070 pushes well beyond the 144 fps mark that you'd want for a higher refresh rate 1440p display. Performance is again a bit behind the 2080 Ti, but well ahead of the 2070 Super and more than double the GTX 1070 and Vega 56. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Based on what we've seen already, the 4K results are likely to favor the 2080 Ti just slightly, and the 3070 will need to run many games at medium to high settings to get 4K performance above 60 fps. That's basically the same as the 2080 Ti, with a much more palatable price tag. Whether you'll be able to buy one later this week is another matter. 

9 Game Average

We'll just focus on the overall picture at 4K once again, as the individual results are below and mostly tell the same story. At 4K ultra, the 3070 averages 68 fps across our nine-game suite. That's good, but four of the games are below 60 fps, while two of the games run at more than 100 fps, somewhat skewing the results. 4K60 is within reach of the 3070, just not at maxed-out settings in all games. The 4K medium results meanwhile range from a low of 73 fps in Metro Exodus to a high of 140 fps in Strange Brigade, and every game breaks 60 fps.

Looking at the competition, the 2080 Ti is about three percent faster than the 3070, so again, the memory capacity and bandwidth deficit didn't really factor into things as much as we expected. The 3070 is also 35 percent faster than the 2070 Super, 53 percent faster than the 2070 FE, 130 percent faster than the GTX 1070, and 95 percent faster than the Vega 56.

If you're running a card that's three or more years old, upgrading to the 3070 should provide a significant boost to performance. If you have anything more recent — at least in the high-end category of GPUs — you can probably stick with whatever you have for a couple of more years.

Borderlands 3

The Division 2

Far Cry 5

Final Fantasy XIV

Forza Horizon 4

Metro Exodus

Red Dead Redemption 2

Shadow of the Tomb Raider

Strange Brigade

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

As we did on the GeForce RTX 3080 and GeForce RTX 3090 reviews, we've run a bunch of additional benchmarks at 4K and maxed out settings, with ray tracing and DLSS enabled where applicable. Well, except for 3DMark Port Royal, which runs at 1440p by default, but who's counting? We're only testing a subset of GPUs here, basically RTX 2080 and above — AMD GPUs can't run these same benchmarks in many cases since they don't support ray tracing or DLSS yet, and they're not going to support DLSS ever, most likely.

That's a big problem, actually, as DLSS 2.x really does work well and makes Nvidia's RTX cards far more capable of running many games at higher resolutions. Is 4K with DLSS the same as 4K native? No. But if 4K DLSS looks similar and performs much better, it's the clearly superior choice that most gamers would want to use. Even if AMD's RX 6800 XT and other Big Navi GPUs can match the RTX 3080 in standard performance metrics, not having a comparable alternative to DLSS 2.x is going to be a concern.

There may only be 20 or so games with DLSS right now, but Unreal Engine 4 has apparently made it quite easy to switch on DLSS — there are a few demos we're looking at here that use Unreal Engine, and even the solo developer effort Pumpkin Jack managed to enable ray-traced shadows, reflections, diffuse lighting, and DLSS thanks to Unreal Engine. Considering how many games use variants of Unreal Engine, we'll likely see even more DLSS enabled games in the coming years. 

15 Game Average

Looking at the combined results of all 15 games/benchmarks, the results are very close to what we saw with our primary test suite. The RTX 3070 ends up 28 percent faster than the RTX 2080 FE, and one percent slower than the RTX 2080 Ti. It's also 25 percent slower than the RTX 3080 and 34 percent slower than the RTX 3090. Average fps is just above 60 for the 15 tests, but individual results will definitely dip below that mark in some games. 

3DMark Port Royal

We're going in alphabetical order, so 3DMark Port Royal is the first in our list of bonus 'games.' It's quite demanding, averaging just 38 fps at native 1440p rendering, and the 3070 is about 29 percent slower than the 3080 here. It's also 8 percent slower than the 2080 Ti, which is a larger gap than we've seen in any actual games. 

Battlefield V

Battlefield V was the first game to release with ray tracing and DLSS support. RT is only used for reflections, and this is DLSS 1.0 — meaning you can't even use DLSS with higher-end Nvidia GPUs at less than 4K. Still, we enabled ultra ray tracing and DLSS, and that's enough to average 64 fps on the 3070. Like Port Royal, the 3070 is about 25 percent slower than the 3080 and 7 percent slower than the 2080 Ti, and it's only 15 percent faster than the RTX 2080. 

Bright Memory

Bright Memory Infinite is a benchmark for a game that's currently in development. I have no idea what the game is about, but it looks absolutely amazing. There are multi-bounce reflections, refractions, caustics, and more at the highest settings, and even with DLSS in performance mode (4x upscaling from 1080p), the RTX 3070 only manages 38 fps. Using the high preset doesn't help much, but the good news is that the game supports DLSS 2.0, so running at 1440p with 4x upscaling ends up being a far better result. Interestingly, while the 3070 is 28 percent behind the 3080, it's also 17 percent faster than the 2080 Ti at maxed-out settings. Considering the number of RT effects being used, future games may also favor the 30-series by a wider margin. 

Boundary

It's a different developer, but Boundary otherwise has a lot in common with Bright Memory Infinite. Both use Unreal Engine, support a bunch of RT effects, DLSS 2.0, and will eventually be full games on Steam. Maybe. I'd say Bright Memory Infinite currently looks better, or at least the benchmark does, but we'll have to see what the final games are like — and if they're available in English (both devs are Chinese). As for performance, Boundary is perhaps slightly more demanding, depending on the settings used. With DLSS in performance mode, the 3070 averages 39 fps, but minimums are only 23 fps. 

Call of Duty: Modern Warfare

Call of Duty: Modern Warfare (2019) doesn't support DLSS and only uses ray tracing for shadows. It makes a difference, but it tends to be less noticeable than some of the other RT options. Also, it still drops performance quite a bit. In our benchmark sequence, which uses the single player mode, performance is just a bit shy of 60 fps. In multiplayer, especially frantic shootouts, performance can be quite a bit lower. Most competitive players will turn down a lot of the effects in this game, though playing at 1440p or 1080p should also boost framerates a lot. 

Control

Up to now, Control remains the best implementation of ray tracing and DLSS effects in a shipping game. It has reflections and transparent reflections, diffuse lighting, and ambient occlusion. We tested in both DLSS Quality (1440p) and DLSS Performance (1080p) modes. The 3070 comes out slightly ahead of the 2080 Ti in both cases, and is about 35 percent faster than the RTX 2080. 

Crysis Remastered

Can it run Crysis Remastered? At 4K with maxed out settings, including ray tracing, the answer is no — your PC cannot handle it. Even the RTX 3090 gets just 41 fps with ray tracing enabled, while the 3070 manages a rather pathetic showing of 24 fps, and that's still faster than the 2080 Ti. Crysis Remastered lives up to its progenitor's legacy, though I'm not convinced the ray tracing effects even look that good. It's also using some weird hybrid ray tracing solution that runs on top of DirectX 11. Even without ray tracing, the 3070 plugs along at 47 fps, just 1 fps behind the 2080 Ti. If you want to play Crysis Remastered with RT enabled, plan on running at 1440p or 1080p. 

Doom Eternal

Sort of the opposite end of the spectrum from Crysis, Doom Eternal rips and tears along at an impressive 115 fps on the 3070. This is also a game that tends to like memory bandwidth, so it's no surprise that the 2080 Ti ends up 12 percent faster than the 3070. The 3070 does beat the 2080 FE by 37 percent, at least. 

Death Stranding

Sorry, Death Stranding, no all-caps for you! But DLSS in Quality mode looks and runs great. The 3070 gets over 100 fps, and even the 2080 gets 86 fps. Norman Reedus would be proud of the little guy (little relative to the 3090 at least). 

Flight Simulator

Microsoft Flight Simulator is another game that likes lots of VRAM, so the 2080 Ti posts a larger lead over the 3070 again. The 3070, meanwhile, is only 13 percent faster than the RTX 2080 this time. Also, it can't break 30 fps in our test sequence, which consists of the autopilot bringing in a plane for a landing at my local airport. This is a brute of a game to run at ultra settings, needless to say. 

Horizon Zero Dawn

There are no fancy RT or DLSS effects in Horizon Zero Dawn, but it does look quite nice. The 3070 and 2080 Ti are basically tied in average fps, but the 3070 has lower minimum fps — probably due to having 8GB VRAM instead of 11GB. A few tweaks should get 4K performance above 60 fps. 

Metro Exodus

Metro Exodus was already demanding without enabling its ray traced global illumination. Even with DLSS enabled (DLSS 1.0, sadly — I'd love to see 4A Games release a patch that supports DLSS 2.1 instead!), the 3070 only manages 45 fps, about 6 percent behind the 2080 Ti. 

Project Cars 3

We looked at Project CARS 3 at launch, and we'll probably replace Forza Horizon 4 with the game in the near future. It's newer and more demanding, though it's missing a few features like TAA. It's also one of the games where the 3070 comes out a few percent ahead of the 2080 Ti for a change. 

Shadow of the Tomb Raider

The shadows in Shadow of the Tomb Raider don't really look that much better with ray tracing, but it was at least a somewhat clever marketing schtick. Performance takes a pretty big hit, even with DLSS 1.0 enabled, and the 2080 Ti again gets a relatively large win over the 3070. 

Wolfenstein Youngblood

Last on our list, we have Wolfenstein Youngblood, another game that uses the Vulkan API and the only game we're testing that uses VulkanRT. Performance is very high, even with DLSS in quality mode, and the 3070 comes out ahead of the 2080 Ti by 6 percent — and 9 percent with DLSS performance mode. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The first two RTX 30-series Ampere GPUs pushed power ratings to the highest levels we've ever seen for a 'reference' design. RTX 3070 dials things back to relatively sane levels of just 220W, basically matching the 215W TDP of the RTX 2080 and RTX 2070 Super. 

It's a bit surprising that Nvidia only uses a single 8-pin PEG connector to feed the 12-pin connector, though, since the 8-pin is rated for 150W. Add in 75W from the PCIe slot, and it's basically right at the limit — the 2070 Super and 2080 both had an 8-pin plus 6-pin PEG configuration. At stock, the GeForce RTX 3070 Founders Edition should be within spec, but overclocking might have some oddities to investigate further.

For these tests, we run Metro Exodus at 1440p ultra without ray tracing or DLSS. We also use FurMark running at 1600x900 in a window using the stress test mode. We'll look at power, GPU clocks, temperatures, and fan speeds. We also measured noise levels of the RTX 3070, from a distance of 15cm. At idle, from that distance, our PC's noise floor is 46.0 dB (or 34 dB from where my head is relative to the case on the floor). After running Metro for 15 minutes, the 3070 FE noise levels were barely above that — 46.8 dB. Needless to say, the fans are doing an excellent job of cooling without making a lot of noise.

Clock speeds, temperatures, fan speeds, and power are all interrelated. Drop the clocks, and you reduce the power and temperature. Raise the fan speed, and you reduce the temperature and power use, potentially allowing for higher clocks. It's a four-way balancing act, and utilities like EVGA Precision X1 and MSI Afterburner allow you to customize your particular card (within limits).

Power use is just a bit below the rated 220W during Metro Exodus testing, while FurMark comes in at 224W. That's for stock operation. Overclocked with the power limit raised by 9 percent, power use reached 238W during Metro and 245W with FurMark.

Incidentally, that also pushes the card beyond the 225W theoretical limit of an 8-pin connector and the PCIe slot. Our PCIe slot measurements did go just above 75W during the overclocked tests, reaching 78.8W, while the maximum from the PCIe slot was 70W at stock. Meanwhile, the 8-pin connector had a peak power draw of 178.4W — even stock saw the 8-pin hit a peak of 165.3W, but we're not particularly worried about that since the power cable has two 8-pin connectors.

Average GPU clocks for the RTX 3070 Founders Edition are above the advertised boost clock, but what's interesting is that they're also slightly below the measured clocks for the RTX 3080. The boost clock is officially 1730 MHz on the 3070 and 1710 MHz on the 3080, but the 3080 ran at 1876 MHz average compared to 1863 MHz average on the 3070. FurMark flips the tables, with the 3070 averaging 1595 MHz and the 3080 getting 1576 MHz — both below the rated boost clock but above the base clock (which Nvidia doesn't actually list these days).

With the overclocked settings, the 3070 also averaged nearly 2GHz, coming in just 37MHz short. It periodically boosts above 2000 MHz but won't maintain those clocks — at least with the Founders Edition and a 220W + 9% (240W) power limit. Some third-party cards will very likely stay above 2GHz while gaming.

The RTX 3070 FE at stock runs pretty cool, reaching a peak temperature of 69C after a couple of minutes of gaming. The average temperature during our Metro testing was 66C, but note that the longer you game, the higher the average will get. When overclocked, the peak temperature and average temperatures were basically unchanged, but that's because we applied a custom fan curve that was a bit more aggressive. In FurMark, clock throttling kicks in to keep temperatures in check, with a peak stock result of 68C and an overclocked result of 70C, with average temperatures of 66.5C and 68C, respectively. 

Here you can better see the results of our fan speed customization with the overclocked 3070. It settled in at about 1600 RPM when overclocked, compared to 1300 RPM at stock. Also, note that the GTX 1070 is getting old and can be a bit of a curmudgeon, and the fans appear to be working extra hard to keep it running at around 75C. A good deep cleaning of the radiator fins and fan blades would likely help — something to keep in mind if you have an older GPU that's overheating or running loud.

In general, the GeForce RTX 3070 Founders Edition performs well, and the new cooler and fan design get the job done. Considering this is a much smaller card than the 3080, the thermal and noise results, in particular, are welcome. You could definitely fit the card into a smaller case, though we'd still recommend something with a decent amount of airflow. If you want a cooler-running RTX 3070, third party cards will definitely fill that role — some companies are basically taking their RTX 3080 coolers and slapping them on a 3070, which will provide ample cooling for the lower-power GPU.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The GeForce RTX 3070 Founders Edition is everything we expected. It's a lower power card with a smaller footprint, and it basically trades blows with the previous generation king of the hill, the RTX 2080 Ti. Two years later, and $500 now potentially gets you the same performance as the old $1,200 GPUs. If there's one constant in the world of GPUs, it's the ever-increasing performance at any given price point. But we're in the midst of a lot of GPU stuff, and without seeing what AMD's Big Navi brings to the table, it's impossible to give a final verdict for the RTX 3070.

AMD will spill the beans on Big Navi tomorrow, but we don't expect the hardware to arrive for review until mid-November or later. That's fine because, in all likelihood, almost no one is going to be able to buy an RTX 3070. Just like the RTX 3080 and RTX 3090 sold out within minutes (or seconds) of launch, we expect the initial batches of RTX 3070 will go fast. Historically speaking, that's what always happens for new high-end cards. RTX 2070, 2080, and 2080 Ti sold out at launch; GTX 1070, 1080, and 1080 Ti also sold out at launch; and so did GTX 970 and 980. The question isn't so much whether the new GPUs will sell out, but rather how long they'll be harder to purchase.

Launching in the midst of the COVID-19 pandemic, plus the holiday shopping season, is only going to exacerbate things. That's why Nvidia's CEO already said supply likely wouldn't keep up with demand until 2021. Hopefully, the supply of the RTX 3070 is better than it's bigger siblings — it should be since it's a smaller chip that will likely yield better. Plus, Nvidia knows it sells more $500 GPUs than it does $700 and $1000+ GPUs, so it almost certainly ordered more GA104 chips than GA102.

Besides AMD's incoming RX 6000 series, you can also count on Nvidia releasing RTX 3060 or RTX 3060 Ti or something similar in the not-too-distant future. It's basic silicon manufacturing: The GA104 chip has up to 48 SMs, 46 of which are enabled on the RTX 3070. While there will be some chips that are fully functional and could run with all 48 SMs, only selling 46 SM chips means overall improved yields. But inevitably, there will be a bunch of chips that can't meet the 46 SM requirement, so there will be another graphics card that's a step down from the 3070, and probably one more that's a step down from there. That's where the RTX 3060 Ti and RTX 3060 come into play. What we don't know is exactly when they'll launch or how much they'll cost, but we'd bet on a $399 replacement for the RTX 2060 Super, and probably a $349 or $299 replacement for the RTX 2060. It all hinges on how well AMD's RX 6000 does.

Speaking of which, early indications of Big Navi and RX 6000 are promising. There's a good chance AMD will have something that comes pretty darn close to the RTX 3080, or maybe even exceeds it, and then there will be lower-tier parts trickling down from there. The catch is that most of the rumors and purported leaks are for games running without ray tracing. AMD will support DirectX 12 Ultimate, which includes ray tracing and VRS (variable rate shading), as well as VulkanRT. What we don't know is how fast AMD's chips will be in such workloads. The general opinion is that AMD won't match Ampere when it comes to RT performance, which could prove to be increasingly important, what with the next-gen consoles adding RT support, not to mention games like Cyberpunk 2077.

While ray tracing is arguably going to become even more important in the coming years, I think the bigger problem for AMD is DLSS. AMD has made no mention of anything equivalent to the Tensor cores present on Turing and Ampere RTX GPUs. For any game that supports DLSS — and the list is growing — Nvidia will probably offer a superior experience. Given a choice between 4K DLSS running at 75 fps or 4K native with TAA running at 45 fps, as an example, I'd take the DLSS option every time.

Right now, there are at least 19 games that support DLSS (Anthem (1.0), Battlefield V (1.0), Bright Memory (2.0), Control (2.1), Death Stranding (2.1), Deliver Us the Moon (2.0), F1 2020 (2.0), Final Fantasy XV (1.0), Fortnite (2.1), Justice (1.0?), Marvel's Avengers (2.1), Mechwarrior 5 Mercenaries (2.0), Metro Exodus (1.0), Minecraft RTX (2.0), Monster Hunter World (2.0), Pumpkin Jack (2.0), Shadow of the Tomb Raider (1.0), Watch Dogs Legion (2.1), and Wolfenstein Youngblood (2.1) — I might have missed a couple). DLSS 2.x integration appears to be quite easy, relatively speaking, since it doesn't require game-specific training from Nvidia's supercomputer, and support from Unreal Engine makes it almost silly not to support DLSS.

Besides those already available (or available tomorrow, if you count Watch Dogs Legion), at least 14 more games are coming that support DLSS 2.x (Amid Evil, Atomic Heart, Boundary, Call of Duty Black Ops, Cyberpunk 2077, Edge of Eternity, Ghost Runner, JX3, Mortal Shell, Mount & Blade 2: Bannerlord, Ready or Not, Scavengers, Vampire the Masquerade — Bloodlines 2, and Xuan-Yuan Sword VII). Several of those games aren't even on my radar, but a handful are right at the top of most 'anticipated games' lists.

Let me put it another way: If you give me an option of basically equivalent image quality, with 50-70 percent higher performance, what games would I not want to support that? Maybe pixelart indie games, because it could throw off the aesthetic (and they already run fast enough), but everything else? Yes, please, I'd like some DLSS 2.x. There are games where DLSS lets the RTX 2060 perform more like a 2080 Super, and the 2080 Super ends up looking like an RTX 3080. I didn't think DLSS 1.0 was all that great, but DLSS 2.0 changed things. It's not that it always looks better (sometimes it does, sometimes it doesn't), but it almost always runs a lot faster.

It's possible AMD will have some DLSS alternative that works equally well available at some point. I'd really love to see that happen because TAA tends to be way too blurry, and FidelityFX/CAS, while good, isn't quite at the level of DLSS 2.x. But then AMD will have to get developers to utilize its DLSS alternative, which puts it about two years behind Nvidia's RTX and DLSS efforts. For games that don't support ray tracing or DLSS, it's easier to see how AMD can win the coming battle. But pricing is still likely to be key.

On that point, we don't expect AMD to severely undercut Nvidia. It will simply try to offer an attractive option. The reason is that Zen 3 looks ready to clean Intel's clock come November. Even the Ryzen 9 5950X only uses two 7nm Zen 3 chiplets, each of which measures just 75-80mm square. That's an $800 chip that only requires at most 160mm square of TSMC N7 manufacturing. Sure, CPUs are different than GPUs, but TSMC is basically at capacity for how many wafers it can produce each month. Given the size of Zen 3, and an estimated size of 500-536mm square for Navi 21? AMD will want to charge quite a bit for the GPUs, as much as it can reasonably manage. That or it would just produce more Zen 3 chips, which will probably sell like hotcakes.

The bottom line is that we can't declare a winner right this moment. Nvidia's Ampere RTX 30-series GPUs are potent, and the RTX 3070 brings new levels of performance to the $500 market. We expect to see 30-series parts push down into the $300-$400 range in the coming months as well. AMD's Big Navi is more of a wildcard since we don't quite know what to expect in terms of ray tracing performance or DLSS alternatives. AMD may have as many as four Navi 2x GPUs launching in the next month or two (or three or four), also with prices ranging from perhaps $250 up to $600 or more.

If you're already set on going with Nvidia and don't want to spend more than $500, you can try to pick up an RTX 3070 on Thursday. If you're willing to spend a bit more money, we'd argue the added VRAM, bandwidth, and performance of the RTX 3080 means it's the better option at $700 — not that you can find RTX 3080 in stock, but you can keep trying. For the undecided, we suggest waiting to see what happens with Big Navi, and of course, those who prefer AMD GPUs will want an RX 6000 regardless of how it stacks up.

Whatever your GPU choice, it's an exciting time after about two years of waiting for the next big thing to arrive. Of course, there will be even faster options next year, so if you already have a capable GPU, just enjoy it until it's no longer sufficient. Maybe by then, we'll be talking about RTX 6070 and RX 9000.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The versatility of the Raspberry Pi never ceases to amaze us--today's project reaches into the field of photography. Maker Benjamin Bezine's project, RoboScan uses a Raspberry Pi to automatically scan old rolls of analog film to develop them into individual digital photos.

According to Bezine, this Pi creation was developed as a lockdown project. In addition to the Raspberry Pi, it uses a DSLR camera to capture individual photos on the roll. The frame is constructed using every developer's foundational tool, Lego.

Bezine integrated machine learning with the help of Google's Coral Tensor Processing Unit to help determine when a photo is within the proper frame boundaries. He also added a web interface that can be used to both control and monitor the system while it scans.

Photos are automatically saved to the Pi as they are scanned. Users can then use  Samba to sync the photo data to a remote PC or other storage device. Bezine reported a performance output of around 6 frames per minute, including the time it takes to sync the photo to his PC.

You can check out this project in action and read more in the full post on Reddit. If you want to recreate it at home, check out the source code on GitHub. Visit our list of Best Raspberry Pi Projects for more cool creations from the Pi community.

Shanghai Micro Electronic Equipment (SMEE) is reportedly on track to deliver its second-gen deep ultraviolet (DUV) lithography scanner by the fourth quarter of 2021, according to a media report. The tool can produce chips using 28 nm process technologies and relies on components produced in China and Japan. Therefore, the tool does not rely on devices made in the USA, which is increasingly important amid the ongoing trade war between China and the U.S that has found the US barring China-based companies from purchasing some types of chip fabrication equipment. 

There are a number of fairly competitive producers of semiconductors in China that make chips developed in the country using fabrication technologies designed in Tianxia. But all of these companies use production equipment developed and made in other countries, such as Japan, the Netherlands, and the U.S.

As part of its multi-faceted plan to make its semiconductor industry self-sufficient, China is not only encouraging chip development and local manufacturing, but also supports the fabrication of semiconductor production equipment. 

Founded in 2002, SMEE is a highly-integrated developer and manufacturer of semiconductor production equipment (and a provider or support services) that makes a broad range of products that includes scanners and inspection tools. Today, SMEE's most advanced devices are its 600-series scanners that can be used to make chips using 0.28-micron (280 nm), 0.11-micron (110 nm), and 0.09-micron (90 nm) process technologies.  

The top-of-the-range SMEE SSA600/20 machine is an immersion deep ultraviolet lithography tool that is equipped with a 193-nm argon fluoride (ArF) laser. Companies like Intel and TSMC started to use immersion DUV lithography back in 2004, so the SSA600/200 can hardly be called a leading-edge piece of equipment.

The successor of the SSA600/20 machine will continue to use an ArF light source, but for considerably thinner process technologies. That upcoming scanner promises to be advanced enough to make chips using a 28nm process technology, according to Verdict. Evidently, these scanners could be used for 40 nm-class, as well as 55nm/65nm fabrication processes, which are quite popular for multiple applications. By 2023, SMEE wants to produce machines good enough for a 20 nm node, the report says. The upcoming scanners are said to use certain components made in Japan, but they don't use any ingredients from the USA. 

TSMC adopted 28 nm process technology back in 2011 using tools from ASML, so even when SMEE ships its 28 nm-capable scanner in Q4 2011, it will still be more than a decade behind the world's No. 1 supplier of lithography tools. Meanwhile, 28 nm process technology is used rather widely today and will continue to be used for many years to come for chips that do not need FinFET transistors. For example, TV maker Konka Group last month disclosed plans to build a $4.5 billion semiconductor industrial park with the local authorities in Nanchang, eastern Jiangxi province. TVs and consumer electronics hardly need leading-edge process technologies, so a 28nm node might be good enough for them. Therefore, SMEE's upcoming DUV tool promises to become an important workhorse for SMEE's customers.  

Shanghai Micro Electronic Equipment has been producing scanners for years, so it is likely that it will be able to assemble a significant number of its next-generation scanners to equip an advanced fab. Meanwhile, chipmakers that have already adopted their 28 nm process technologies using scanners from ASML and Nikon will likely have to redesign their nodes to use SMEE's new tools. Therefore, while advanced Chinese scanners are about a year away, it will take quite some time before they will be widely adopted by the local semiconductor industry.  

SMEE's next-generation ArF DUV scanner is certainly a step towards China's self-sufficiency in the semiconductor industry. Meanwhile, there is one more critical element that the country yet has to develop in a bid not to depend on technologies from overseas: All advanced electronic design automation (EDA) tools that are crucial for modern chip development come from companies based in the USA. Until Chinese companies can create tools that are competitive with those from Cadence, Mentor Graphics, or Synopsys, all the chips developed in China will be designed using software from the U.S.

Nvidia's automatic overclocking algorithm, OC Scanner is getting a major rework in the way it runs and the way it's incorporated into overclocking utilities such as MSI Afterburner and EVGA's Precision X1 software. Fixing a lot of issues the previous OC Scanner had. MSI's latest patch notes for Afterburner version 4.6.3 Beta 2 showcase all the new changes made to Nvidia's automatic overclocking program.

• OC Scanner is now integrated into Nvidia's drivers, instead of being integrated into the overclocking utility. This allows Nvidia to update OC Scanner on the fly without needing to consult programmers for each overclocking utility and get them to update their software. Third-party software will communicate with OC Scanner via the NVAPI.
• MSI Afterburner will support both versions of OC Scanner, depending on what graphics drivers you are using. you'll need Nvidia driver 456.xx or newer to use the new OC Scanner implementation.
• OC Scanner now supports automatic overclocking for both GPU and graphics memory.
• OC Scanner will save and apply its overclocks on a driver level and automatically apply them on reboot. So leaving your overclocking utility running in the background is no longer necessary.
• Manual adjustment of the voltage/frequency curve is no longer available. (i.e. this applies to OC Scanner overclocks only, you should have manual control of the voltage/frequency curve still.)

These changes fix a lot of headaches with the original OC scanner, namely better stability and identical overclocking results across all overclocking utilities which was not the case with the original implementation.

On a side note, this new implementation of OC Scanner should have good implications on Nvidia's new Ampere GPUs as they are very power limited and struggle to maintain stability above 2GHz. GPUs inherently have excellent efficiency at lower clock speeds and voltages, as a result, OC Scanner will take advantage of this and keep your GPU frequency more stable and allow core clocks to drop less as you start hitting power limit or temperature limits.

You can try the new OC scanner yourself, all you need is an Nvidia 456.xx driver or newer plus MSI Afterburner 4.6.3 Beta 2 or the latest EVGA Precision X1 update. For now, it appears Ampere GPUs are only supported according to EVGA Precision X1's latest patch notes. Hopefully, this is temporary and support will extend to Turing GPUs as well.

There are four lights! Right? You can double-check and know for sure with this portable Star Trek inspired Raspberry Pi scanner created by Gabe Emerson of the saveitforparts channel on YouTube.

Emerson's handheld scanner is built on top of a Raspberry Pi and uses various sensors to scan for RF signals and thermal bodies. It even has a built-in speaker to play the radio signals out loud.

This was Emerson’s first foray into Raspberry Pi territory and the results were absolutely worth it. The test version of the Raspberry Pi scanner was housed inside a Tupperware container which he affectionately dubbed the world's first redneck tricorder and, according to Emerson, was able to accurately detect a Klingon trailer park.

He ordered more parts for the final build which ended up inside a clear waterproof shell. It has a display operated by the Raspberry Pi which puts out visual data for both the RF scanner and thermal camera. He also included a battery pack so the unit is completely portable.

In the video, he experimented with an environmental sensor which reads both temperature and humidity but didn't have much luck, but this was likely due to a bad sensor module. There are plans in the works to update the scanner with more sensors and functions. In the meantime, check out his work on the official saveitforparts website. You may enjoy his steam-powered RC goose project.

At CES 2020, EVGA announced a new SKU in its Nvidia Turing-based lineup, the RTX 2060 KO Gaming. The announcement coincided with AMD’s RX 5600 XT release later that month. At the time, the AMD card positioned itself to be a disruptive product at the $300 price point, with an aim to perform better than the more-expensive RTX 2060 while costing less. It's an important and hotly contested price segment among the best graphics cards. EVGA pared some features down on the RTX 2060 KO, offering basic I/O connectivity and a smaller heatsink in order to meet the $299.99 price point—$50 less than the launch price of the Founder’s Edition RTX 2060 (which Nvidia also dropped to $299).

Here, we’re looking at the EVGA RTX 2060 KO Ultra Gaming, which is still aggressively priced at $319.99, but comes with increased core clocks and better performance. At that price, it’s still one of the least-expensive RTX 2060s, and EVGA’s KO SKUs fired a preemptive shot across the bow of AMD's RX 5600 XT priced from $279.99 to $339.99.

Performance of our EVGA RTX 2060 KO Ultra Gaming card landed where we expected. That is, it’s faster than the RTX 2060, slower than the RTX 2060 Super (both Founders Editions) and trades punches with the higher-performing variants of the RX 5600 XT. The RTX 2060 KO Ultra is a very capable 1080p card that easily averages over 60 fps across all our test titles, except for Metro Exodus. Moving up in resolution but down to medium settings, we experienced similar results at 1440p with the EVGA RTX 2060 KO Ultra Gaming.

Below, we’ll take a closer look and see how the EVGA KO Ultra Gaming stacks up against the Gigabyte RX 5600 XT Gaming OC 6G, along with other comparable cards. We’ll also check how well the smaller cooler works with the TU104 die, as well as other performance details.

Features

As previously reported in January, our EVGA RTX 2060 KO Ultra Gaming uses the TU104 GPU— normally found in the RTX 2070 Super and RTX 2080/2080 Super—but it's cut down to match the RTX 2060 specifications (both the RTX 2060 and 2060 Super typically use TU106 silicon). It’s a reasonable assumption that these TU104-150 dies didn’t make the cut for higher-end cards, and here only 60% of the CUDA cores and six of the eight memory channels need to be “good” to work as an RTX 2060. Given the low price of both the cards in this KO series, the likely reason for using this die in an RTX 2060 cost savings—Nvidia likely has a stock of TU104 chips that couldn't meet the requirements of the higher end models.

The odd choice of underlying silicon here is an interesting fact, but it has few real-world implications. For gaming purposes, this is simply an RTX 2060. However, as other sites have shown, when rendering in Blender, the card performs better than its namesake, landing between the RTX 2060 FE and the RTX 2070 Super Founders Edition. Those considering this card for compute tasks like Blender may be wondering if there’s a way to tell before buying what die is under the hood. According to an email we received from EVGA, the answer no. However, an EVGA rep told us the KO card “...is and will continue to be TU104 for the foreseeable future”. That makes sense, as the PCB is physically different from a regular 2060 card. TU104 is larger and has more solder points than TU106, so manufacturers can't use the same board for both chips.

The die is manufactured using TSMC’s 12nm FinFET Nvidia (FFN) process, placing 13.6 billion transistors in a 545 mm² area. The RTX 2060 KO and KO Ultra feature a 30 SM configuration versus the RTX 2060 Super's 34 SMs and the RTX 2070’s 36 SMs. This yields 1,920 shaders, 48 ROPs and 120 TMUs. Since the Turing architecture scales with the number of SMs, all core compute values scale down with it. This means a 17% drop in ray tracing core count, tensor core count, TMUs and in the L0/L1 cache.

Clock speeds on the KO Ultra Gaming are set at 1,365 MHz base with a boost rated to 1,755 MHz. The latter is a 75 MHz bump over reference speeds (1,365/1,680 MHz base/boost respectively), and the clock speed increase will bring better performance than a reference-clocked card. On the memory side, the KO Ultra Gaming includes the same 6GB of Samsung-made GDDR6, which sits on a 192-bit bus. The memory speed of 14 Gbps (1,750 MHz) matches the reference part, as do most other cards from board partners. This configuration yields 336 GBps bandwidth.

The EVGA RTX 2060 KO Ultra Gaming specifies 160W for power consumption. This value is less than the 180W AMD RX 5700 and more than the 150W of the RX 5600 XT. Actual power use will vary between partner cards, due to higher clock speeds and where the power limit is set. Feeding power to the VRMs is a single 8-pin PCIe connector. EVGA recommends at least a 500W power supply.

The following table summarizes the specifications of Nvidia GPUs and the RTX 2060 KO Ultra Gaming:

Design

The EVGA RTX 2060 KO Ultra Gaming is a factory tweaked, dual-slot video card measuring 7.9 x 4.3 x 1.5 inches (202.1 x 111.1 x 39.6mm). The card is smaller than most full-size GPUs and sits flush with the top and I/O plate. Given the size, this card should be able to fit in many small form factor (SFF) systems as well as larger chassis. Be sure to check the space inside your case before buying.

EVGA’s cooling solution uses two 85mm fans and is designed to cool all critical components, including the GPU, VRAM and MOSFETs. The GPU makes contact with a copper plate and two copper heat pipes that move the heat away from the core. The memory and MOSFETs keep cool through thick thermal pads that make contact with the fins. In fact, the company doubled up the thermal pads on the three memory ICs. Using such thick thermal pads isn’t the most efficient way to remove heat from these components, and a fair amount will migrate into the PCB. The backplate doubles as protection and also uses thermal pads to help remove heat from the memory chips.

Compared to more expensive RTX 2060 SKUs, the smaller heatsink uses fewer heat pipes, likely also contributing to the cost cutting here. We’ll find later that while it was effective at keeping the critical components cool during testing, fan noise was louder than the competition. EVGA chose not to include RGB lighting, which also helps the company hit a lower price point.

The RTX 2060 KO Ultra Gaming sends power to the GPU and memory via a 4+2-phase VRM. Managing the power delivery is an ON Semiconductor NCP81611 four-channel controller. Handling the memory is a UPI uP1666Q controller. Feeding power to the board is a single 8-pin PCIe connector. Theoretically, this configuration allows up to 225W of power, well over the 160W power limit, giving the card plenty of power for stock and overclocked use.

The EVGA card has minimal video output options, consisting of one each for DVI-D, HDMI and DisplayPort. Compare this to the reference RTX 2060 with two DisplayPorts, one HDMI and DVI-D, as well as a USB Type-C VirtualLink port, and you see another area where EVGA cut corners to save money. That said, the three ports here should suffice for most of gamers.

How We Tested the EVGA RTX 2060 KO Ultra Gaming

Our current graphics card test system consists of Intel's Core i9-9900K, an 8-core/16-thread CPU that routinely ranks as the fastest overall gaming CPU. The MSI MEG Z390 Ace motherboard holds down the fort, with 2x16GB Corsair Vengeance Pro RGB DDR4-3200 CL16 memory (CMK32GX4M2B3200C16). Keeping the CPU cool is a Corsair H150i Pro RGB AIO, along with a 120mm Sharkoon fan for general airflow across the test system. Storing our OS and gaming suite is a single 2TB Kingston KC2000 NVMe PCIe 3.0 x4 drive.

The motherboard is running BIOS version 7B12v16. Optimized defaults were used to set up the system. We then enabled the memory's XMP profile to get the memory running at the rated 3200 MHz CL16 specification. No other BIOS changes or performance enhancements were enabled. The latest version of Windows 10 (1909) is used and is fully updated as of February 2020.

Our GPU hierarchy provides a complete overview of graphics cards and how the various models stack up against each other. For these individual third-party card reviews, we include GPUs that compete with and are close in performance to the card being reviewed. On the AMD side, that consists of the Gigabyte RX 5600 XT Gaming OC and a reference Radeon RX 5700. For Nvidia, we've included anEVGA GTX 1660 Ti XC, RTX 2060 Founders Edition and RTX 2060 Super Founders Edition for comparison.

Our list of test games is currently Tom Clancy's The Division 2, Ghost Recon: Breakpoint, Borderlands 3, Gears of War 5, Strange Brigade, Shadow of the Tomb Raider, Far Cry 5, Metro Exodus, Final Fantasy XIV: Shadowbringers, Forza Horizon 4 and Battlefield V. These titles represent a broad spectrum of genres and APIs, which gives us a good idea of the performance differences between the cards. We're using driver build 441.20 for the Nvidia cards and Adrenalin 2020 Edition 19.12.2 for AMD cards, although the 5600 XT was tested using 20.1.2 beta drivers.

We capture our frames per second (fps) and frame time information by running OCAT during our benchmarks. For clock and fan speed, temperature and power, we use GPU-Z's logging capabilities. We'll be resuming our use of the Powenetics-based system from previous reviews in the near future.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Looking at the 1080p ultra results, the EVGA RTX 2060 KO Ultra Gaming averaged 105 frames per second (fps) across all games. Every title except for Metro Exodus (55.7 fps) was well above 60 fps, with most significantly over 80 fps. Performance in our lighter titles (Strange Brigade, Far Cry 5, Final Fantasy XIV, Forza Horizon 4, and Battlefield V) pushed well over 100 fps.

Compared to the Gigabyte RX 5600 XT Gaming OC 6G, the AMD card wins in 6 out of 11 titles over the KO Ultra Gaming, though not by much. An average across all titles shows the overclocked KO is faster than the Gigabyte 5600 XT by 2%. This isn’t a big difference but it's more than the margin of error. Compared to the Founders Edition RTX 2060 (99 fps average) the KO card is 5% faster. Our KO Ultra gaming is also 5% and 6% slower (respectively) than an RTX 2060 Super Founders Edition and a reference AMD RX 5700.

Adding price to the variables, the $319.99 EVGA RTX 2060 KO Ultra Gaming is $20 higher than the Gigabyte RX 5600 XT Gaming OC 6G listed in the charts. The biggest difference users will notice between these cards isn’t the frame rates, but pricing and features. The Nvidia card supports hardware ray tracing while AMD does not. If you value this technology, the choice is obvious. But keep in mind the 2060 is the lowest-end card in Nvidia’s current RTX lineup, often struggling to run titles with ray tracing enabled and other settings set to high (unless you enabled DLSS, which has its own pros and cons). So if you truly want to enjoy a ray traced gaming experience, both in current and future titles, you may want a higher-end RTX card.

The Division 2

Ghost Recon: Breakpoint

Borderlands 3

Gears of War 5

Strange Brigade

Shadow of the Tomb Raider

Far Cry 5

Metro: Exodus

Final Fantasy XIV: Shadowbringers

Forza Horizon 4

Battlefield V

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Raising the resolution to 1440p and lowering image quality to medium, the EVGA card averages 103 fps across all games—2 fps less than the 1080p results. Here, all 11 titles are above 65 fps including Metro Exodus at 65.5 fps. Yes, Metro's ultra settings outweigh the reduction in resolution. The rest of the games are 80s-plus or well over 100 fps. Like the RX 5600 XT, some but not all titles could run using high or ultra settings. Shadow of the Tomb Raider and Metro Exodus will likely not be able to maintain a smooth gaming experience at these settings. Outside of that, the EVGA card pounds through this resolution at medium settings.

At the higher resolution, we saw similar performance gaps between all tested cards. The Gigabyte RX 5600 XT managed to close the meager 2% performance gap at 1080p, yielding effectively equal results—there's a one fps difference, now in favor of the 5600 XT. But whether at 1440p medium or 1080p ultra, few games show more than a 3% difference, which you’ll only notice in benchmarks.

The Division 2

Ghost Recon: Breakpoint

Borderlands 3

Gears of War 5

Strange Brigade

Shadow of the Tomb Raider

Far Cry 5

Metro: Exodus

Final Fantasy XIV: Shadowbringers

Forza Horizon 4

Battlefield V

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

We use GPU-Z logging to measure each card's power consumption with the Metro Exodus benchmark running at 2560 x 1440 using the default Ultra settings. The card is warmed up prior to testing and logging is started after settling to an idle temperature (after about 10 minutes). The benchmark is looped a total of five times, which yields around 10 minutes of testing. In the charts, you will see a few blips in power use that are a result of the benchmark ending one loop and starting the next.

We also use FurMark to capture worst-case power readings. Although both Nvidia and AMD consider the application a "power virus," or a program that deliberately taxes the components beyond normal limits, the data we can gather from it offers useful information about a card's capabilities outside of typical gaming loads. For example, certain GPU compute workloads including cryptocurrency mining have power use that can track close to FurMark, sometimes even exceeding it.Power Draw

Power Draw

Starting with the power chart for gaming, the EVGA RTX 2060 averaged 164W. This puts it between the Founders Editions of the RTX 2060 at 158W and RTX 2060 Super pulling 173W. The EVGA RTX 2060 KO Ultra is a 160W card, but the BIOS on it allows for a bit more during the gaming test. When comparing the KO Ultra to the Gigabyte RX 5600 XT Gaming OC 6G, we see the AMD card use a lot less power, pulling an average of 122W for similar performance. However, GPU-Z only gets chip power and not board power for AMD GPUs, which means actual power use is likely very similar to the 2060.

Our Furmark results show the KO Ultra averaging 150W, which is less than what we saw during the gaming test. The built-in power limits worked hard here, lowering the clock speeds and voltage to fit within the defined parameters of the card. The Founder’s RTX 2060 checked in at 159W and the RTX 2060 Super 174W—both results are like the gaming values. If we compare this to the Gigabyte RX 5600 XT Gaming OC 6G, the AMD card now uses more power, drawing 158W—a reversal from the more realistic gaming loads.

Temperatures, Fan Speeds and Clock Rates

To see how each video card behaves, like the power testing, we use GPU-Z logging in one-second intervals to capture data. These items are captured by looping the Metro Exodus benchmark five times, running at 2560x1440 and ultra settings.

Additionally, we also use FurMark to capture the data below, which offers a more consistent load and uses slightly more power, regardless of the fact that the clock speeds and voltages are limited. These data sets give insight into worst-case situations along with a non-gaming workload.

Gaming

Temperatures in our game tests peaked at 71 degrees Celsius on both the KO Ultra and the Founder’s Edition RTX 2060. The Founder's RTX 2060 Super peaked a couple of degrees higher at 73C, in part due to the higher power draw versus the other two cards. The results for all cards were well within operating specifications.

Fan speeds on the KO Ultra Gaming ramped up smoothly and averaged 2,250 RPM in the game test. Fan speeds were stable, only rising with temperature and not varying wildly. While speeds were stable when tested, the higher RPMs on the EVGA card made it louder than the two Founders Edition cards. The noise wasn’t intrusive, but it is clear the acoustic profile is different using the smaller heatsink.

Clock speeds on the EVGA RTX 2060 KO Ultra Gaming averaged 1,878 MHz during the final loop of Metro Exodus. This value is well over the rated boost of 1,755 MHz. The RTX 2060 FE averaged 1,831 MHz in this test, 55 MHz less than the overclocked EVGA. The higher clock speeds translate to the higher performance seen in the game tests.

I have to admit, I prefer the way Nvidia lists their clocks as compared to AMD. Although AMD lists a boost clock, it’s the maximum boost for the card whereas Nvidia lists a conservative minimum boost. All AMD cards we’ve tested have, so far, not been able to reach the listed boost clock while Nvidia cards often run at speeds well over the listed value.

FurMark

Temperatures when running Furmark ended up lower than when running a game. Our EVGA KO Ultra Gaming peaked at 67 degrees Celsiu,s which is four degrees less than the gaming result. The temperature is lower due to the card fighting to keep within its power limits, lowering both clock speed and voltage to compensate. The RTX 2060 FE peaked at 70 degrees Celsius. Although the cooler is smaller than on other EVGA models, it does a good job keeping the card running within specifications.

When looking at the fan speed results, not much changed here. With Furmark pumping out a more constant load, we saw more of a flat fan curve. The RPM on the EVGA card stayed around the 2,250 RPM mark. There is nothing out of the ordinary to see here.

The most interesting part of testing using Furmark is to see how hard the cards work trying to run without banging off the power limit. In this case, the EVGA RTX 2060 KO Ultra Gaming averaged 1,595 MHz—almost 300 MHz less than the average gaming clock speed. As is typical with modern Nvidia cards, throttling here is more significant than the AMD Navi-based cards tested.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Moving on to software, by design EVGA’s Precision X1 monitors and overclocks EVGA video cards. The application can overclock the core and memory speed (manually or using OC Scanner), as well as power and temperature thresholds via large sliders on the main (VGA1) page. Precision X1 also controls fan speeds manually and has the ability to set up a custom fan curve.

On the monitoring front, the software displays real-time graphs with critical GPU data. This includes GPU temperature, clock speeds for the memory and core, GPU memory use and a framerate chart. Overall, the software works well for its intended purpose.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

EVGA’s RTX 2060 KO Ultra is intended to steal some thunder from AMD’s RX 5600 XT. The end result is an improved RTX 2060 that matches the performance of AMD’s RX 5600 XT graphics cards. The performance difference is so small that In most games, users would be hard-pressed to notice a difference. Select games may show more than a 5% difference, but that goes both ways, leading to an overall tie in performance. The price points for the KO and KO Ultra cards are $299.99 and $319.99, respectively. Other RTX 2060s are priced higher and lose out on the performance per dollar metric the EVGA RTX 2060 KO offers, though there are other factors at play.

With this price drop comes a few changes. First, and generally not a big deal, is the I/O situation. While most Nvidia Turing RTX GPUs include multiple DisplayPorts, HDMI, (sometimes) DVI-D, and USB-C VirtualLink ports, the KO Ultra includes single DisplayPort, HDMI, and DVI-D ports. Along with cutting back on the I/O, EVGA uses a smaller heatsink that isn’t as good as some of the other larger heatsinks on competing models. This, in turn, causes the fans to run faster and make more noise. While the card isn’t loud, it’s louder than the RTX 2060 Founders Edition used in this comparison.

Around the $300 price point, the EVGA RTX 2060 KO Ultra Gaming does succeed at stealing some thunder away from AMD’s popular RX 5600 XT series cards. Without specifically checking what video card is in a PC, you can’t tell the difference in performance between the two. Although the cooling solution is louder than other cards we’ve tested, it isn’t a deal-breaker unless the system is intended for a quiet build like an HTPC. The RTX 2060 cards also support ray tracing, which is another potential deciding factor. Put simply, if you want to use it, currently the only choice is an Nvidia card, but the image quality improvement and subsequent performance loss may or may not be worth it to you.

The EVGA RTX 2060 KO certainly disrupts the market solely based on price. The EVGA KO SKUs are the least expensive RTX models, without any major compromises. While it doesn't score a  knock out or even a technical knockout against the AMD competition, it does squeak out a win by a very close judge's decision. It's a good performance option within its own ranks as well as against AMD’s RX 5600 XT. If you're looking for solid 1080p performance at the $300 price point and would like ray tracing capabilities, the EVGA RTX 2060 KO Ultra Gaming should be on the shortlist.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Security company Cofense has discovered a new phishing campaign that has been using CAPTCHA boxes to hide a fake Microsoft login page from secure email gateways (SEGs). 

CAPTCHA systems create challenges, such as having to identify a blurred word or an object in an image, that you have to complete before you can go on to the next page. These systems are meant to block malicious automated bots from visiting a page by only allowing humans to bypass them. 

However, the group behind this latest phishing campaign has turned the typical anti-bot usage of CAPTCHA on its head and is using CAPTCHA to stop malware scanners from checking if a web page was built to steal user credentials.

Many businesses use SEGs to scan incoming emails for malware or phishing attempts. However, SEGs are not sophisticated enough to pass CAPTCHA systems. As CAPTCHA systems weren’t typically used by malicious actors, the vendors of SEGs seem to have been caught off-guard in this case, as their malware scanner were unprepared to meet a CAPTCHA challenge.

"The SEG cannot proceed to and scan the malicious page, only the CAPTCHA code site. This webpage doesn’t contain any malicious items, thus leading the SEG to mark it as safe and allow the user through.," the Cofense researchers said. 

After the email receivers pass the CAPTCHA test, they are served with a fake Microsoft login page meant to capture their Microsoft account credentials.

The email delivering the malicious phishing link comes from a hijacked email account from avis.ne.jp and pretends to be a notification for a voicemail message. Ironically, both the CAPTCHA and the phishing page are hosted on Microsoft’s cloud servers. 

However, the attackers aren’t doing this just to spite Microsoft, but to make it more difficult for either humans or automated scanners to be able to tell that the page isn’t legitimate. SEGs typically look at the domain reputation of the links found in emails, so when the malware is hosted on Microsoft’s own public cloud servers, it passes the reputation test with flying colors.

Attackers are evolving their techniques and are continuing to integrate tools that would normally be used against them to deploy their own malware. Things such as cryptographic signatures, HTTPS encryption and now CAPTCHA systems are now increasingly used to bypass or prevent automated scanners from identifying malware. Even AI-powered anti-malware systems aren’t safe from this technique.

There's nothing better than unlocking an extra bit of 'free' performance from our components, but overclocking can be a daunting task for the uninitiated. Nvidia's auto-overclocking OC Scanner feature looks to simplify the process. Nvidia originally launched this feature with its RTX-20 series of graphics cards, but a new beta version of MSI's Afterburner software (Version 4.6.0 Beta 10) expands the feature to Nvidia's Pascal graphics cards.

Nvidia developed the Scanner API/DLL package so that its partners, such as MSI and EVGA, can bake the auto-overclocking feature into their graphics cards utilities. With the click of a single button, Scanner runs an arithmetic-based routine that evaluates the graphics card at various voltages and frequencies, testing stability along the way. The program runs as its own process, as we noted in our Nvidia GeForce 2080 Ti Founders Edition Review: 

Although Nvidia says the metric usually encounters math errors before crashing, the fact that it’s contained means the algorithm can recover gracefully if a crash does occur. This gives the tuning software a chance to increase voltage and try the same frequency again. Once the Scanner hits its maximum voltage setting and encounters one last failure, a new frequency/voltage curve is calculated based on the known-good results.

Our own testing with the GeForce RTX 2080 Ti found that the program works as advertised, though as per usual with any type of automatic overclocking feature, we met with more success through manual tuning.

Opening the feature up to the Pascal cards is a nice addition to MSI's Afterburner, and it's rational to expect other graphics card vendors to add Scanner to their utilities soon. Nvidia hasn't made an official announcement either, but we'll update as necessary.

Meet TU102 and GeForce RTX 2080 Ti

Nvidia calls Turing its biggest architectural leap forward in more than 10 years. To prove it, the company is showing off a compendium of forward-looking capabilities that speed up performance in today’s games, introduce artificial intelligence to desktop graphics, make real-time ray tracing viable for the first time, accelerate video streaming, and support the next generation of VR hardware.

But there’s one problem with taking a victory lap before the opening bell rings: expectations get set very, very high.

Most of Turing’s flashiest features can’t even be tested yet. And although GeForce RTX 2080 Ti and 2080 cards are starting to show up in the Tom’s Hardware lab, drivers remain closely guarded by Nvidia. Really, there’s no way to tell how these things perform across our benchmark suite. But we do know quite a bit about the underlying Turing architecture. We can also tell you about TU102, TU104, and TU106—the first three Turing GPUs—plus the Founders Edition products based on those processors.

It’s abundantly clear to everyone that Nvidia will emerge on the other side of this Turing launch with the fastest gaming graphics cards you can buy. What remains uncertain is whether the company’s eyebrow-raising prices, ranging from $600 to $1200, justify an upgrade now or encourage gamers to hold off until ray tracing gains momentum.

Grand Turing: Meet the TU102 GPU

The centerpiece of today’s graphics-focused smorgasbord is TU102, a 754-square-millimeter GPU that sits at the heart of Nvidia’s GeForce RTX 2080 Ti. Its 18.6 billion transistors are fabricated on TSMC’s 12nm FinFET manufacturing process, which purportedly reflects a slight density improvement over TSMC’s previous 16nm node. The foundry even classifies 12nm technology under the same umbrella as 16nm on its website. We’re not accustomed to covering Nvidia’s “big” gaming GPU at the same time as a new architecture. But Nvidia knows that for real-time ray tracing to entice enthusiasts, it needs to run at smooth frame rates. Getting TU102 into the hands of early adopters was critical this time around.

Compared to the biggest Pascal-based GPU used in a desktop graphics card, GP102, Nvidia’s TU102 is 60% larger with a 55%-higher transistor count. But it’s not the company’s most massive processor. The Turing-based flagship is eclipsed by GV100, a 21.1 billion-transistor behemoth measuring 815mm². That GPU was introduced in 2017 with an emphasis on data center applications, and is still found on the $3000 Titan V.

TU102 is aimed at a different target market than GV100, and it’s consequently provisioned with a list of resources to match. While elements of Turing do borrow from Nvidia’s work in Volta/GV100, pieces of the architecture that either don’t benefit gamers or aren’t cost-effective on the desktop are deliberately stripped out.

For example, each Volta Streaming Multiprocessor (SM) includes 32 FP64 cores for fast double-precision math, adding up to 2688 FP64 cores across GV100. They aren’t really useful in games though, and they eat up a lot of die space, so Nvidia pulled all but two of them from each Turing SM. As a result, TU102’s double-precision rate is 1/32 of its FP32 performance, leaving just enough FP64 compute to maintain compatibility with software dependent on it. Similarly, GV100’s eight 512-bit memory controllers attached to four stacks of HBM2 would have ended up being very expensive (just ask AMD about the trouble it had pricing HBM2-equipped Radeons competitively). They were consequently replaced with Micron-made GDDR6, facilitating a cheaper solution that’s still able to serve up a big bandwidth upgrade over Pascal-based predecessors.

A complete TU102 processor comprises six Graphics Processing Clusters (GPCs) made up of a Raster Engine and six Texture Processing Clusters (TPCs). Each TPC is composed of one PolyMorph Engine (fixed-function geometry pipeline) and two Streaming Multiprocessors (SMs). Again, at the SM level, we find 64 CUDA cores, eight Tensor cores, one RT core, four texture units, 16 load/store units, 256KB of register file space, four L0 instruction caches, and a 96KB configurable L1 cache/shared memory structure.

Multiply all of that out and you get a GPU with 72 SMs, 4608 CUDA cores, 576 Tensor cores, 72 RT cores, 288 texture units, and 36 PolyMorph engines.

Those resources are fed by 12 32-bit GDDR6 memory controllers, each attached to an eight-ROP cluster and 512KB of L2 cache yielding an aggregate 384-bit memory bus, 96 ROPs, and a 6MB L2 cache.

Putting It All Together: GeForce RTX 2080 Ti

The TU102 found on GeForce RTX 2080 Ti isn’t a complete processor, though. Whether Nvidia wanted to leave room for a Titan-class model or found yields of fully-functional GPUs unsatisfactory above a certain bin, the RTX 2080 Ti has two of its TPCs disabled, leaving the card with 4352 CUDA cores, 544 Tensor cores, 68 RT cores, 544 texture units, and 34 PolyMorph engines.

Moreover, one of TU102’s 32-bit memory controllers is turned off, creating an aggregate 352-bit bus that moves data to 88 ROPs and 5.5MB of L2 cache. Nvidia matches its strategically-hobbled GPU to Micron’s MT61K256M32JE-14:A modules. Eleven of these populate the RTX 2080 Ti’s PCB, leaving one emplacement vacant. Nevertheless, theoretical peak bandwidth rises sharply compared to the previous generation cards due to GDDR6’s higher data rate: at 14 Gb/s on a 352-bit interface, you’re looking at 616 GB/s. In comparison, GDDR5X at 11 Gb/s held GeForce GTX 1080 Ti to 484 GB/s.

At least on the Founders Edition card, a base core frequency of 1350 MHz jumps all the way up to a typical GPU Boost rate of 1635 MHz, so long as GeForce RTX 2080 Ti is running cool enough. And because Nvidia cites peak compute performance using GPU Boost numbers, its top-end model achieves up to 14.2 TFLOPS of single-precision math.

The reference specification calls for a GPU Boost frequency of 1545 MHz and a slightly lower TDP. Whereas the Founders Edition card’s overclock imposes a maximum board power of 260W, reference-class implementations should duck in around 250W.

Both configurations feature two NVLink interfaces for multi-GPU connectivity, though. This technology is covered in greater depth further along, but in short, each x8 link enables 50 GB/s of bi-directional bandwidth to support higher resolutions and faster refresh rates. On GeForce RTX 2080 Ti, 100 GB/s of total throughput is enough for 8K monitors in Surround mode.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Meet TU104 and GeForce RTX 2080

TU104: Turing With Middle Child Syndrome

It’s not that TU104 goes unloved, but again, we’re not used to introducing three GPUs alongside a new architecture. Then again, with GeForce RTX 2080 Ti starting at $1000, the RTX 2080, priced from $700, is going to find its way into more gaming PCs.

Similar to TU102, TSMC manufactures TU104 on its 12nm FinFET node. But a transistor count of 13.6 billion results in a smaller 545 mm² die. “Smaller,” of course, requires a bit of context. Turing Jr out-measures the last generation’s 471 mm² flagship (GP102) and comes close to the size of GK110 from the 2013-era GeForce GTX Titan.

TU104 is constructed with the same building blocks as TU102; it just features fewer of them. Streaming Multiprocessors still sport 64 CUDA cores, eight Tensor cores, one RT core, four texture units, 16 load/store units, 256KB of register space, and 96KB of L1 cache/shared memory. TPCs are still composed of two SMs and a PolyMorph geometry engine. Only here, there are four TPCs per GPC, and six GPCs spread across the processor. Therefore, a fully enabled TU104 wields 48 SMs, 3072 CUDA cores, 384 Tensor cores, 48 RT cores, 192 texture units, and 24 PolyMorph engines.

A correspondingly narrower back end feeds the compute resources through eight 32-bit GDDR6 memory controllers (256-bit aggregate) attached to 64 ROPs and 4MB of L2 cache.

TU104 also loses an eight-lane NVLink connection, limiting it to one x8 link and 50 GB/s of bi-directional throughput.

GeForce RTX 2080: TU104 Gets A (Tiny) Haircut

After seeing the GeForce RTX 2080 Ti serve up respectable performance in Battlefield V at 1920x1080 with ray tracing enabled, we can’t help but wonder if GeForce RTX 2080 is fast enough to maintain playable frame rates. Even a complete TU104 GPU is limited to 48 RT cores compared to TU102’s 68. But because Nvidia goes in and turns off one of TU104’s TPCs to create GeForce RTX 2080, another pair of RT cores is lost (along with 128 CUDA cores, eight texture units, 16 Tensor cores, and so on).

In the end, GeForce RTX 2080 struts onto the scene with 46 SMs hosting 2944 CUDA cores, 368 Tensor cores, 46 RT cores, 184 texture units, 64 ROPS, and 4MB of L2 cache. Eight gigabytes of 14 Gb/s GDDR6 on a 256-bit bus move up to 448 GB/s of data, adding more than 100 GB/s of memory bandwidth beyond what GeForce GTX 1080 could do.

Reference and Founders Edition RTX 2080s have a 1515 MHz base frequency. Nvidia’s own overclocked models ship with a GPU Boost rating of 1800 MHz, while the reference spec is 1710 MHz. Peak FP32 compute performance of 10.6 TFLOPS puts GeForce RTX 2080 Founders Edition behind GeForce GTX 1080 Ti (11.3 TFLOPS), but well ahead of GeForce GTX 1080 (8.9 TFLOPS). Of course, the faster Founders Edition model also uses more power. Its 225W TDP is 10W higher than the reference GeForce RTX 2080, and a full 45W above last generation’s GeForce GTX 1080.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Meet TU106 and GeForce RTX 2070

A Turing Baby Is Born

GeForce RTX 2070 is the third and final card Nvidia announced at its Gamescom event. Unlike GeForce RTX 2080 and 2080 Ti, the 2070 won’t be available until sometime in October. Gamers who wait can expect to find reference models starting around $500 and Nvidia’s own Founders Edition model selling for $100 more.

The 2070 is built around a complete TU106 GPU composed of three GPCs, each with six TPCs. Naturally, the TPCs include two SMs each, adding up to 36 SMs across the processor. Those blocks are unchanged between Turing GPUs, so RTX 2070 ends up with 2304 CUDA cores, 288 Tensor cores, 36 RT cores, and 144 texture units. TU106 maintains the same 256-bit memory bus as TU104, and it’s likewise populated with 8GB of 14 Gb/s GDDR6 modules capable of moving up to 448 GB/s. A 4MB L2 cache and 64 ROPs carry over as well. The only capability blatantly missing is NVLink, which isn't supported on RTX 2070.

Although TU106 is the least-complex Turing-based GPU at launch, its 445 mm² die contains no fewer than 10.8 billion transistors. That’s still pretty enormous for what Nvidia might have once considered the middle of its portfolio. In comparison, GP106—“mid-range Pascal”—was a 200 mm² chip with 4.4 billion transistors inside. GP104 measured 314 mm² and included 7.2 billion transistors. Targeting greater-than GTX 1080 performance levels, RTX 2070 really does seem like an effort to drive Tensor and RT cores as deep as possible down the chip stack, while keeping those features useful. It’ll be interesting to see how practical they remain in almost-halved quantities versus RTX 2080 Ti once optimized software becomes available.

Pumped-up die size aside, reference GeForce RTX 2070 cards based on TU106 have a 175W TDP. That’s less than GeForce GTX 1080.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Turing Improves Performance in Today’s Games

Some enthusiasts have expressed concern that Turing-based cards don’t boast dramatically higher CUDA core counts than their previous-gen equivalents. The older boards even have higher GPU Boost frequencies. Nvidia didn’t help matters by failing to address generational improvements in today’s games at its launch event in Cologne, Germany. But the company did put a lot of effort into rearchitecting Turing for better per-core performance.

To start, Turing borrows from the Volta playbook in its support for simultaneous execution of FP32 arithmetic instructions, which constitute most shader workloads, and INT32 operations (for addressing/fetching data, floating-point min/max, compare, etc.). When you hear about Turing cores achieving better performance than Pascal at a given clock rate, this capability largely explains why.

In generations prior, a single math data path meant that dissimilar instruction types couldn’t execute at the same time, causing the floating-point pipeline to sit idle whenever non-FP operations were needed in a shader program. Volta sought to change this by creating separate pipelines. Although Nvidia eliminated the second dispatch unit assigned to each warp scheduler, it also claimed that instruction issue throughput rose.

How is that possible? It's all about the composition of each architecture's SM.

Check out the two block diagrams below. Pascal has one warp scheduler per quad, with each quad containing 32 CUDA cores. A quad's scheduler can issue one instruction pair per clock through the two dispatch units with the stipulation that both instructions come from the same 32-thread warp, and only one can be a core math instruction. Still, that's one dispatch unit per 16 CUDA cores.

In contrast, Turing packs fewer CUDA cores into an SM, and then spreads more SMs across each GPU. There's now one scheduler per 16 CUDA cores (2x Pascal), along with one dispatch unit per 16 CUDA cores (same as Pascal). Gone is the instruction-pairing constraint. And because Turing doubles up on schedulers, it only needs to issue an instruction to the CUDA cores every other cycle to keep them full (with 32 threads per warp, 16 CUDA cores take two cycles to consume them all). In between, it's free to issue a different instruction to any other unit, including the new INT32 pipeline. The new instruction can also be from any warp.

Turing's flexibility comes from having twice as many schedulers as Pascal, so that each one has less math to feed per cycle, not from a more complicated design. The schedulers still issue one instruction per clock cycle. It's just that the architecture is better able to utilize resources thanks to its improved balance throughout the SM.

According to Nvidia, the potential gains are significant. In a game like Battlefield 1, for every 100 floating-point instructions, there are 50 non-FP instructions in the shader code. Other titles bias more heavily toward floating-point math. But the company claims there are an average of 36 integer pipeline instructions that would stall the floating-point pipeline for every 100 FP instructions. Those now get offloaded to the INT32 cores.

Despite the separation of FP32 and INT32 paths on its block diagrams, Nvidia says each Turing SM contains 64 CUDA cores to keep things straightforward. The Turing SM also comprises 16 load/store units, 16 special-function units, 256KB of register file space, 96KB of shared memory and L1 data cache, four texture units, eight Tensor cores, and one RT core.

On paper, an SM in the previous-generation GP102 appears more complex, sporting twice as many CUDA cores, load/store units, SFUs, texture units, just as much register file capacity, and more cache. But remember that the new TU102 boasts as many as 72 SMs across the GPU, while GP102 topped out at 30 SMs. The result is a Turing-based flagship with 21% more CUDA cores and texture units than GeForce GTX 1080 Ti, but also way more SRAM for registers, shared memory, and L1 cache (not to mention 6MB of L2 cache, doubling GP102’s 3MB).

That increase of on-die memory plays another critical role in improving performance, as does its hierarchical organization. Consider the three different data memories: texture cache for textures, L1 cache for load/store data, and shared memory for compute workloads. As far back as Kepler, each SM had 48KB of read-only texture cache, plus a 64KB shared memory/L1 cache. In Maxwell/Pascal, the L1 and texture caches were combined, leaving 96KB of shared memory on its own. Now, Turing combines all three into one shared and configurable 96KB pool.

The benefit of unification, of course, is that regardless of whether a workload is optimized for L1 or shared memory, on-chip storage is utilized rather than sitting idle as it may have before. Moving L1 functionality down has the additional benefit of putting it on a wider bus, doubling L1 cache bandwidth (at the TPC level, Pascal supports 64 bytes per clock cache hit bandwidth, while Turing can do 128 bytes per clock). And because those 96KB can be configured as 64KB L1 and 32KB shared memory (or vice versa), L1 capacity can be 50% higher on a per-SM basis.

Combined, Nvidia claims that the effect of its redesigned math pipelines and memory architecture is a 50% performance uplift per CUDA core. To keep those data-hungry cores fed more effectively, Nvidia paired TU102 with GDDR6 memory and further optimized its traffic reduction technologies (like delta color compression). Pitting GeForce GTX 1080 Ti’s 11 Gb/s GDDR5X modules against RTX 2080 Ti’s 14 Gb/s GDDR6 memory, both on an aggregate 352-bit bus, you’re looking at a 27%-higher data rate/peak bandwidth figure across the board. Then, depending on the game, when RTX 2080 Ti can avoid sending data over the bus, effective throughput increases even more by double-digit percentages.  

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Designing for The Future: Tensor Cores and DLSS

Although the Volta architecture was full of significant changes compared to Pascal, the addition of Tensor cores was most indicative of GV100’s ultimate purpose: to accelerate 4x4 matrix operations with FP16 inputs, which form the basis of neural network training and inferencing.

Like the Volta SM, Turing exposes two Tensor cores per quad, or eight per Streaming Multiprocessor. TU102 does feature fewer SMs than GV100 (72 versus 84), and GeForce RTX 2080 Ti has fewer SMs enabled than Titan V (68 versus 80). So, the RTX 2080 Ti only has 544 Tensor cores to Titan V’s 640. But TU102’s Tensor cores are implemented differently in that they also support INT8 and INT4 operations. This makes sense of course; GV100 was designed to train neural networks, while TU102 is a gaming chip able to use trained networks for inferencing.

Nvidia claims that TU102’s Tensor cores deliver up to 114 TFLOPS for FP16 operations, 228 TOPS of INT8, and 455 TOPS INT4. The FP16 multiply with FP32 accumulation operations used for deep learning training are supported as well, but at half-speed compared to FP16 accumulate.

Most of Nvidia’s current plans for the Tensor cores revolve around neural graphics. However, the company is also researching other applications of deep learning on desktop cards. Intelligent enemies, for instance, would completely change the way gamers approach boss fights. Speech synthesis, voice recognition, material/art enhancement, cheat detection, and character animation are all areas where AI is already in use, or where Nvidia sees potential.

But of course, Deep Learning Super Sampling (DLSS) is the focus for GeForce RTX. The process by which DLSS is implemented does require developer support through Nvidia’s NGX API. But the company claims integration is fairly easy, and has a list of games with planned support to demonstrate industry enthusiasm for what DLSS can do for image quality. This might be because the heavy lifting is handled by Nvidia itself. The company offers to generate ground truth images—the highest-quality representation possible, achieved through super-high resolution, lots of samples per frame, or lots of frames averaged together. Then, it’s able to train an AI model with the 660-node DGX-1-based SaturnV server to get lower-quality images as close to the ground truth as possible. These models are downloaded through Nvidia’s driver and accessed using the Tensor cores on any GeForce RTX graphics card. Nvidia claims that each AI model is measured in megabytes, making them relatively lightweight.

While we hoped Nvidia's GeForce Experience (GFE) software wouldn't be a requisite of DLSS, we suspected it probably would be. Sure enough, the company confirmed that the features of NGX are tightly woven into GFE. If the software detects a Turing-based GPU, it downloads a package called NGX Core, which determines if games/apps are relevant to NGX. When there's a match, NGX Core retrieves any associated deep neural networks for later use.

Will DLSS be worth the effort? It’s hard to say at this point. We’ve seen one example of DLSS from Epic’s Infiltrator demo and it looked great. But it’s unclear if Nvidia can get the same caliber of results from any game, regardless of genre, pace, environmental detail, and so on. What we do know is that DLSS is a real-time convolutional auto-encoder trained on images sampled 64 times. It’s given a normal-resolution frame through the NGX API, and spits back a higher-quality version of that frame.

Shortly after its Gamescom announcement, Nvidia started teasing performance figures of GeForce RTX 2080 with DLSS enabled versus GeForce RTX 2080 versus GTX 1080. Those results made it seem like turning DLSS on improved frame rates absolutely, but there just wasn't much backup data to clarify how the benchmark numbers were achieved. As it turns out, DLSS improves performance by reducing the card's shading workload, while achieving similar quality.

Turing can produce higher-quality output from a given number of input samples compared to a post-processing algorithm like Temporal Anti-Aliasing (TAA). For DLSS, Nvidia turns that into a performance benefit by reducing input samples to the network until the final output (at the same resolution as TAA) is close to a similar quality. Even though Turing spends time running the deep neural network, the savings attributable to less shading work are greater. Set to 2x DLSS, Nvidia says it can achieve the equivalent of 64x super-sampling by rendering its inputs at the target resolution, while side-stepping the transparency artifacts and blurriness sometimes seen from TAA.

Twenty-five games are already queued up with DLSS support, including existing titles like Ark: Survival Evolved, Final Fantasy XV, and PlayerUnknown’s Battlegrounds, plus several others that aren’t out yet.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Hybrid Ray Tracing in Real-Time

Beyond the scope of anything that Volta touched, and arguably the most promising chapter in Turing’s story, is the RT core bolted to the bottom of each SM in TU102. Nvidia’s RT cores are essentially fixed-function accelerators for Bounding Volume Hierarchy (BVH) traversal and triangle intersection evaluation. Both operations are essential to the ray tracing algorithm. For more background about ray tracing and why it’s so visually appealing, see our feature: What is Ray Tracing and Why Do You Want it in Your GPU?

In short, BVHs form boxes of geometry in a given scene. These boxes help narrow down the location of triangles intersecting rays through a tree structure. Each time a triangle is found to be in a box, that box is subdivided into more boxes until the final box can be divided into triangles. Without BVHs, an algorithm would be forced to search through the entire scene, burning tons of cycles testing every triangle for an intersection.

Running this algorithm today is entirely possible using the Microsoft D3D12 Raytracing Fallback Layer APIs, which use compute shaders to emulate DirectX Raytracing on devices without native support (and redirect to DXR when driver support is identified). On a Pascal-based GPU, for instance, the BVH scan happens on programmable cores, which fetch each box, decode it, test for an intersection, and determine if there’s a sub-box or triangles inside. The process iterates until triangles are found, at which point they’re tested for intersection with the ray. As you might imagine, this operation is very expensive to run in software, preventing real-time ray tracing from running smoothly on today’s graphics processors.

By creating fixed-function accelerators for the box and triangle intersection steps, the SM casts a ray into the scene using a ray generation shader and hands off acceleration structure traversal to the fixed-function RT core. All of the intersection evaluation happens much more quickly as a result, and the SM’s other resources are freed up for shading just as they would for a traditional rasterization workload.

According to Nvidia, a GeForce GTX 1080 Ti can cast about 1.1 billion rays per second in software using its CUDA cores capable of 11.3 FP32 TFLOPs. In comparison, GeForce RTX 2080 Ti can cast about 10 billion rays per second using its 68 RT cores. It’s important to note that neither of those figures are based on calculated peaks like a lot of speeds and feeds. Rather, Nvidia took the geometric mean of results from several workloads to settle on its “10+ gigarays”value.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

NVLink: A Bridge To…Anywhere?

TU102 and TU104 are Nvidia’s first desktop GPUs rocking the NVLink interconnect rather than a Multiple Input/Output (MIO) interface for SLI support. The former makes two x8 links available, while the latter is limited to one. Each link facilitates up to 50 GB/s of bidirectional bandwidth. So, GeForce RTX 2080 Ti is capable of up to 100 GB/s between cards and RTX 2080 can do half of that.

One link or two, though, SLI over NVLink only works across a pair of GeForce RTX boards with at least one empty slot between them for airflow. Officially, Pascal-era GPUs endured the same two-card maximum. Technically, however, as many as four top-end GeForce GTXes could be made to work together in a handful of benchmarks. These days, you’ll also have to purchase your own GeForce RTX NVLink Bridge for multi-GPU connectivity. Three- and four-slot sizes are both available for $80 from Nvidia’s website.    

Some of the trouble last generation was caused by bandwidth constraints between SLI bridges. Compared to the original SLI interface’s 1 GB/s MIO link, Pascal’s implementation drove ~4 GB/s. That was fast enough to get the second card’s rendered frame back to the primary board in time for smooth output to a 4K monitor at 60 Hz. But it wouldn’t have been able to keep up at 120 Hz and higher, which is where today’s highest-end gaming displays operate.

Even in a single-link configuration, NVLink can move data so quickly that SLI on an 8K screen is possible. Driving three 4K monitors at 144 Hz in Surround mode is no problem at all. Two x8 links have the throughput needed for 8K displays in Surround.

Really, the question is: who cares anymore? AMD and Nvidia did such a good job of pumping the brakes on multi-GPU configurations that Tom’s Hardware readers rarely, if ever, ask for benchmark results from an SLI setup. Back in the day, value-minded gamers used SLI to match the performance of higher-end cards. Nvidia put a stop to that by removing support from lower-end models in its product stack. Now, even GeForce RTX 2070 lacks an NVLink connector. Older DirectX 11-based games still run well across two cards, and a handful of DirectX 12-based titles do exploit the API’s explicit multi-adapter control. But the fact that developers like EA DICE are pouring time into taxing features like real-time ray tracing and ignoring multi-GPU says a lot about SLI’s current importance.

We’ve heard Nvidia representatives say they’ll have more to discuss on this front in the future. For now, NVLink support on GeForce RTX 2080 Ti and 2080 is a novelty, particularly as we’re able to focus on playable frame rates at 4K and G-Sync technology to keep the action smooth.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Mesh Shading: A Foundation for More On-Screen Objects

Architectural enhancements are commonly split between changes that affect gaming today and new features that require support from future titles. At least at launch, Nvidia’s GeForce RTX cards are going to be judged predominantly on the former. Today’s games are what we can quantify. And although veterans in the hardware field have their own opinions of what real-time ray tracing means to an immersive gaming experience, I’ve been around long enough to know that you cannot recommend hardware based only on promises of what’s to come.

The Turing architecture does, however, include a few more advanced graphics features loaded with potential but not yet accessible. Mesh shaders, for instance, augment the existing DirectX 11/12 graphics pipeline whereby host processors are responsible for calculating levels of detail, culling objects that aren’t in view, and issuing draw calls for each object. Up to a certain point, CPUs are fine for this. But in complex scenes with hundreds of thousands of objects, they have a hard time keeping up.

By using a mesh shader instead, game developers can offload LOD calculation and object culling to a task shader, which replaces the vertex shader and hull shader stages. Because the task shader is more general than a vertex shader, it’s able to take a list of objects from the CPU and run it through a compute program that determines where objects are located, what version of each object to use based on LOD, and if an object needs to be culled before passing it down the pipeline.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Variable Rate Shading: Get Smarter About Shading, Too

In addition to optimizing the way Turing processes geometry, Nvidia also supports a mechanism for choosing the rate at which 16x16 blocks of pixels are shaded in different parts of a scene to improve performance. Naturally, the hardware can still shade every single pixel in a 1x1 pattern. But the architecture also facilitates 2x1 and 1x2 options, along with 2x2 and 4x4 blocks.

Nvidia offers several use cases where variable rate shading is practical (you don’t want to apply it gratuitously and negatively affect image quality). The first is content-adaptive shading, where less detailed parts of a scene don’t change as much and can be shaded at a lower rate. There’s actually a build of Wolfenstein II with variable rate shading active. In it, you can turn on the shading rate visualization to watch how complex objects aren’t affected at all by this technology, while lower-frequency areas get shaded at a lower resolution. A number of intermediate steps facilitate multiple rates. We must imagine that game developers looking to exploit variable rate shading in a content-adaptive manner will prioritize quality over performance. Still, we’d like to see this enabled as a toggleable option so third parties can draw comparisons with the feature on and off. 

Motion-adaptive shading is another interesting application of Nvidia’s variable rate shading technology, where objects flying by are perceived at a lower resolution than whatever subject we’re focused on. Based on the motion vector of each pixel, game developers can determine how aggressively to reduce the shading rate and apply the same patterns seen in the content-adaptive example. Doing this correctly does require an accurate frequency response model to ensure the right rates are used when you spin around, sprint forward, or slow back down.

Again, Nvidia presented a demo of Wolfenstein II with content- and motion-adaptive shading enabled. The performance uplift attributed to variable rate shading in that title was on the order of ~15%, if only because Wolfenstein II already runs at such high frame rates. But on a slower card in a more demanding game, it may become possible to get 20%+-higher performance at the 60ish FPS level. Perhaps more important, there was no perceivable image quality loss.

Although Nvidia hasn’t said much about how these capabilities are going to be utilized by developers, we do know that its Wolfenstein II demo was made possible through Vulkan extensions. The company is working with Microsoft to enable DirectX support for Variable Rate Shading. Until then, it'll expose Adaptive Shading functionality through the NVAPI software development kit, which allows direct access to GPU features beyond the scope of DirectX and OpenGL.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

RTX-OPS: Trying to Make Sense of Performance

As modern graphics processors become more complex, integrating resources that perform dissimilar functions but still affect the overall performance picture, it becomes increasingly difficult to summarize their capabilities. We already use terms like fillrate to compare how many billions of pixels or texture elements a GPU can theoretically render to screen in a second. Memory bandwidth, processing power, primitive rates—the graphics world is full of peaks that become the basis for back-of-the-envelope calculations.

Well, with the addition of Tensor and RT cores to its Turing Streaming Multiprocessors, Nvidia found it necessary to devise a new metric that’d suitably encompass the capabilities of its INT32 and FP32 math pipelines, its RT cores, and the Tensor cores. Tom’s Hardware doesn’t plan to use the resulting “RTX-OPS” specification for any of its comparisons, but since Nvidia is citing it, we want to at least describe the equation’s composition.

The RTX-OPS model requires utilization of all resources, which is a bold and very future-looking assumption. After all, until games broadly adopt Turing’s ray tracing and deep learning capabilities, RT and Tensor cores sit idle. As they start coming online, though, Nvidia developed its own approximation of the processing involved in one frame rendered by a Turing-based GPU.

In the diagram above, Nvidia shows roughly 80% of the frame consumed by rendering and 20% going into AI. In the slice dedicated to shading, there’s a roughly 50/50 split between ray tracing and FP32 work. Drilling down even deeper into the CUDA cores, we already mentioned that Nvidia observed roughly 36 INT32 operations for every 100 FP32 instructions across a swathe of shader traces, yielding a reasonable idea of what happens in an “ideal” scene leveraging every functional unit.

So, given that…

FP32 compute = 4352 FP32 cores * 1635 MHz clock rate (GPU Boost rating) * 2 = 14.2 TFLOPS

RT core compute = 10 TFLOPS per gigaray, assuming GeForce GTX 1080 Ti (11.3 TFLOPS FP32 at 1582 MHz) can cast 1.1 billion rays using software emulation = ~100 TFLOPS on a GeForce RTX 2080 Ti capable of casting ~10 billion rays

INT32 instructions per second = 4352 INT32 cores * 1635 MHz clock rate (GPU Boost rating) * 2 = 14.2 TIPS

Tensor core compute = 544 Tensor cores * 1635 MHz clock rate (GPU Boost rating) * 64 floating-point FMA operations per clock * 2 = 113.8 FP16 Tensor TFLOPS 

…we can walk Nvidia’s math backwards to see how it reached a 78 RTX-OPS specification for its GeForce RTX 2080 Ti Founders Edition card:

(14 TFLOPS [FP32] * 80%) + (14 TIPS [INT32] * 28% [~35 INT32 ops for every 100 FP32 ops, which take up 80% of the workload]) + (100 TFLOPS [ray tracing] * 40% [half of 80%]) + (114 TFLOPS [FP16 Tensor] * 20%) = 77.9

Again, there are a lot of assumptions made in this model, we see no way to use it for generational or competitive comparisons, and we don’t want to get in the habit of generalizing ratings across many different resources. At the same time, it’s clear that Nvidia wanted a way to represent performance holistically and we cannot fault the company for trying, particularly since it didn’t just add the capabilities of each subsystem but rather isolated their individual contributions to a frame.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Display Outputs and the Video Controller

Display Outputs: A Controller For The Future

The GeForce RTX Founders Edition cards sport three DisplayPort 1.4a connectors, one HDMI 2.0b output, and one VirtualLink interface. They support up to four monitors simultaneously, and naturally are HDCP 2.2-compatible. Why no HDMI 2.1? That standard was released in November of 2017, long after Turing was finalized.

Turing also enables Display Stream Compression (DSC) over DisplayPort, making it possible to drive an 8K (7680x4320) display over a single stream at 60 Hz. GP102 lacked this functionality. DSC is also the key to running at 4K (3840x2160) with a 120 Hz refresh and HDR.

Speaking of HDR, Turing now natively reads in HDR content through dedicated tone mapping hardware. Pascal, on the other hand, needed to apply processing that added latency.

Finally, all three Founders Edition cards include VirtualLink connectors for next-generation VR HMDs. The VirtualLink interface utilizes a USB Type-C connector but is based on an Alternate Mode with reconfigured pins to deliver four DisplayPort lanes, a bi-directional USB 3.1 Gen2 data channel for high-res sensors, and up to 27W of power. According to the VirtualLink Consortium, existing headsets typically operate within a 15W envelope, including displays, controllers, audio, power loss over a 5m cable, cable electronics, and connector losses. But this new interface is designed to support higher-power devices with improved display capabilities, better audio, higher-end cameras, and accessory ports as well. Just be aware that VirtualLink’s power delivery is not reflected in the TDP specification of GeForce RTX cards; Nvidia says using the interface requires up to an additional 35W.

Video Acceleration: Encode And Decode Improvements

Hardware-accelerated video features don’t get as much as attention as gaming, but new graphics architectures typically do add improvements that support the latest compression standards, incorporate more advanced coding tools/profiles, and offload work from the CPU.

Encode performance is more important than ever for gamers streaming content to social platforms. A GPU able to handle the workload in hardware alleviates other platform resources, so the encode has a smaller impact on game frame rates. Historically, GPU-accelerated encoding didn’t quite match the quality of a software encode at a given bit rate, though. Nvidia claims Turing changes this by absorbing the workload and exceeding the quality of a software-based x264 Fast encode (according to its own peak signal-to-noise ratio benchmark). Beyond a certain point, quality improvements do offer diminishing returns. It’s notable, then, that Nvidia claims to at least match the Fast profile. But streamers are most interested in the GPU’s ability to minimize CPU utilization and bit rate.

This generation, the NVEnc encoder is fast enough for real-time 8K HDR at 30 FPS. Optimizations to the encoder facilitate bit rate savings of up to 25% in HEVC (or a corresponding quality increase at the same bit rate) and up to 15% in H.264. Hardware acceleration makes real-time encoding at 4K a viable option as well, although Nvidia doesn’t specify what CPU it tested against to generate a 73% utilization figure in its software-only comparison.

Additionally, the decode block supports VP9 10/12-bit HDR. It’s not clear how that differs from GP102-based cards though, since GeForce GTX 1080 Ti is clearly listed in Nvidia’s NVDec support matrix as having VP9 10/12-bit support as well. Similarly, HEVC 4:4:4 10/12-bit HDR is listed as a new feature for Turing.  

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Nvidia’s Founders Edition: Farewell, Beautiful Blower

The GeForce RTX 2080 Ti, 2080, and 2070 Founders Edition cards represent a departure from the “reference” design we were first introduced to in 2012 with GeForce GTX 690. After the 690, we grew to appreciate cards like the original GeForce GTX Titan with its centrifugal fan that blew heated air out the dual-slot bracket and away from your PC’s guts. Many enthusiasts felt differently, though. Because centrifugal fans move air through heat sinks quickly, they tend to be noisier under load than more free-flowing thermal solutions with multiple axial fans. And because Nvidia tried to keep their reference boards running quietly, they were often accused of limiting peak GPU Boost clock rates or even outright throttling performance.

Sadly for us, those days are gone. The new Founders Edition design eschews a centrifugal fan for two axial fans. They sit atop a dense fin stack that surrounds what Nvidia calls the largest vapor chamber ever used on a graphics card. A forged aluminum cover encircles the cooler’s length but leaves the top and bottom open. Heated air is consequently directed up out the top and down towards your motherboard, possibly in the direction of an M.2-based SSD installed underneath. The aesthetic just isn’t as distinct, sacrificing the prestige of Nvidia’s previous-gen reference boards.  

Gamers who put a lot of care into building PCs with plenty of airflow should enjoy a better experience overall though; the axial fans can dissipate more power or drive lower temperatures at a given noise level. Alternatively, the axial fans offer improved acoustics compared to blowers in a power-limited condition. Either way, there’s a silver lining for anyone who preferred the elegance of a centrifugal fan exhausting heated air but is willing to entertain the merits of Nvidia’s latest creation.

Another potential selling point is improved overclocking headroom. Some of this comes from the cooler’s increased capacity. But Nvidia also says its power supply facilitates close to 60W of additional capacity beyond the stock 260W. At the same time, efficiency is optimized using an eight-phase power supply able to dynamically turn phases on and off based on load.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Overclocking: Making The Most Of Headroom With Nvidia Scanner

The Founders Edition cards are already overclocked beyond Nvidia’s base specification, but they aren’t tuned right up to their breaking points. In order for the company to define a frequency floor and typical GPU Boost rate, all samples must be capable of the same clocks at minimum. From there, though, every GeForce RTX hits a different limit before becoming unstable. That ceiling even changes depending on workload. Enthusiasts often make it a point of pride to push right up to this threshold by finessing whatever gears, levers, and dials are exposed through popular apps like Precision XOC and Afterburner. 

To the best of its ability, Nvidia is taking the trial and error out of overclocking with an API/DLL package that partners like EVGA and MSI can build into their utilities. Instead of an enthusiast going back and forth, testing one part of the frequency/voltage curve at a time and adjusting based on the stability of various workloads, Nvidia’s Scanner runs an arithmetic-based routine in its own process, evaluating stability without user input. Although Nvidia says the metric usually encounters math errors before crashing, the fact that it’s contained means the algorithm can recover gracefully if a crash does occur. This gives the tuning software a chance to increase voltage and try the same frequency again. Once the Scanner hits its maximum voltage setting and encounters one last failure, a new frequency/voltage curve is calculated based on the known-good results. From start to finish, the process purportedly takes fewer than 20 minutes.

Interestingly, Nvidia Scanner functionality won’t be limited to owners of GeForce RTX graphics cards. The company says it’ll characterize older boards as well (though it isn’t specific about how far back support will stretch).

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Ray Tracing And AI: Betting It All on Black

Nvidia could have poured all of its efforts into features that make Turing faster than the Pascal architecture in today’s games, rather than splitting attention between immediate speed-ups and investments in the future. But the company saw an opportunity to drive innovation and took a calculated risk. And why not? Its performance and efficiency lead over AMD’s Vega RX series creates a comfortable margin to push next-gen technology more than we’ve seen in years.

Dedicating gobs of transistors to Tensor cores and RT cores doesn’t pay off today. It doesn’t pay off tomorrow. Rather, it pays off months or even years down the road as developers become comfortable optimizing their games for a combination of real-time ray tracing and rasterization. Nvidia needs them to spend time experimenting with neural graphics and AI. Technologies like mesh shading and variable rate shading aren’t used anywhere currently; they require adoption and support. With all of that said, developer enthusiasm for real-time ray tracing appears unanimous. This truly seems like a vision everyone wants to see realized.  

In the meantime, gamers aren’t going to spend their hard-earned dollars on a future promise of better-looking visuals. They want instant gratification in the form of higher frame rates. We can talk architecture ad nauseam. However, the features found in today’s discussion only amount to half of the story (the academic half, no less). In a few days, we’ll have a battery of performance, power, and acoustic results to present. Only then can we tell you if GeForce RTX 2080 and 2080 Ti warrant the high prices that have enthusiasts gnashing their teeth.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

LAS VEGAS, NV -- In 2014, HP introduced the Sprout, an all-in-one PC with a built-in 3D camera for digital creators. Now it has a standalone version of the Sprout’s 3D scanner called the HP Z 3D Camera that brings the technology down to a more attainable price and form factor that’s easier to share.

“The demand for 3D content has skyrocketed, and 3D pros are constantly looking for new hardware and software innovations to increase their end-to-end productivity, from conception to production,” said Louis Kim, global head of Immersive Computing, Personal Systems, HP Inc. “HP Z 3D Camera helps speed up modeling, reinvent the creative palette and protect the artistic point of view as more people enter the profession and workflows become increasingly automated.”

HP said that it built the HP Z 3D Camera for creators such as game developers and 3D artists who need an easy way to digitize content but can’t afford, or don’t want to afford, the $3,700 Sprout Pro desktop. The HP Z 3D Camera is easier to share with a group of people, too. HP created a magnetic mount that allows you to easily move it from PC to PC, or take it home with you after work. The HP Z 3D Camera bundle includes three adhesive bases that you can stick to the back of different displays so you can attach the camera when you need it.

The HP Z 3D Camera features a 14MP down-facing camera that can capture 2D images, which enables you to scan documents or images. The camera can also capture 2D video, so you can use to record yourself writing or drawing. The camera also features SLAMscan technology to provide 3D scanning capability, which allows you to create 3D models of real objects in real time.

The HP Z 3D Camera pairs with the HP WorkTools applications suite, which enables several advanced features. The suite offers a solution to automatically remove your hands from a 3D model after you hold an object in front of the camera for capture; the SLAMscan Technology can determine what part of the scan is part of your body and what is the object you’re scanning.

The HP WorkTools software outputs .OBJ files from the scans that the HP Z 3D camera captures. These files can be imported into popular 3D modeling software for further refinement. You can also drag and drop the scans into the PowerPoint 3D and Paint 3D applications on Windows 10.

The WorkTools software also automatically adjusts the white balance of the camera, and it features automated keystone software, so you never have to calibrate the device.

HP said the HP Z 3D Camera should work with most modern PCs that have a proper GPU. The minimum requirements include a GTX 960M mobile GPU or better and two USB Type-A port (one for power, one for data). HP also recommends that you use a “good monitor” when working with the HP Z 3D Camera.

The HP Z 3D Camera isn’t available yet, but it’s not far off. HP said the camera would be available in March and that it would retail for $599.

Update, 7/28/17, 8:05am PT: Microsoft announced that the Modern Keyboard is now available in the U.S, Canada, and China for $140 USD. The company also released a new Modern Mouse designed to complement the Modern Keyboard's looks; it's available in the same regions for $50. 

Original article: 6/16/17, 10:10am PT:

Passwords aren't the perfect authentication method. People often share them, forget them, or use simple passwords that are easy to guess. That's why many companies are pushing for biometric authentication by adding iris scanners, facial recognition features, and fingerprint sensors to their products. Microsoft joined their ranks with the new Modern Keyboard with Fingerprint ID, which features a "hidden" fingerprint scanner.

That scanner is housed inside a key with a fingerprint icon. Microsoft published a video demonstrating its use in which a woman named Miranda signs on to her Windows 10 device by pressing the key with her index finger. Windows Hello--an authentication tool that supports facial recognition and device pairing as well as fingerprint scanning--handles everything on the software side. The whole process is supposed to be quick and easy.

Microsoft said on Windows Hello's page that the feature can sign you into a Windows 10 device in less than two seconds. That's 3x faster than if you had to enter a password, the company said, but the real draw is the added security. The page opens with "Windows Hello: They can guess your password--not your face." (We're going to assume the company also stands by the security of Windows Hello's fingerprint support.)

That is marginally true--it's often harder for someone to mimic your body than it would be for them to guess your password. But like other companies offering biometric authentication, Microsoft has effectively glossed over the real security and privacy problems associated with them.

The first problem: Your body can be spoofed. Authentication tools have been tricked by high resolution images of eyes and faces before, and someone can replicate your fingerprint, too. The rise of social media has made it easy for someone to get those high resolution images; your own selfies could be used against you. You probably won't unintentionally share a password the same way you share biometric data.

The second problem: Your body can be used against you. Several U.S. courts have ruled that law enforcement officials can force you to sign in to a device if you use biometric authentication. (And that's ignoring other, more painful and gruesome ways your body parts can be used to log in against your will.) That isn't true of passwords. And even if it were, it's easier to protect an idea than it is to stop someone from forcing your finger onto a scanner.

This isn't a black and white issue. Biometric authentication makes sense for people who don't have to worry about law enforcement forcing them to unlock their devices. Passwords make sense for people who do worry about those things, or who simply want to make sure they can still post pictures of themselves to social media without inadvertently compromising their security. Everyone has to account for their own personal needs.

Windows Hello is going to be around either way. The Modern Keyboard doesn't change that--it merely gives people another way to use the feature. Before, you had to purchase a webcam or fingerprint sensor to use the feature (unless you have a laptop, many of which have those features built-in). Not that those are hard to come by; Microsoft said in January that "nearly 100" biometric devices work with Windows Hello.

Those products were more obtrusive than the Modern Keyboard, though, because basically everyone needs a keyboard. Windows Hello makes biometric authentication a core part of Windows 10; the Modern Keyboard seems like a (friendly?) Trojan horse meant to make fingerprint scanners a common aspect of future keyboards. Microsoft is proving that it can be done while simultaneously appealing to people who like its peripherals.

The Modern Keyboard features a number pad and, aside from the key housing the fingerprint sensor, a more or less standard layout. The F buttons pull double duty as media controls and shortcuts to other functions, such as search, home, or switching between desktops. The keyboard can be used wired or wirelessly via Bluetooth Low Energy 4.0 or 4.1 with a maximum range of 50 feet (in "open air") or 23 feet (in an "office environment.") It measures 16.57 x 4.43 x 0.76" (LxWxH) and weighs 14.79 oz. with batteries installed.

Microsoft said the Modern Keyboard can last up to two months on a pair of AAA batteries, which come with the device. The keyboard is compatible with Windows 10 / 8.1 / 8, Windows 10 Phone, macOS 10.10.5 or later, Android 4.4.2-5, and iOS 8.1-9.2.1. Those last two are kind of surprising--Google plans to release Android 8.0 later this year, and Apple's set to release iOS 11 this Fall. It's not clear if the Modern Keyboard will experience some problems with those versions of the mobile operating systems or if Microsoft simply hasn't optimized it for them.

Microsoft's Modern Keyboard with Fingerprint ID costs $130 from the company's online store. It's listed as "coming soon"--a release date wasn't provided.

You might know Introversion Software for its most recent game, Prison Architect. The small studio has continued to work on game prototypes since that title's release, and it's finally coming out with a new game on April 26. Instead of building and managing an entire prison complex, you’ll explore a vast and mysterious cave with the help of a scanner in Scanner Sombre.

The objective of Scanner Sombre is simple: You have to make your way through a complex cave network in an attempt to return to the surface. However, most of the cave paths are in complete darkness, and you can only navigate them with the help of a LIDAR (light detection and ranging) scanner. This scanner will map out your surroundings with a plethora of light points, which translates into a 3D map of the cave. Because of the many pathways available throughout the cave network, you’ll probably find some dead ends or paths that loop around to the same sections. You might also encounter some strange figures, which suggests that you might not be alone in the subterranean caverns.

The game was revealed last year as part of an update video for Prison Architect. It was one of two prototype games created by the studio, and even though it was in development for last year, a release date wasn’t announced until today.

When it comes out on Wednesday, you can purchase it through Steam, GOG, and the Humble Store for $12, but it will be available for a limited time with a 20% discount, which brings it down to roughly $10.

At Mobile World Congress (MWC), which begins next week, biometrics is a big deal. Synaptics just announced its latest fingerprint scanner, and Vkansee has a new one to show off at the show, as well.

We encountered Vkansee at MWC last year and witnessed an impressive demo of its unique multi-layered UTFIS optical sensor that can read a print even if the finger is wet or dirty. The company is back at it this year, bringing out a new iteration of the sensor.

It appears that this new sensor employs all the same technology as the one we saw a year ago, but Vkansee said that it’s smaller, with an active area of 6 x 3.5 mm. Like its predecessor, the (pre-production) VK2108 sensor is just 1.5 mm thick, and it captures “third-level” data such as sweat pores and fingerprint ridges and offers the same 2000 PPI resolution. It can be bonded under 0.4-0.8 mm cover glass.

Typically, both component makers and device OEMs play coy about who they’re working with. Thus, we aren’t sure where or if Vkansee has landed any deals with manufacturers. In a press release, Jason Chaikin, President of Vkansee, stated, “When we first set out to disrupt the biometric authentication space a year ago, we didn’t expect to receive interest from the major players so quickly. With the sensors now in theirhands, we’re looking forward to finalizing our design and seeing them hit the market soon.”

Reading between the lines there, it would seem that Vkansee is still working towards getting its sensor onto shipping devices.  A year ago, Vkansee told us the same thing -- that several OEMs were evaluating its sensor. In any case, Vkansee now has an even smaller version of its sensor to show off, and the pre-production version of it is in the hands of “leading PC and smartphone manufacturers,” according to the company.

Seth Colaner is the News Director for Tom's Hardware. Follow him on Twitter @SethColaner. Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Next Biometrics Group revealed the latest iteration of the company's fingerprint sensors. The new sensors are thinner, and feature a frameless and bezel free design, which NEXT said makes them perfect to implement into smartphones, tablets and portable computers. With its latest product, Next is expecting to expand its market share in these key markets.

Next Biometrics Group, based in Oslo, Norway, is one of the world's leading fingerprint sensor manufacturers, building sensors for a wide range of products. Next's sensors are built using the company's patented Active Thermal Principle, which it credits to keeping the cost of its sensors low while maintaining high quality. Next stated that the cost of the new sensors is further reduced by removing the bezel material from the package.

The tiny footprint of the sensor makes it possible for product designers to introduce a fingerprint sensor without compromising the aesthetics of the device. Next said that the new sensors will sit nearly flush with the surface onto which it's embedded. The sensors are designed to be sunken only 0.4 to 0.6 mm. The company said that the recessed position aids in locating the sensor by touch.

The company stated that it wants to meet the requirements of product designers by allowing them to define the size and shape of the opening for the sensor that best suits the design of the product in which it will be used.

Follow Kevin Carbotte @pumcypuhoy. Follow us @tomshardware, on Facebook and on Google+.

Among several new releases from EVGA, which included graphics cards with Nvidia's GTX 980 Ti GPU, the Z10 keyboard with the built-in EVBot, the new TORQ X5 mouse featuring a fingerprint scanner and a gaming case, we also spotted two new EVGA PSUs.

EVGA's SuperNOVA line will be enriched with a Titanium unit with 1 kW capacity and an 850 W PSU with Platinum efficiency. Both of the aforementioned PSUs are made by Super Flower and are based on the company's Leadex platform, which is among the best platforms money can buy. EVGA utilizes a fully modular cabling design and features a semi-passive mode for total silent operation at light loads.

According to EVGA, these two new PSUs will be available in either August or September 2015. Pricing information has not yet been determined.

Follow us @tomshardware, on Facebook and on Google+.

Optical fingerprint scanners are regularly used in high-traffic areas such as border crossings or fitness clubs to verify membership. The rugged but bulky construction of optical scanners are well suited to these applications, but not for razer-thin mobile devices, where capacitive sensors are currently used due to their low cost and compact size.

The exclusivity of capacitive sensors will soon end, however, because other methods are making the move to mobile, too. Qualcomm recently announced its ultrasonic fingerprint scanner, Sense ID, for mobile applications, and now Vkansee is doing the same with a new type of optical sensor.

Vkansee's new multi-layered UTFIS sensor replaces the space-consuming prism and lens system of traditional optical scanners with a transparent sheet and a metal film filled with pinholes, respectively. These sit atop a CMOS imaging device, creating a sensor that's only 1.5 mm thick and can be bonded below the cover glass of mobile devices. This solution, together with the company's proprietary software, produces an image with 2000 PPI resolution, four times the pixel density found in standard fingerprint sensors.

With such a high-resolution scan, it can detect third-level features like sweat pores, making it more difficult to fool than capacitive sensors. At MWC 2015, Vkansee demonstrated how a rubber mold of a fingerprint was sufficient to gain access to an iPad Air 2 using Touch ID; however, the same hack failed to bypass the UTFIS sensor. Also, you no longer have to worry about some rogue cutting off your finger and using it to unlock your stolen phone, because your sweat pores would close and foil a match, keeping your sensitive data safe (you would still be missing a finger and your phone, though).

Optical sensors are less sensitive to variations in temperature and moisture than capacitive sensors because they are non-conductive, which should reduce the number of false rejections. In the video below, Jason Chaikin, Vkansee's President, demonstrates how the UTFIS sensor still works even if his finger is wet. Optical sensors, because they image the surface of the skin, encounter problems when the finger is dirty, however.

Vkansee provides OEMs with several different color options for its sensor. The imaging surface, which can be seen below the cover glass, comes in black, white, silver or gold. The sensor is also encircled by an LED light, which provides the near-infrared light necessary for imaging and also serves as a vanity light, indicating where to place the finger or flashing to alert the user to a message or notification. The LED lights are available in white, green, blue or red, and there's a clear option, as well.

According to Vkansee, several OEMs are currently evaluating its new sensor, and we could potentially see it integrated into products by the end of this year or early next.

Matt Humrick is the Mobile Editor at Tom's Hardware. Contact him at mhumrick@tomshardware.com and follow him on Twitter @digitalOut_net. Follow us @tomshardware, on Facebook and on Google+.

Greetings From Barcelona

Sometimes we must make sacrifices for our readers. That may mean playing with heavily testing hot new tech weeks before it's released to the general public. Sometimes we have to game for hours to fully evaluate some hardware. In other cases, we must travel to Barcelona, one of the most beautiful cities in the world, for Mobile World Congress (MWC), which we did recently. You're welcome, folks. 

MWC is, of course, rife with new handsets, but there's so much more to see there. There's new mobile component technology, new standards being implemented, emerging trends and so on. That's not to mention plenty of startups pushing new ideas, concepts and technologies.

This MWC was a study in contrasts. Some of the major developments, such as OEM strategies in emerging markets, were important but not exactly sexy, while others such as HTC's and Valve's shocking Vive VR announcement left even the most jaded techies in a downright tizzy. 

What follows is what we believe to be some of the most noteworthy products and technologies from MWC.

Best To Watch: Yezz Project Ara Module Development

Although some would debate the need for or excitement over Google's Project Ara modular smartphone, the prospect of a true enthusiast-type smartphone — that is, one with discrete parts that users can upgrade piece by piece, just as they do a PC — is tantalizing for many.

So much of Project Ara is promises and plastic prototypes, but there is a pilot launch happening this year (in Puerto Rico), and at least one company, Yezz, is actually producing modules for Project Ara. The company stated that it will have at least 10 modules for the Spiral 3 pilot (exactly enough to flesh out a full Endo, or Ara phone skeleton).

It does not appear as though many companies are diving into module-making, likely because they're nervous about the eventual success of Project Ara. That's unfortunate, because if indeed Project Ara works out, forward-thinking and brave companies like Yezz will be the ones to hit the ground running on a new and exciting (and potentially quite lucrative) cottage industry.

It's still possible that Project Ara amounts to nothing more than vaporware, but keep your eyes peeled for developments. Yezz deserves recognition for being gutsy enough to plan a full spate of modules for Project Ara.

Best Security: Qualcomm Sense ID & Vkansee Fingerprint Scanner

Security was an important topic again this year at MWC. With apps for encrypting messages, to virtual machines for sandboxing apps, to encrypted phones claiming to be unhackable, it was difficult to put our finger on just a single solution. So instead, we chose two different fingerprint scanners that overcome the limitations of current capacitive sensors.

Qualcomm's Sense ID is a FIDO-compliant ultrasonic fingerprint scanner that uses sound waves to penetrate the outer layer of skin, imaging not just the ridges and valleys but the sweat pores of your fingertip. Because sound travels well through water, this sensor still works even if your finger is wet or covered in hand lotion, something existing capacitive sensors like Apple's Touch ID have trouble with.

If improved security and accuracy aren't enough, Sense ID's ability to work through glass, metal and plastic means OEMs don't need to cut holes in the cover glass or housing for the sensor, leading to sleeker looking designs.

Qualcomm wasn't the only company to show off cool fingerprint scanning technology, though. Vkansee, a company with a background in forensic fingerprint scanning, revealed its new UTFIS optical scanner.

Replacing the bulky prism-and-lens system used by traditional optical scanners with a thin film Matrix Pinhole Imaging Sensor (MAPIS), this new technology can be easily integrated below the cover glass of smartphones and tablets. With a 2000ppi imaging resolution, Vkansee's scanner can also detect details such as sweat pores. Based on the demo we saw, this optical scanner also works with wet fingers.

Both of these technologies offer big improvements in security and accuracy over existing capacitive sensors and should find their way into products later this year.

Best IoT Innovation: Rambus Lensless Smart Sensor (LSS)

Even though the Internet of Things (IoT) may be the most overused tech buzzword in recent memory, what it actually refers to is a huge change in how technology will impact our daily lives. In the near future, our interactions with technology won't just be limited to when we use our computers, phones, wearables and now increasingly, our cars. Soon almost all our everyday objects, from kitchen appliances to street lights to the next generation of medical devices will all be interconnected and sending data about their surroundings to each other. By the year 2020, there could be up to 50 billion connected devices, and many of them will require sensors to interact with the world.

Rambus, who many of you probably only know from the days of RDRAM, was at MWC to show off the final version of its revolutionary Lensless Smart Sensor (LSS). What Rambus has come up with is a tiny (as in smaller than a human hair) image sensor that doesn't use a lens like a traditional optical sensor but collects data through an inexpensive diffraction grating.

Although this initially produces an unintelligible image, by using algorithmic computations built right into the sensor, a recognizable, yet still low-resolution image can be reconstructed. This data can then be used for gesture recognition, object recognition, and even depth and range tracking. Another benefit of the LSS's simplicity is that it has extremely low power requirements. We were told it can run for 10 years on a single watch battery.

At the same time, the information capture is not detailed enough to infringe on your privacy. Although it can recognize that it is looking at a face, it can't tell whose face it is. With the growing number of IoT devices sensing and potentially recording our every action, designing technology that respects your privacy is paramount, and Rambus understands this.

Best Value: Sony Xperia M4 Aqua

There were many new phones released at MWC, and two of them were flagship devices from two of the leading smartphone vendors. We definitely thought both the HTC One M9 and Samsung Galaxy S6 were great devices and could have given either them an award for best phone of the show (in fact we did give the S6 another kind of honor).

However, this year we wanted to recognize that not everybody needs (or can even afford) a $600 - $700 (and up) flagship phone. There were quite a few good mid-range phones announced at the show, and while many of them won't be finding their way to North America, one of them, the Sony Xperia M4 Aqua, will be, and it's our pick for the best value in a handset.

At first glance, the Xperia M4 Aqua looks very similar to Sony's current flagship, the Xperia Z3 — which is great, because the Z3 is one of the best-looking of the current Android flagships. The M4 trades much of the Z3's metal for plastic, but it retains the attractive glass front and back. Like the Z3, the M4 is also waterproof, but it one-ups its older brother by having an uncovered USB charging port while maintaining the IP68 rating.

Inside, it is powered by Qualcomm's mid-range Snapdragon 615 SoC, which sports a 64-bit octa-core CPU (made up of two quad-core A53s, one running at 1.7GHz and one at 1.0GHz) and an Adreno 405 GPU. This should be more than powerful enough to keep things running buttery-smooth on the M4's 5-inch IPS HD display.

The M4 Aqua has a good camera for a mid-range phone, with a 13MP Exmor RS sensor on the back and a 5MP on the front. For software, the M4 Aqua will be the first Sony handset to ship with Android 5 Lollipop, and the build of the OS we tried at MWC was an attractive combination of Google's Material Design language and Sony's own look and feel.

The Xperia M4 Aqua will start at €300, which is about $350 USD, and will be coming this spring to the U.S. initially as an unlocked device directly from Sony. It will also be coming to Canada, both as an unlocked device and from yet-unnamed carriers.

Biggest Surprise: HTC Vive VR

On the first day of this year's MWC, we expected to see two important things: the latest phones from HTC and Samsung. However, in the run-up to Sunday's HTC event, hints had been dropped by HTC on social media that we should expect a surprise. However, when HTC's CEO Peter Chou announced the HTC Re Vive we were shocked!

Everyone expected that the rumored surprise would either be a bigger phablet-sized phone or some kind of wearable, not a virtual reality headset. What's more, the Vive has been developed in partnership with Valve. We all knew that headsets were coming for Valve's SteamVR platform, but we never guessed that its initial partner would be HTC. We also didn't expect to see a VR headset announced at MWC when the Game Developers Conference (GDC) was taking place the same week. (It turned out, though, that the Vive was demoed at both events).

Still, while the HTC Vive was the biggest surprise of the show, the question remained how good it would actually be. Could HTC, a company one usually doesn't associate with gaming products, create a VR headset that could compete with the efforts of Oculus and Sony, even with the backing of gaming giant Valve? Well, after trying it out at both MWC and GDC, we can report that HTC and Valve have knocked it out of the park. The Vive is currently the best VR experience you can get (that is, if you can get your hands on a developer kit). This year's MWC "best surprise" was also one of the best surprises of any tech show in recent memory.

Most Controversial: Samsung Galaxy S6 & S6 Edge

Although we didn't give it an official award, Samsung's Galaxy S6 and Galaxy S6 Edge couldn't escape our scrutiny. Perhaps placing form over function, the sleek, all-metal and -glass construction of its latest flagship phones are like no other Galaxys Samsung has ever produced.

Inside the shimmering exterior, Samsung packed in some serious technology, including an Exynos 7420 SoC built on the latest 14nm FinFET process with four of the fastest 64-bit ARM Cortex-A57 CPUs we've seen (along with four lower power A53 cores). The speedy SoC is paired with 3GB of high-bandwidth LPDDR4 RAM and up to 128GB of internal UFS 2.0 NAND.

All of the latest wireless capabilities are also onboard, including 802.11a/b/g/n/ac 2x2 MIMO Wi-Fi with speeds up to 620Mbps, Bluetooth 4.1 LE, NFC, IrLED, and a LTE Cat 6 modem. There's even built-in wireless charging, covering both the WPC (Qi) and PMA standards.

The fingerprint scanner has been upgraded from a swipe-style to a more user-friendly touch capacitive sensor, which pairs nicely with Samsung Pay, a wireless payment solution supporting both NFC and Magnetic Secure Transmission (MST). Both cameras are vastly improved, and the Android Lollipop-based OS is more refined and easier to use.

With such a long list of enhancements, you might be wondering why it didn't win a Best of Show award. Although the technology inside the new S6 models is certainly impressive (and might not be matched by any other phone this year), Samsung sacrificed the very features that attracted people to the Galaxy brand — namely, a removable battery and microSD card support (IP67 dust and water resistance is another casualty of the new design).

Will the radical new look of the S6 and S6 Edge attract new buyers to the Galaxy brand as Samsung hopes, or will its betrayal of the Galaxy faithful backfire? It's going to take a while for this controversy to be resolved.

Does Samsung's Star Still Shine Bright?

Samsung’s Galaxy S line of smartphones employs cutting-edge hardware and more features than any human can remember to use, making it one of the most anticipated flagship phones every year. The Galaxy S5 was no exception, generating a lot of hype amidst swirling rumors regarding new hardware and features.

The prelaunch S5 certainly sounded spectacular, with rumored features including a larger than five-inch display, QHD resolution, a 64-bit Exynos or Snapdragon 805 SoC, 4GB RAM, a 16MP camera with optical image stabilization, fingerprint scanner, metal body, a larger and faster-charging battery, along with some top secret alien technology. However, the actual Galaxy S5 that Samsung produced wasn’t nearly that otherworldly, and many of Samsung’s fans were disappointed.

While some of those features did find their way into the Galaxy S5 LTE-A variant released later in the year, the original Galaxy S5 still includes with some great hardware. Screen size increases slightly to 5.1 inches, compared to the S4's five-inch screen, but maintains the same HD resolution. The new SAMOLED panel delivers better peak brightness and color accuracy, however, than the S4's previous-generation display.

The S5’s rear camera uses Samsung's own 16MP ISOCELL sensor, which improves color fidelity and reduces noise at higher ISO settings by limiting crosstalk between adjacent pixels. The S5 is also the first smartphone to use phase detection autofocus (PDAF), vastly improving autofocus performance.

Apple’s Touch ID feature in the iPhone 5s, though not the first smartphone to use a fingerprint sensor, seems to have sparked a new trend, with several flagship devices now using them. Samsung, unwilling to cede features to Apple, also includes a fingerprint scanner in the S5, a first for its mobile devices.

Another interesting addition, one that carries over from the Galaxy S4 Active, is IP67 dust and water resistance. While Sony has included ingress protection on its Xperia phones for some time, the S5 is the first non-specialized, flagship model from Samsung to include it. This is definitely a feature that makes sense for a smartphone.

Samsung Galaxy S5 Tech Specs

While the S5 didn't receive the newer Snapdragon 805 SoC, Qualcomm's 801 is a significant upgrade over the Snapdragon 600 its predecessor came with. Its 2GB of RAM is standard for five-inch flagship phones, as is the maximum internal storage of 32GB. The S5 does support microSD cards up to 128GB if your storage needs grow beyond the S5's internal NAND.

Along with the bigger screen, the S5 grows slightly in all three dimensions, resulting in a 12% weight gain over the S4. While smaller and lighter overall than the HTC One (M8), it's roughly equivalent to the Motorola Moto X in size and weight, despite the Moto's larger screen.

More internal volume creates space for a bigger battery, which should give the S5 more longevity than the S4 and newer Moto X, and be comparable to other current-gen five-inch phones.

Availability And Options

The Galaxy S5 is selling in 125 countries from a large number of retailers. It's available for all four major U.S. carriers and almost every national and regional carrier in Canada, including Bell, Eastlink, Fido, Koodoo, MTS, Rogers, Sasktel, Telus, Videotron, Virgin Mobile and Wind.

Besides the choice between 16 or 32GB of internal storage, the S5 is available in four different color options (from left to right): Copper Gold, Electric Blue, Charcoal Black and Shimmery White. Unfortunately, most carriers only offer the Charcoal Black and/or Shimmery White variants.

Accessories

Included with the Galaxy S5 is a 2.0A wall charger, USB 3.0 cable and stereo headphones with inline controls and extra silicone ear tips.

In terms of hardware, the S5 struggles to float to the surface in a rising sea of Android-based smartphones. Its most compelling feature is dust and water resistance, which it'll need if Samsung's fingerprint scanner, new 16MP ISOCELL camera sensor or software additions don't provide some buoyancy.

Galaxy S5 Look And Feel

Samsung has invested heavily into building marketshare and a reputation for creating all-plastic, cheap-looking devices. The Galaxy S5 is no exception. Offering no visual improvement over the Galaxy S4, the S5 highlights a period in Samsung's history of increasing profits and an apathy towards design. Since the S5's release, increasing competition, primarily from Chinese handset makers, has led to a marked downturn in sales and, subsequently, more attractive phones (Galaxy Alpha, A3/A5, Note 4). The design of the S5 is only significant in that it marks the end of an era.

The S5 is patently Samsung, and even Apple's lawyers will have a hard time confusing it with an iPhone 6. The clickable home button remains an easily recognizable rounded rectangle, and while it requires a larger lower bezel than a phone with on-screen controls, it does allow for a smaller bezel than the iPhone with its circular button. Capacitive back and app-switching buttons flank the home button on the right and left, respectively. Since the inclusion of a fingerprint scanner necessitates a physical home button, it makes sense to use the required bezel area for capacitive buttons and forgo on-screen controls.

Above the screen and to the left of the centered speaker and Samsung logo are a notification LED and ambient light sensor. The proximity sensor, gestures sensor (for detecting Air View and Air Gesture motions) and front-facing camera are on the right side, with the camera in the upper-right corner.

Around the perimeter of the S5 is a ribbed chrome band made of plastic. In my opinion, this is one of the ugliest phone adornments I've had the displeasure to see. It looks cheap and out-of-date. With Samsung moving to a much more stylish and premium-feeling aluminum band in later devices, hopefully we will never see this corrugated siding again.

The 3.5mm headphone jack is located on the top edge, as is the noise canceling microphone. There's also a temperature and humidity sensor that's used for calculating calories burned within S Health and an IR transmitter for controlling devices old-school.

The long, slim volume rocker is on the left edge near the top, while the power button is in a standard position on the right edge, also near the top. Both buttons make a positive click when pressed, and they fit snugly in the frame without any looseness or rattle.

The second microphone is located on the bottom edge. Like the Note 3 before it, the S5 uses the rather ridiculous micro-USB 3.0 port, but hides it behind a flap to keep out dust and water.

For this iteration, Samsung ditches the shiny plastic back, replacing it with a stippled pattern and a matte finish. The soft-touch, almost rubbery plastic makes the S5 easy to grip and less likely to slip from your hand. While we definitely like the in-hand feel of the new back panel, opinions regarding its appearance vary; some people like the professional look. Personally, I think it looks like a golf club handle.

The primary camera is centered on the back near the top edge and sits within Samsung's signature protruding rounded rectangle. Below the camera and recessed below the surface are the LED flash and heart rate sensors. Both the camera and flash are trimmed with a plastic chrome ring to match the shiny chrome sides.

The single speaker is also located on the back near the bottom-left corner. Why OEMs continue to place the speaker on the back of phones baffles me. In a best-case scenario, it's pointing away from you. If you're holding the phone or resting it on a table, the speaker is muffled by your hand or the table top.

Overall, the Galaxy S5 doesn't look much different than the S4, and it isn't going to win any design awards. Opinions will vary, but to me the S5 is the ugliest flagship phone on the market. The only real positive is the new soft-touch feel on the back, which makes it easy to hang on to, assuming the dated aesthetics don't keep you from picking it up in the first place.

Taking A Look Inside: Dust And Water Resistance

One feature that's been gradually gaining popularity in smartphones over the past few years is protection from dust and water. While it does add to the BOM cost of a phone, its usefulness is readily apparent to anyone who has brought their phone to the beach or dropped it in the toilet. Sony is one of the biggest proponents of this functionality, with several of its Xperia phones offering protection from the elements. Several other OEMs, like Motorola and now HTC with the Desire Eye (IPX7), also include some form of protection for select phones. Samsung too has offered specialized versions of its phones with environment protection like the Galaxy S4 Active. The Galaxy S5, however, marks the first time Samsung offers this feature on a conventional flagship model.

With an IP67 rating, the S5 is completely sealed against dust contamination and can be immersed in liquid up to 1m deep for 30 minutes. This presents special design challenges to ensure all buttons, speakers and ports are sealed. Other parts like the camera module and LED flash/heart rate sensor module are already airtight. The removable back cover requires a rubber gasket, not to keep water out of the cover, but to isolate the battery and openings that lead to to the interior of the device (such as the microSD and microSIM slots).

While the headphone jack does not require any sort of plug, the micro-USB 3.0 port requires a gasketed flap to keep dust and water out. This flap is not unique to the S5, as other IP-rated phones like Sony's Xperia Z series also use a flap to seal the USB port. This flap is a mild inconvenience, since it needs to opened and closed every time you want to charge the phone; however, the added protection it provides is probably worthwhile.

After every charge and boot sequence, the phone reminds the user to preserve the IP67 rating by securing the back cover and making sure the USB flap is closed. This reminder gets old fast, but is necessary considering how rushed and absentminded people can be—just think about how many folks you see driving around with their car's gas cap dangling against the fender.

Removing the back cover reveals the sensitive parts the rubber gasket needs to protect. In addition to the battery and card slots, the pins for connecting the optional wireless charging cover also need protection.

Above the battery and to the right of the camera are the microSIM and microSD card slots. While it's good to see the inclusion of the microSD card slot (a staple of Samsung phones), we are a bit surprised to see Samsung continue to use microSIM cards.

The company continues to provide a removable battery for its Galaxy phones (in this case 2800mAh). Since the battery houses the NFC components, care needs to be taken when ordering additional batteries to ensure they support NFC.

Having dust and water resistance is a plus for the Galaxy S5. The flap covering the USB port adds some inconvenience, but it's a small price to pay for the additional protection it affords your phone.

Fingerprint Scanner

The Galaxy S5 is one of the few smartphones that offers a fingerprint scanner as a means for authentication. Incorporated into the Home button, Samsung reduces cost and size by using a swipe-style sensor. Since the sensor is too small to scan your entire fingerprint at once, you must slide your finger over the sensor, which takes a series of measurements and then stitches them together via software. The drawback to this approach is reduced accuracy, as the process can be sensitive to the speed and angle with which you slide your finger over the sensor and algorithmic artifacts created while stitching the individual images together. The passive capacitive sensor in the S5 is also less accurate than the active capacitive sensor technology Apple acquired from AuthenTec in 2012. It will be interesting to see how well the S5’s sensor works compared to Touch ID.

Registering a fingerprint is a simple affair: just swipe your finger across the sensor eight times. Unlike Apple's Touch ID setup process that requires you to vary how the finger is placed on the sensor, the S5 doesn't ask for any variation in how you swipe during its setup procedure. Up to three fingerprints can be stored, versus five for Touch ID.

When the S5 first launched, the fingerprint scanner was ridiculed for its poor accuracy. In our initial tests, we saw our success rate hover around 50%. However, subsequent software updates have improved accuracy substantially. I'm actually quite surprised at how well it works now, with accuracy nearly identical to Touch ID. I say nearly because the S5's swipe sensor is still less forgiving than Touch ID when it comes to finger orientation. Still, I found that I could vary the angle of my finger by up to about 30º before the scan would fail and even varying the swipe speed didn't affect accuracy.

With performance nearly equal, the distinguishing factor between Apple's Touch ID and Samsung's implementation is ease of use. Only requiring a finger to touch the sensor, in any orientation, makes Touch ID more pleasant in practice. The S5's sensor is bit awkward to use because your finger needs to be swiped over the sensor, which is located close to the bottom edge of the phone. Swiping a finger lengthwise either requires the use of two hands or setting the phone on a flat surface. The scanner can actually be used while holding the phone normally with one hand by swiping your thumb diagonally. This method is much more convenient and natural, although accuracy decreases slightly.

By default, the fingerprint scanner can only be used to unlock the phone or access Samsung services. However, with the right apps installed, it can also be used for authenticating PayPal transactions and other FIDO Ready online services, adding to its usefulness.

Similar to Apple's Touch ID, the S5's fingerprint scanner didn't work exceptionally well when it first shipped. But with some additional fine tuning in software, both biometric solutions have matured into useful authentication features.

Camera: Hardware

When the Galaxy S5 launched as Samsung's flagship device at the beginning of 2014, it came with a couple of key advancements, including Samsung's own ISOCELL camera sensor and phase detection autofocus (PDAF). In the months that followed, several other flagship phones appeared, each with their own camera improvements. So, can the S5's camera still compete with these newer devices?

Samsung Galaxy S5 Camera Specs

The Galaxy S5 still has one of the highest megapixel counts, the newer Xperia Z3 being one of only a few phones to have more. However, packing more pixels into the same sensor area results in smaller pixels that capture less light.

Balancing resolution with pixel size is challenging, but as you’ll see later in our image quality section, Samsung's ISOCELL sensor is reasonably successful at combining lots of detail with respectable low-light performance. But the cameras in the HTC One (M8) and iPhone 6 still outperform the S5 in challenging light due to their sensors' much larger pixels.

As for the optics in front of the sensor, you can see that all of the phones being compared have similar specifications, apart from the Z3. It does have a lens with both a wide aperture and a wide focal lens to allow you to fit more of what you see in the frame. Other than the new sensor, the S5’s other camera specs match those found on its predecessor, the S4.

Samsung's ISOCELL Sensor and Phase Detection AF

In previous Galaxy phones, Samsung relied on using image sensors from other vendors, namely Sony. The Galaxy S4 used the 13MP IMX091PQ and the Note 3 used the 13MP Sony IMX135 Exmor RS, which added 4K video (the Note 3 being one of the first phones to record in UHD). For the S5, Samsung decided to change things up and use one of its own ISOCELL sensors announced back in 2013.

Samsung claims that its ISOCELL technology “substantially increases light sensitivity” and allows for “higher color fidelity even in poor lighting conditions.” This is despite the fact that the 1/2.6” sensor’s pixels are still a small 1.12µm, the same as the S4 and many other phones using a 13MP Sony sensor. The way this is achieved is by adding additional barriers between adjacent cells, thereby reducing the crosstalk between them. This short video by Samsung gives you a better idea of how its ISOCELL technology works:

It's interesting to note that the Galaxy S5’s sensor shoots natively at 16:9. At the time of its release, it was the only other phone besides HTC's One M7 and M8 that shot in this aspect ratio. When the Note 4 was released later in the year, its native aspect ratio (despite using a different sensor) was also 16:9. It appears that Samsung will be adopting this standard for the foreseeable future.

The other innovation that Samsung introduced with the Galaxy S5 is phase detection autofocus (PDAF), the S5 being the first smartphone to implement this technology. PDAF works in addition to the traditional contrast AF system (found in most other smartphones) to give the S5 much faster focusing times. Contrast AF uses the SoC’s ISP to compare the contrast of nearby pixels at different focal distances to determine focus, and while this system is not processor-intensive, it is relatively slow. PDAF is a lot faster (Samsung claims the S5 can focus on its subject in as little as 300ms), and works by calculating the distance between the light refracted by microlenses superimposed on the sensor. Samsung's implementation works well, and seems just as fast as the PDAF system used by the iPhone 6.

Camera: Software

The Galaxy S5’s camera UI is laid out well and uncluttered. Since the native 16MP resolution shoots in a 16:9 aspect ratio, the viewfinder fills the entire screen and some of the buttons on the sides can obscure what you are shooting.

Samsung keeps the HDR on/off button and the selective focus mode (more on that later) always displayed on the left side of the screen. The grid shown in the image above is not on by default.

The screenshot above shows the variety of user-adjustable settings for both stills and video. Even though the default mode is Auto, there are still a number of manual adjustments that can be made, such as white balance, metering modes, exposure compensation and ISO. They are grayed-out above, since when HDR is selected you are more limited as to what adjustments you can make. The button for picture stabilization is not for OIS (optical image stabilization), but for the much less effective digital stabilization.

As mentioned, the Galaxy S5’s native 16MP resolution shoots in a 16:9 aspect ratio. To shoot in the more common 4:3 aspect ratio, the S5’s resolution needs to be lowered to 12MP, which in most cases should still produce sufficiently detailed images.

When the Galaxy S4 launched in 2013, it was ridiculed for its overwhelming number of camera modes. Thankfully Samsung saw the error of its ways and reduced the number of pre-installed modes on the S5 to just six (though "Shot & more" includes five sub-modes that were separate modes on the S4). At the end of the list is a download option for acquiring additional modes, most of them being ones that were originally included with the Galaxy S4. Along with these modes, the S5 also has the obligatory selection of Instagram-like photo filters.

Selective Focus

Selective focus is a new mode that was not found on the Galaxy S4. It leverages the power of the Snapdragon 801’s ISP to create DSLR-like images with an out-of-focus background. This helps separate and emphasize the subject in the foreground from the rest of the image.

When the Galaxy S5 was released, this mode was quite novel. But ten months later we now see some variation of it on nearly every flagship Android smartphone. For this effect to work properly, the subject has to be quite close to the phone and reasonably far from the background. Also, the S5 only provides two focal distances, near and far, which is unlike similar modes on phones like the LG G3 or HTC One (M8) that provide additional focal points in between these two extremes.

"Shot & more" Mode

The "Shot & more" mode applies different effects to a burst sequence of images. There are options for selecting the best photo within a sequence, combining multiple images to create a shot with multiple exposures and for creating a panning shot with a dramatically blurred background. The "Eraser" effect is the most interesting. It uses the sequence of images to remove moving objects from the scene.

Dual Camera Mode

Dual camera mode creates picture-in-picture images with both the front and rear camera. You can resize and position the smaller image anywhere in the picture, and there are different frames and shapes for the smaller image to choose from, including the postage stamp shown above, a heart, a Polaroid, and more.

Video

UHD (or 4K) video was introduced on the Snapdragon 800-powered Samsung Galaxy Note 3 in 2013. In 2014, it became almost a standard feature, with nearly every flagship phone other than the HTC One (M8) and iPhone 6 supporting it. UHD video on the S5 is recorded at 30 FPS and about 50 Mb/s, which is almost as high as the OnePlus One's 57 Mb/s. Unfortunately, HDR isn't available while recording in 4K, so video quality suffers in lighting conditions with high dynamic range.

While being able to record UHD video was of questionable value when the S5 launched last year, with the wider availability of lower-priced 4K monitors and TVs, it is now a useful feature to have. Still, that doesn’t change how much capacity it monopolizes, so investing in a large microSD card for additional storage is a must if shooting a lot of UHD video.

The S5 also captures 1080p at 30 FPS and ~17 Mb/s, lower than both the LG G3 and OnePlus One, which record at a slightly higher 20 Mb/s rate. Unlike the G3 however, the S5 supports 1080p at 60 FPS and an average of 28 Mb/s with its smooth motion setting that reduces the jerkiness of moving objects.

There are several other recording modes available, as the image above shows. Slow-motion video can be captured at three different speeds. At 1/2 speed, the S5 captures 720p content at 60 FPS and outputs a file that plays at a standard 30 FPS. At 1/4 speed, the S5 should capture 720p video at 120 FPS. However, it doesn't appear to do so. Instead, it appears to reduce resolution in order to reach 120 FPS, upscaling it back to 720p at 30 FPS in the final video file. This results in a noticeably lower-quality video. Quality gets even worse or rather unusable at the 1/8 setting, which outputs a heavily upscaled video (with very blocky artifacts) at 720p and 15 FPS. It doesn't appear that the S5 is utilizing the Snapdragon ISP for this feature, since it can only capture true 720p at 60 FPS and a lower resolution at 120 FPS. Other phones using the Snapdragon 801 SoC don't have any issues capturing a full 720p at 120 FPS.

The three fast-motion modes (x2, x4, x8) don't actually record at a higher speed, instead recording 30 FPS video and then compressing the time in software. For instance, the x8 fast mode requires eight seconds of actual recording time at 30 frames per second to produce one second of video, taking 240 frames of video and showing them in one second at 30 FPS. This effect works as advertised, but it doesn't seem particularly useful.

Camera: Photo Quality

In this section, we will evaluate the quality of images shot with the Galaxy S5 in different lighting conditions and examine its HDR mode, which is important for getting good results from the S5’s camera. All images were taken using the Auto mode unless noted. Also, you can view the full-sized image for each photo by either clicking on the stand-alone image or clicking the text links below the images that are within a slideshow album.

Daylight Images

Full Size Images: [Galaxy S5 outdoor daylight sample 1: f/2.2, 1/438 sec, ISO 40], [Galaxy S5 outdoor daylight sample 2: f/2.2, 1/50 sec, ISO 40], [Galaxy S5 outdoor daylight sample 3: f/2.2, 1/33 sec, ISO 64], [Galaxy S5 outdoor daylight sample 4: f/2.2, 1/848 sec, ISO 40], [Galaxy S5 outdoor daylight closeup 1: f/2.2, 1/432 sec, ISO 40], [Galaxy S5 outdoor daylight closeup 2: f/2.2, 1/100 sec, ISO 80]

The first image with the car was taken in bright outdoor conditions without the use of HDR. There's a lot of detail in that shot thanks to the S5’s 16MP sensor, and the colors are bold but still accurate and natural-looking. At ISO 40 there is no noise, and the PDAF quickly focused on the car and captured the shot before it drove away. The only issue is that the dynamic range of this picture isn’t particularly good, with the shadows being too dark. In general, we found that without HDR turned on, the photos we took with the S5 often had dynamic range issues, which can be seen in some of the other daylight images above.

Indoor Images

Full Size Images: [Galaxy S5 indoor sample 1: f/2.2, 1/33 sec, ISO 64], [Galaxy S5 indoor sample 2: f/2.2, 1/17 sec, ISO 320], [Galaxy S5 indoor sample 3: f/2.2, 1/109 sec, ISO 40], [Galaxy S5 indoor sample 4: f/2.2, 1/50 sec, ISO 40], [Galaxy S5 indoor sample 5: f/2.2, 1/33 sec, ISO 125], [Galaxy S5 indoor sample 6: f/2.2, 1/33 sec, ISO 40], [Galaxy S5 indoor sample 7: f/2.2, 1/33 sec, ISO 400]

The Galaxy S5 struggles a little more used indoors. The image of the ties above again demonstrates that it reproduces colors accurately, but the Auto mode favors ISO over shutter speed, resulting in a little blurring from camera shake if you look closely. We’d have preferred to see the ISO raised a little to speed up the shutter, which would result in a crisper hand-held image. The rest of the indoor images have the same problem of shutter speed being set too low. But again, the colors look great, and the white balance is spot on. However, as the ISO value increases, even at the still relatively low ISO 320 and 400 of the first and last image, the S5’s noise reduction post-processing starts to muddy up the picture and destroy detail.

Low Light Images

Full Size Images: [Galaxy S5 low light outdoor sample 1: f/2.2, 1/213 sec, ISO 40], [Galaxy S5 low light outdoor sample 2: f/2.2, 1/33 sec, ISO 200], [Galaxy S5 low light outdoor sample 3: f/2.2, 1/17 sec, ISO 2000], [Galaxy S5 low light outdoor sample 4: f/2.2, 1/33 sec, ISO 125], [Galaxy S5 low light indoor sample 1: f/2.2, 1/17 sec, ISO 2000], [Galaxy S5 low light indoor sample 2: f/2.2, 1/17 sec, ISO 2000], [Galaxy S5 low light indoor with flash: f/2.2, 1/30 sec, ISO 125]

Even though the first photo was taken at dusk, there was still enough light for the S5’s camera to set the ISO to a low 40 and shutter speed to a fast 1/213s, creating a great-looking image even zoomed in close. In the remaining low-light pictures, the S5 really begins to struggle.

At higher ISO settings, noise becomes a problem. The first outdoor image in the slideshow above was taken at ISO 200, and its noise reduction algorithm is already smearing detail and creating a "soft"-looking picture. The third one, shot at ISO 2000, is an almost unusable blotchy mess. The last two images shot indoors also have the same problems, while the last image shows that the S5’s flash is bright and effective. Unfortunately, Samsung's ISOCELL sensor doesn't handle low-light conditions well enough to make the S5 a top performer. Still, all smartphone cameras struggle with low light and the S5's performance isn't substantially worse than most of its competitors.

HDR Mode

Full Size Images: [Galaxy S5 HDR: f/2.2, 1/247 sec, ISO 40], [Galaxy S5 No HDR: f/2.2, 1/334 sec, ISO 40]

The Galaxy S5’s HDR mode is an integral part of its imaging experience. Unlike most phones, where HDR is a setting buried in the options menu, the HDR button is right on the main screen of the camera UI. It's a good thing too because the S5 has a bit of a problem with dynamic range when shooting outdoors (see image above). Therefore, it's a good idea to keep HDR on in that environment.

Full Size Images: [Galaxy S5 example 1: f/2.2, 1/264 sec, ISO 40], [Galaxy S5 example 1 (HDR): f/2.2, 1/289 sec, ISO 40], [Galaxy S5 example 2: f/2.2, 1/265 sec, ISO 40], [Galaxy S5 example 2 (HDR): f/2.2, 1/254 sec, ISO 40], [Galaxy S5 example 3: f/2.2, 1/1532 sec, ISO 40], [Galaxy S5 example 3 (HDR): f/2.2, 1/1316 sec, ISO 40]

The most impressive aspect of the Galaxy S5’s HDR mode is that it's displayed real-time in the viewfinder, leveraging the power of the Snapdragon 801’s ISP. This lets you see how the HDR picture will appear before taking it, and removes the processing delay after each shot. We also found that the S5’s HDR algorithms produced a fairly natural-looking picture with more detail in the shadows.

Overall, the S5's camera performs well in good lighting as long as HDR mode is left on. The 16MP sensor provides a lot of detail and white balance is accurate, but when the light dims, so does does its performance. Samsung's ISOCELL sensor fails to offer a substantial improvement in low-light capability, leaving noise reduction to overly aggressive software. It's too bad Samsung didn't give the S5 OIS, which would allow it to leave the shutter open longer and capture more light.

Software Tour: TouchWiz

The Galaxy S5 comes with Android 4.4.2 (an update to Android 5.0 has already begun rolling out in select countries) and Samsung's latest version of the infamous TouchWiz interface. Criticized for its cluttered UI, laggy performance and glut of gimmicky features, the sight of TouchWiz sends Android fans searching for custom ROMs. With its latest incarnation, however, Samsung scales things back at least a little bit.

The UI design is ostensibly TouchWiz, from the home screen layout to the brightly-colored icons. At least until the Android 5.0 update arrives, Samsung uses black backgrounds (helping the AMOLED screen save power) with blue overlays and green accents. Common apps like Phone, Contacts and Calculator use this same theme; however, the visual cohesiveness breaks down when using apps from Google or carrier-specific apps, since they all use a different design language.

The left-most home screen is reserved for My Magazine, a Flipboard-powered news and social media aggregator that groups articles, posts, pictures and status updates into a scrollable list, not unlike BlinkFeed from HTC. A Flipboard account is required to use this feature, or it can be turned off completely from within the home screen settings.

As with any Android device, the home screen is customizable. Long-pressing on the wallpaper brings up the standard options to add home screens, set the default home screen, select new wallpaper and add widgets. Unlike other OEM ROMs, TouchWiz doesn't provide any specialized widget pages, animated wallpapers or theming capability (themes can be found in the Samsung app store though).

Samsung provides a few options for sorting and organizing the App Drawer. Apps can be sorted alphabetically or arranged manually however you like. There's also an option to cut through the clutter and only show downloaded apps. To keep icons from piling up, they can be organized into folders, hidden from view or uninstalled directly from the app drawer.

The notification shade carries the TouchWiz style, but unlike LG's cluttered implementation that leaves little room for actual notifications, buttons and controls occupy less than half the screen. The side-scrolling top row of hardware toggles is customizable, and by clicking the icon in the upper-right corner, it expands into a grid view. The overall functionality is pretty standard for Android.

Like many larger-screened phones, Samsung offers a one-handed mode for the S5. Activated by quickly swiping your thumb inwards from the right edge of the screen and then back again, it collapses the screen into a smaller, resizable window that's anchored to the lower-right corner. With the shrunken screen, it's easy to reach all of the on-screen elements with your thumb while holding the phone normally with one hand. Samsung even adds a toolbar along the bottom of the display that contains the standard Android navigation buttons and volume controls.

Samsung utilizes the front-facing camera and proximity sensor for a few of its gesture-based controls. With "Smart stay", the screen remains on as long as you're looking at it, and if you look away while watching a video, it will pause it for you. This last feature sounds more annoying than useful to us, so we're glad it can be turned off. There's also "Air browse", which allows you to scroll lists by moving your hand over the screen in the direction you want to scroll, either up-down or left-right. This actually does work well with either your palm facing down or to the side and up to about six inches above the screen. Its scrolling utility is limited, though, to only the body of emails, moving between pictures in Gallery and changing the tracks in the Music app or lock screen.

Rather than overload the S5 with apps that most people will never use, Samsung makes them optional downloads from its app store. This keeps clutter to a minimum and doesn't waste storage space. Unfortunately, most carriers won't be this sensible, continuing to ruin the user experience with bloatware.

Overall, the S5's version of TouchWiz is an improvement over the Galaxy S4's variant. Excess apps have been relocated to Samsung's app store and some of the extra features that didn't add any value have been removed. TouchWiz still isn't our favorite OEM ROM, but we can at least live with it now.

How We Tested

Benchmark Suite

Our current Android test line-up comprises six key sections: CPU, Web, GPU, GPGPU, Display and Battery.

Test Methodology

All handsets are benchmarked on a fully updated copy of the device's stock software. The table below lists other common device settings that we standardize to before testing.

Furthermore, for browser-based testing on Android, we're employing a static version of the Chromium-based Opera in order to keep the browser version even across all devices. Due to platform restrictions, Safari is the best choice for iOS-based devices, while Internet Explorer is the only game in town on Windows RT.

Comparison System Specs

Qualcomm had a strong year in 2014, with nearly every flagship smartphone running a Snapdragon SoC. The Snapdragon 801 was particularly popular, showing a good balance between performance and efficiency. Its ubiquity isn't necessarily bad for consumers, but it does make our benchmark charts rather boring.

The table below contains all the pertinent technical specifications for today’s comparison units:

Other than the iPhone 6, all of the Android phones we'll be comparing use the Snapdragon 801 SoC. The Galaxy S5 uses the highest-clocked 8974AC version just like the G3, One and Z3. Only the HTC One (M8) uses the lower-clocked 8974AB variant, which should give it slightly lower CPU performance.

Again, other than the iPhone 6 with its PowerVR GPU, all of the Android phones use the Adreno 330 graphics engine. The only differentiator for GPU performance should be the G3's higher WQHD screen resolution.

We should see some variation in longevity, though. Both the G3 and OnePlus One have slightly larger batteries than the S5, but they also have larger screens to power. The iPhone 6 has by far the smallest battery. Will its smaller, lower-resolution screen make up for this deficit? The Sony Z3 has a bigger battery than the S5 and almost the same size screen. This should give the Z3 an edge in battery life.

Results: CPU Core Benchmarks

AnTuTu X

AnTuTu is an Android system benchmark designed to test the performance capabilities of four major aspects of mobile devices: Graphics (encompassing 2D, UI and basic 3D), CPU (fixed, floating-point and threading), RAM (read and write) and I/O (read and write).

The results for AnTuTu X are sorted based on the overall scores, which are the values shown within the left end of each bar. The overall score is not simply a cumulative value based on the individual test scores. Instead, it’s a separate calculation with its own distinct value. This is why the length of the bars don’t correlate with the overall score.

The total spread between the LG G3 and the OnePlus One is 17%. Both the S5 and G3 do well in the UX, RAM and CPU tests, all of which stress threaded CPU performance to some degree, but fall short when testing the GPU. The opposite is true for the OnePlus One and Sony Z3; they both score higher in the GPU test than the other three. The One and Z3 are both newer phones running Android 4.4.4 (versus 4.4.2 for the others), so it's possible they have newer GPU drivers as well.

As expected, the HTC One (M8) posts the lowest scores in the CPU-centric tests due to its lower CPU clock frequency.

AndEBench Pro

AndEBench Pro is produced by The Embedded Microprocessor Benchmark Consortium (EEMBC) and uses carefully designed, low-level algorithms for testing CPU, GPU, memory and storage subsystem performance. The benchmark workload includes XML parsing, data compression, GUI rendering, photo manipulation and cryptography tasks.

We don't see any surprises in CPU or GPU performance: all of the devices demonstrate similar performance except the HTC One's lower score in the CPU-centric CoreMark-HPC test. There's some variation in memory performance where the OnePlus One shows about a 8% and 9% advantage in memory bandwidth and memory latency, respectively.

The S5 does poorly in the Storage test, achieving less than half the throughput as the OnePlus One and G3. Looking more closely at the Storage sub-tests in the table below, the G3 performs 2x-3x better in each test. The 256K sequential and random read tests show the G3 with a 3x advantage over the S5.

First letter: S=sequential, R=random
Second letter: R=read, W=write

Basemark OS II Full (Anti-Detection)

Basemark OS II is an all-in-one tool designed for measuring overall performance of mobile devices. It scores each device in four main categories: System, Memory, Graphics and Web. The System score reflects CPU and memory performance, specifically testing integer and floating-point math, along with single- and multi-core CPU image processing using a 2048x2048, 32-bit image. Measuring the transfer rate of the internal NAND storage (Memory) is done by reading and writing files with a fixed size, files varying from 65KB to 16MB, and files in a fragmented memory scenario. Calculating the Graphics score involves mixing 2D/3D graphics inside the same scene, applying several pixel shader effects and displaying 100 particles with a single draw call to test GPU vertex operations. The benchmark is rendered at 1920x1080 off-screen 100 times before being displayed on-screen. Finally, the Web score stresses the CPU by performing 3D transformations and object resizing with CSS, and also includes an HTML5 Canvas particle physics test.

There are no real surprises in Basemark OS II. The overall scores for the 8974AC Snapdragon 801-based devices fall within a 9% range. The score for the lower-clocked 8974AB SoC in the HTC One (M8) is 13% lower than the quickest 801, the Sony Z3.

Geekbench 3 Pro (Anti-Detection)

Primate Labs' Geekbench offers a wide selection of cross-platform compatibility, with apps available for Windows, OS X, Linux, iOS and Android. This simple system benchmark produces two sets of scores: single- and multi-threaded. For each, it runs a series of tests in three categories: Integer, Floating Point and Memory. The individual results are used to calculate category scores, which, in turn, generate overall Geekbench scores.

Both the OnePlus One and HTC One (M8) pull ahead of the S5 in the Memory test, just as they did in AndEBench Pro. The Snapdragon 801 devices all perform the same in the single-threaded CPU tests, with the HTC One (M8) falling behind by 16%

The multi-threaded tests yield more variation. The S5 scores 13% better in both the Integer and Floating Point Multi-Core tests.

The reason we see more variation in the Multi-Core benchmark is because each OEM configures their CPU governors differently. Some optimize for performance, while others optimize for power consumption, placing additional constraints on CPU clock frequency or on how many cores can be active in a given time slice. This behavior is readily apparent when monitoring the real-time CPU usage while running the benchmark. The S5 keeps all four cores running at ~2.5GHz for the majority of the benchmark run. The G3, however, never seems to run all four cores at peak frequency, electing to either run all four cores at reduced frequency or letting two cores go idle while the other two run at max speed. Most of the time this isn't an issue, since apps are generally single-threaded and background tasks are usually short duration. Where this effect will be noticeable is when playing games, which are multi-threaded, and loading apps.

MobileXPRT 2013

Principled Technologies' MobileXPRT 2013 is a modern SoC benchmark for Android. It consists of 10 real-world test scenarios split into two categories of testing: Performance and User Experience. The Performance suite contains five tests: Apply Photo Effects, Create Photo Collages, Create Slideshow, Encrypt Personal Content, and Detect Faces to Organize Photos. Performance results are measured in seconds. The User Experience suite also has five tests: List Scroll, Grid Scroll, Gallery Scroll, Browser Scroll, and Zoom and Pinch. These results are measured in frames per second. The category scores are generated by taking a geometric mean of the ratio between a calibrated machine (Motorola's Droid Razr M) and the test device for each subtest.

The Galaxy S5 performs on par with the Sony Z3, as expected. The HTC One (M8) is 8% slower than the S5, which equals their difference in CPU clock frequency.

The S5 does well in every test except Gallery Scroll. As expected, there isn't much variation in this test, with an overall performance spread of 9%.

In terms of CPU and memory performance, the Galaxy S5 does as well or a little better than its Snapdragon 801 peers, particularly in multi-threaded scenarios. It's unfortunate then that it performs so poorly in the storage test.

Results: HTML5 And JavaScript Benchmarks

The tests on this page are JavaScript- and HTML5-heavy selections from our Web Browser Grand Prix series. Such tests are extremely meaningful to mobile devices because so much of the in-app content is served via the platform's native Web browser. These tests not only offer a view of each device's Web browsing performance, but since these tasks are traditionally so CPU-dependent, browser benchmarks (especially JavaScript-heavy tests) are a great way to measure SoC performance among devices using the same platform and browser.

Browsermark 2.1

Rightware's Browsermark 2.1 is a synthetic browsing benchmark that tests several performance metrics, including load time, CSS, DOM, HTML5 Canvas, JavaScript and WebGL.

Performance is essentially identical for all the phones in Browsermark, which is to be expected.

JSBench

Unlike most JavaScript performance benchmarks, JSBench could almost be considered real-world, since it utilizes actual snippets of JavaScript from Amazon, Google, Facebook, Twitter and Yahoo.

The Sony Z3 shows a small 7% advantage over the Galaxy S5, but that's nothing to get excited about. The HTC One (M8) takes 5% longer to complete the benchmark due to its lower CPU clock.

The iPhone 6 Plus dominates this benchmark thanks to Safari’s JIT compiler optimizations.

Peacekeeper 2.0

Peacekeeper is a synthetic JavaScript performance benchmark from Futuremark.

Since Peacekeeper is predominantly a single-threaded CPU test, it's not surprising that the results are similar to those from Geekbench 3 Single-Core. So far, the S5 matches the performance of its peers, avoiding any performance anomalies.

WebXPRT 2013

Principled Technologies' WebXPRT 2013 is an HTML5-based benchmark that simulates common productivity tasks that are traditionally handled by locally installed applications, including photo editing, financial charting and offline note-taking.

The HTC One (M8) scores 14% better than the Galaxy S5 in WebXPRT 2013 despite its lower CPU clock rate. This test seems to have more error, since there's no explanation for this result. The HTC One (M8) has the lowest frequency, but finishes first. The OnePlus One has the highest memory bandwidth, yet finishes last. And the S5 has the best multi-threaded performance and finishes next to last.

Results: GPU Core Benchmarks

3DMark (Anti-Detection)

Futuremark has become a name synonymous with benchmarking, and the company's latest iteration of 3DMark offers three main graphical benchmarks: Ice Storm, Cloud Gate and Fire Strike. Currently, the DirectX 9-level Ice Storm tests are cross-platform for Windows, Windows RT, Android and iOS.

Ice Storm simulates the demands of OpenGL ES 2.0 games using shaders, particles and physics via the company's in-house engine. Although it was just released in May of 2013, the on-screen portions of Ice Storm have already been outpaced by modern mobile chipsets, with Nvidia's Tegra 4 and Qualcomm's Snapdragon 800 both easily maxing-out the Extreme version (1080p with high-quality textures). However, Ice Storm Unlimited, which renders the scene off-screen at 720p, is still a good gauge of GPU-to-GPU performance.

The iPhone 6 and its newer PowerVR Rogue GPU post the best Graphics score in this benchmark. Of the devices using the older Adreno 330 GPU, the HTC One (M8) shows a slim lead, pulling ahead of the S5 by 7%. The LG G3 falls behind the HTC by a more noticeable 15%.

In the multi-threaded, CPU-focused Physics test, the HTC One (M8) trails its Snapdragon 801 peers as expected (due to its lower CPU clock), and the S5 scores well, which agrees with the Multi-Core Geekbench results. The one outlier is the G3, which posts the highest score in the Physics test. This is at odds with the Multi-Core Geekbench test, where the G3 didn't fully utilize all four of its cores at max frequency. Monitoring CPU usage during the 3DMark Physics test confirms the G3 does not limit CPU use and runs all four cores at max frequency. Another interesting twist is that while the G3 for the Canadian market (shown in the charts) scores 16630, the U.S. version on Sprint scores 6% lower (15713), slower than the S5. This same U.S. G3 also scores 18% lower than the Canadian model in the Multi-Core Geekbench Integer test, while the Floating Point and Memory scores are the same. Perhaps this is just statistical error based on our limited sample size, or maybe we're seeing the effect of OEMs tuning devices for specific markets, emphasizing performance in one market and power efficiency in another.

Basemark X 1.1

Based on the Unity 4.0 game engine, Rightware’s Basemark X is a cross­-platform graphics benchmark for Android, iOS and Windows Phone 8. This test utilizes Unity’s modern features via the OpenGL ES 2.0 render path to simulate how a modern game might look and run. Basemark X is an aggressive metric that still hasn’t been maxed out by the latest mobile SoCs.

The on-screen results for the iPhone 6 Plus are not accurate, since the current version of this benchmark doesn’t support its native resolution. Instead, it renders at the native resolution of the 5s (640x1136). The Overall scores are not affected by this issue though, since they are based on the off-screen rendering tests.

All of the Adreno 330-equipped devices fall within 8% of each other when looking at the overall score, with the S5 turning in the lowest value. While this difference isn't significant, it is consistent. The HTC One (M8) posts the highest scores in every test, with the S5 falling just shy of its mark.

The OnePlus One's slightly better memory bandwidth (shown in the AndEBench tests) helps it post a better overall score in the high-quality tests. Again, the difference isn't much, there's only a 9% spread between it and the S5, but the results consistently show the OnePlus One scoring slightly better and the Galaxy S5 with the lowest scores.

GFXBench 3.0 Corporate

Kishonti GFXBench 3.0 is a cross-platform GPU benchmark supporting both the OpenGL ES 2.0 and OpenGL ES 3.0 APIs. It comprises both “high-level” game-like scenarios, along with more “low-level” tests designed to measure specific subsystems.

Among the high-level tests are Manhattan and T-Rex. Manhattan is a modern, complex OpenGL ES 3.0-based test, while the OpenGL ES 2.0-level T-Rex is a holdover from GFXBench v2.7.

The low-level tests include Fill, which measures fill rate by rendering four layers of compressed textures; Alpha Blending, a test that renders layers of semi-transparent quads using high-resolution, uncompressed textures; ALU, for measuring shader compute performance; and Driver Overhead, which measures the CPU overhead of the graphics driver and API by making a lot of draw calls and state changes.

See GFXBench 3.0: A Fresh Look At Mobile Benchmarking for a complete test-by-test breakdown of this benchmark.

In both Manhattan and T-Rex, the Galaxy S5 performs the same as all of the other phones with Adreno 330 GPUs. The LG G3 shows a significant drop in performance in both on-screen tests because of its higher WQHD resolution. The Adreno 330 just wasn't designed to handle this many pixels.

AndEBench Pro already showed the OnePlus One with a slight advantage over the Galaxy S5 in both memory bandwidth and memory latency. It should be no surprise, then, that the One shows a 14% advantage over the S5 in the on-screen Alpha Blending test, which fetches high-resolution, uncompressed textures from memory. The margin is even greater when rendering off-screen.

All of the other low-level tests show the S5 with near identical performance to the other Snapdragon 801 phones.

The Galaxy S5 performs as expected in gaming and GPU benchmarks. While it falls near the bottom in our benchmark charts, the difference between it and the top performing Adreno 330-based phone is generally negligible and shouldn't be noticeable in real-world use.

Results: Display Measurements

In the past, Samsung equipped its devices with decent panels but didn't perform any screen calibration, resulting in poor image quality. Fortunately, market pressure forced a change in policy, and Samsung has been gradually improving the calibration of its screens. While the S5's display looks much better than the previous-generation S4, it still doesn't match the performance of some of its peers.

The Galaxy S5 uses a Samsung 5.1-inch 1080p SAMOLED screen paired with several different software modes that tweak the display settings for different use cases. The default mode is Standard, but there are also Cinema, Dynamic and Professional photo modes too. We include results for each of these settings in our benchmark charts when appropriate. However, we weren't able to examine the adaptive display mode, which "automatically optimizes the color range, saturation and sharpness of your display" when using the Gallery, Camera, Video, Smart Remote or Google Play Books apps.

Brightness

Brightness (also known as white level) measurements are taken by recording the luminance output of each device displaying a full white pattern, with the device's brightness slider set to both minimum and maximum values.

Unsurprisingly, the SAMOLED display in the Galaxy S5 turns in the lowest brightness value, although all the phones can dim their screens to an acceptable level for a dark room. At the other end of the scale, the S5 fails to break the 400-nit mark. This is a worst-case value however, based on displaying a full white image (100% APL). If auto-brightness is turned on, the S5 provides additional power to the display to boost its output under bright ambient light conditions. While the value can vary, we measured 452 nits with auto-brightness on and 100% APL. This boost makes the S5's screen much easier to view outdoors, but is detrimental to battery life.

For the Cinema and Professional photo modes, brightness drops slightly to 354 nits, while the Dynamic mode remains at 367 nits.

In order to make device comparison possible, the rest of our display measurements, along with our battery testing, are performed with the screen set to a standardized white level of 200 nits.

Black Level

Our black level measurement is the luminance output of a full black pattern after the luminance output of full white has been standardized to 200 nits. It's important to note that AMOLED displays will always measure a black level of zero, since their pixels simply turn off to render black.

The SAMOLED screen in the Galaxy S5 achieves a true black, since the organic LEDs are switched off and not emitting any light. Turned off, the screen has dark green-gray tint when viewed at an angle.

Contrast ratio, the difference between a full white pattern and a full black pattern, is essentially infinite, because of the S5's zero reading on the black level tests.

Gamma

Gamma compensates for the linear brightness levels displayed by a screen versus the nonlinear way our eyes perceive light. A gamma curve of 2.2 is what we optimally want to see, as a screen with a gamma less than 2.2 appears brighter and with less shadow detail, while a gamma larger than 2.2 displays heavy shadows with fewer highlights.

This is one test where all of the phones do well. The S5 just overshoots the ideal value by an amount not noticeable in everyday use. In the Cinema and Professional photo modes, gamma increases slightly to 2.30.

Color Temperature

Color temperature is a measurement in Kelvin, which is used to describe how “warm” or “cool” a given display is. Ideally, as long as you're not viewing your device in direct sunlight, this should be in the 6500 range. Higher color temperatures result in a bluish hue, while lower temperatures deliver a warm or reddish tone.

All of the smartphone displays we’ve seen lately tend to use cooler temperatures above 6500 Kelvin due to the color of the LED backlights being used. Definitely on the high end of the scale, the S5 in both the Standard and Dynamic mode exhibits a noticeable blue tint when viewing an all-white background. This effect is reduced but not eliminated when using the Cinema or Professional photo modes, where the tint looks a bit more green.

Color Gamut

Our volume measurements are compared against both the sRGB and AdobeRGB color gamuts. A reading of 100 percent on sRGB and 72 percent on AdobeRGB is ideal for viewing the vast majority of digital consumer content. A lower reading is typically accompanied by an overly red or yellow image, and a higher reading is usually too blue/green.

AMOLED panels are capable of reproducing a broader spectrum of color than LCD screens, which the S5 demonstrates. This isn't necessarily a good thing though. When viewing content targeted for the sRGB color space (which is most content) on a wide gamut display like the S5's, colors appear oversaturated, almost neon, in appearance. Some people prefer such vivid colors, which make images "pop", while others find the bright colors overpowering and unpleasant. When comparing the screens of the S5 and iPhone 6, which covers about 100% of the sRGB color space, the effect is especially pronounced; primary colors in particular look very unnatural on the S5. Using the Cinema mode helps produce more accurate colors, but a true sRGB mode would be preferred.

The SAMOLED display in the Galaxy S5 improves upon its predecessor, but still doesn't match the best IPS LCD screens. For starters, Samsung's PenTile sub-pixel matrix creates grainy or noisy-looking images in certain situations, like when turning the brightness down or viewing gray backgrounds. Grayscale gradients also exhibit a noticeable amount of banding, which isn't present when viewing the same image on the iPhone 6. Gray backgrounds also show a green tint, another side effect of the PenTile pixel arrangement, which becomes progressively worse as viewing angle increases.

Results: Battery And Throttling

Basemark OS II Full (Anti-Detection)

The Basemark OS II battery test scores are derived by repeatedly running the devices until enough data has been collected to determine the drain rate of the device.

In our first battery benchmark, the Galaxy S5 turns in an average performance, sitting between the One (M8) and Xperia Z3, both of which have similar size screens and the same HD resolution. Despite having about an 8%-larger battery than the HTC, Samsung's S5 scores 10% lower. However, it performs better than the 11%-larger battery in the Xperia Z3 by about the same amount. The iPhone 6 has a substantially smaller battery than the S5 (1800mAh versus 2800mAh), but manages to perform slightly better, its smaller screen, efficient SoC and software optimizations helping to reduce power drain.

GFXBench 3.0 Corporate

GFXBench's battery test measures battery life and performance stability by logging frame and battery discharge rate as the on-screen T-Rex test runs for 30 consecutive iterations. The results are given in two scores: estimated battery life in minutes and the number of frames rendered on the slowest test run (to gauge if a device is throttling).

This test stresses the GPU and serves as an indicator for battery life when playing games. With its chart-topping performance, one might conclude that the S5 is a clear winner. However, this chart doesn't tell the whole story. We also need to look at the average frames per second during each run to see how much work the phones are actually doing. Noting that the two all-metal phones have the worst battery life here foretells what we'll see in the next chart.

Average performance for the S5 over thirty iterations of T-Rex drops to 18 FPS compared to 27 FPS for a single iteration, indicating thermal throttling. The three metal phones—iPhone 6, One (M8) and OnePlus One (internal magnesium chassis)—are able to dissipate heat more effectively, which means they continue to operate at peak frequency and consume more power.

The diagram below shows how the S5’s performance varies over time.

While not as severe as the thermal throttling we saw with the LG G3, the S5 still can't keep its GPU running at 100% for long before it has to throttle clock rate.

Battery life for the Galaxy S5 is merely average. Unfortunately, the upcoming update to Android 5.0 will likely reduce battery life further, since AMOLED screens are most efficient with dark interfaces (effectively no power draw when displaying black) and Lollipop uses all-white backgrounds.

Did A New Star Form In Galaxy S5?

The S5 is a brown dwarf, failing to achieve star status in Samsung's Galaxy. Sure, it's a definite improvement over the Galaxy S4, and a decent smartphone, but it just does not stand out amongst the brighter stars in its celestial neighborhood.

Packed inside the S5 is an impressive list of hardware: Snapdragon 801 SoC, 2GB RAM, 16MP camera, microSD support, removable battery, all topped off with a 5.1-inch 1080p SAMOLED screen. It even includes support for 802.11ac Wi-Fi, Bluetooth 4, NFC and Cat. 4 LTE; all of the important wireless communication standards in use today. The only issue is that this is basically the same list of features the S5's competitors have. Samsung did add a fingerprint scanner, which, after some software tweaks, works pretty well, falling just shy of Apple's Touch ID in accuracy and convenience. The only other semi-unique hardware feature is IP67 dust and water resistance, a feature shared by Sony's Xperia Z2 and now Z3. The Xperia line offers much of the same hardware, but in a more attractive package with better cameras and audio, although the batteries aren't removable.

Camera performance is reasonably good. The S5's strengths are the easy-to-use UI, excellent HDR mode, high level of detail when shooting in good light, excellent color reproduction and super-fast phase detection autofocus. We recommend keeping HDR mode on at all times because the S5 doesn't handle scenes with much dynamic range well with it turned off. Also, Samsung's ISOCELL sensor, despite the marketing claim of improved low-light performance, really struggles in poor lighting conditions. To compensate, Samsung uses overly aggressive noise reduction algorithms that destroy detail, blend elements together and produce "soft"-looking images. There's still no substitute for pixel size when shooting in low light.

Performance is comparable to other flagship phones, but is not really faster than any of them. The S5's internal NAND storage performance is also rather low, although this was really only noticeable during game loading screens. Also, like all of the other all-plastic phones, the S5 experiences thermal throttling when stressing the GPU, limiting its ability to play the most demanding Android games.

Samsung did a good job cleaning out the excess features and fluff from its latest TouchWiz ROM. Without the extra clutter, the interface is easier to navigate and more responsive than previous versions. TouchWiz also gives you complete control over the organization of the app drawer and a customizable notification shade. While a 5.1-inch screen isn't big by today's standards, it's still impossible for most people to use with only one hand. Fortunately, the S5 includes a one-handed mode that brings the entirety of the screen within reach, including the navigation and volume buttons. Samsung makes good use of the proximity sensor on the front with its "Air browse" feature, allowing you to scroll through lists with a wave of your hand. Too bad it only works in a handful of apps. One thing that TouchWiz lacks is granular control over app permissions.

The design of the Galaxy S5, which isn't much different than the S4, already looked dated the day it launched. Thicker than average bezels, a screen with a gray-green tint when turned off and a chrome ribbed band around its perimeter combine to create the least-attractive flagship phone currently available. The soft-touch plastic on the back at least feels nice and provides ample grip.

Expectations for the S5 were astronomical, but it quickly fell back to Earth after launch day. Samsung's reputation for using cutting-edge hardware and packing in features finally exceeded what it was able to deliver, which is a decent phone that fails to distinguish itself amongst stiff competition. If you specifically need a smartphone with a removable battery, fingerprint scanner and IP67 protection, then the S5 is for you. If you can forgo at least one of those features, there are more compelling phones to consider.

     >> See the Reader Ratings and Advice for the Galaxy S5

HP's future has been slightly murky, as the company has faced serious competition and recently announced that it's splitting itself into two companies, one that would focus on printers and PCs and one that would be dedicated to the enterprise. Many have wondered about the fate of those respective parts. The enterprise segment seemed strong, but would HP's consumer side wither and die? Would it be sold off to a competitor such as Lenovo?

It looks like neither is happening any time soon. Today, HP revealed two new intriguing technologies that are designed to rock the market: one is a 3D printer, and the other is a unique approach to personal computing and creation, but together HP is calling them a "blended reality ecosystem."

sprout

On the consumer side, the most compelling of the two by far is a curious system called sprout. (Yep, lower-case "s" to increase its adorableness.)

Although fundamentally sprout is an AIO with a 3D camera strapped to it, the system is built to enhance users' ability to create, particularly in 3D. There is an optical mouse and a keyboard (a chiclet-style item with volume control and Windows 8 hot keys), but sprout relies heavily on a special 3M-designed mat called the HP Touch Mat for input. The 20-inch mat supports touch input, but you can also use the included stylus, the Adonit Jot Pro.

One of the main components of sprout that makes it unique is the 3D camera, the HP Illuminator. It's powered by an HP DLP Projector to give users that second "screen" on the Touch Mat, and it includes a 14.6 MP camera and an LED desk lamp to further illuminate your work. HP also sneaked in an Intel RealSense 3D Camera to allow for 3D scanning, which is a key element of the system.

The 23-inch (Full HD) AIO (which has a Texas Instruments-made display) itself is powered by a 4th-gen Intel Core i7-4790S CPU, as well as Nvidia GeForce GT 745A and Intel HD Graphics 4600. It has 8 GB of DDR3-1600 SDRAM (2 x 4 GB) and can be expanded to 16 GB. For storage, the system (model number 23-s010) uses a combination of a 1 TB SATA 6 Gbps SSHD (with 8 GB of flash acceleration cache).

There's a PCI-e x1 mini-slot for expandability, and other ports include a 3-in-1 card reader (SD/SDHC/SDXC), two USB 2.0 ports, four USB 3.0 ports, HDMI out, gigabit Ethernet, and separate audio line out and headphone/mic jacks. For video chat, there's a 1 MP HP webcam.

The machine also offers integrated Bluetooth 4.0 and dual-band 802.11a/b/g/n with MIMO technology. There are "premium" stereo speakers with DTS Sound and dual digital MEMS microphones.

There's plenty of software on the sprout. Windows 8.1 is the OS of choice, and thus the system will come with everything you'd expect to be bundled with the OS, but the company also included HP Workspace to help users get the hang of creating with the system. There's also CyberLink PowerDirector for making movies, Fresh Paint, HP Connected Photo for accessing images across all your devices, HP Connected Music, Microsoft's Project Spark, and more.

You don't have long to wait to see sprout; the system will be on sale on November 9 at several outlets including Best Buy, B&H, HSN (really?), the Microsoft Store, and HP's own site. The starting price is $1,899, which seems to place it at the level of a professional-class system. It's hard to imagine that the average household will be willing to drop that kind of scratch on a system that, while functioning perfectly well as a family PC, is loaded with (powerful) gimmicks that most families won't often use.

HP Multi Jet Fusion

The 3D printing market is growing rapidly, and HP does not want to be left out; in fact, HP plans to dominate. Not to put too fine a point on it, but HP said that its new 3D technology is "A tool to trigger the next industrial revolution." However, if everything HP said today at a launch event is true, the company may be on to something.

HP announced a 3D printer that's designed to solve the problems of print speed, printed quality, and cost. The printer uses a new technology called "multi jet fusion," which HP said will be able to offer faster print speeds than any other device on the market by a factor of 10.

HP also said that it can also offer higher precision and greater strength than its competitors and do so with a device that is inexpensive to purchase and costs less to own. Those competitors are not so much the consumer-grade MakerBots of the world, but more like Stratasys; which is to say, this technology is aimed primarily at commercial applications. Of course, these sorts of innovations tend to trickle down, so anything exciting you see happening with multi jet fusion on the commercial side will surely end up in your home office in due time.

HP played a short animation that showed how the process works. There's a high-resolution, page-wide "print bar" that lays down multiple types of liquid material, and then a second arm slides over and selectively "applies energy" (which is such an engineer-like way of describing it) to fuse that material. That selectiveness allows the printer to adjust elasticity within a part or change its texture. HP also said that its new printer will offer a wide array of colors for powerful multi-color prints.

HP is supremely confident in the strength of its printed products. To that end, the company showed a video of a large printed chain link supporting the weight of a car being hoisted by a crane. The car actually belongs to one of the engineers who worked on the product, and after the demo the car was returned to him unharmed. (HP didn't mention whether or not there were previous failed tests with the vehicles of unluckier HP employees.)

Like some other 3D printers in the market, HP's multi jet machine can print some of its own parts, and thus it's a (semi) self-replicating device.

This multi jet fusion will be an open platform. HP wants partners to come in and innovate on its IP to push the technology further; exactly how "open" it will actually be remains to be seen. The company did not specifically say, for example, that it would be open source. (We doubt that would be the case.)

It's taken HP some 20 years of R&D to get to this point where it has combined its 2D printing IP (such as HP Thermal Inkjet) and developed it to create this 3D printing technology, and the fruit of that work is nigh. Partner companies will work with HP for the rest of this year and into 2015, when they'll be a seeded beta version of the printer. In 2016, the printer will go into full production.

Together, sprout and HP's Multi Jet Fusion technology are designed to usher in a new era of "blended reality," and HP appears confident that it's the vanguard.

Follow Seth Colaner @SethColaner. Follow us @tomshardware, on Facebook and on Google+.

On Thursday, Brother International Corporation expanded its Business Smart Series of printers with the launch of the MFC-J4420DW. Designed with small offices and home offices (SOHO) in mind, it's a color inkjet all-in-one with faster print speeds, mobile device connectivity, and a 2.7-inch color touchscreen display.

"By committing to low-cost printing, mobile device connectivity, and versatile paper handling, we're making owning an innovative, hard-working Brother Inkjet All-in-One accessible to SOHO customers who are keeping a careful eye on their budget," said Eric Dahl, Brother Director of Marketing for SOHO products.

According to the specs, the new all-in-one printer is the fastest in its class, with up to 20 ppm black and 18 ppm color. The printer is also compatible with Super High-yield replacement ink cartridges, which delivers approximately 1,200 pages black and 1,200 pages color. That means a lower cost per page, which is ideal in higher volume SOHO printing environments.

Thanks to the printer's Wireless N connectivity, users can print and scan from mobile devices including Apple's iOS, Android, Windows Phone and Amazon's Fire OS. This is accomplished by connecting to the printer via AirPrint, Google Cloud Print, the free Brother iPrint&Scan app, Wi-Fi Direct, or Cortado Workplace. The printer can also scan to Dropbox, Google Drive, Box, OneDrive, Onenote and more.

Brother provides a list of features that includes the ability to print on 11" x 17" paper by using the printer's single-sheet bypass tray. The printer also has a 150-sheet paper tray capacity for paper up to 8.5" x 11," and a 20-sheet Automatic Document Feeder for unattended multi-page scanning, copying and faxing.

In addition to those features, the printer provides automatic duplex (two-sided) printing. The printer also includes Landscape Print Technology, which stores and feeds paper in a landscape orientation for a "space-saving, desk-friendly" design.

The MFC-J4420DW is available now for $169.99 USD.

On Thursday, the company also introduced the $179.99 MFC-J4620DW. This model builds upon the MFC-J4420DW by adding NFC “touch to print” capability and a larger 3.7-inch Color TouchScreen Display. The company also introduced the $149.99 MFC-J4320DW, which has the same features as the MFC-J4420DW, but doesn’t provide an automatic document feeder.

Follow Kevin Parrish @exfileme. Follow us @tomshardware, on Facebook and on Google+.

Amazon has quietly announced a new gadget that should make it easier on AmazonFresh shoppers to purchase goods. Called Amazon Dash, it's essentially a barcode scanner that is 6.375 inches long, 1.125 inches wide and features wireless connectivity so that customers can add to their shopping list on the fly.

"Every member of the family can use Dash to easily add items to your AmazonFresh shopping list," reads the product page. "Keep it on your kitchen counter or hang it on the refrigerator. Did your kids just eat the last of the cereal? Conveniently refill and restock your home's everyday essentials, and have fun doing it."

Amazon's Dash has two input modes: microphone and scanner. The microphone allows users to simply say what they want to order by pressing the microphone button and speaking into the built-in mic. "It's a voice search that actually works," reads the product page, which is what Amazon said about the just launched Fire TV set-top-box. Could they use the same voice tech?

To scan a barcode, users merely hit the scan button and point it at the product box. Users can scan in a week's worth of groceries in mere minutes. According to the company, Dash recognizes millions of items available on AmazonFresh and Amazon.com.

"Dash connects to your home Wi-Fi network and works directly with your AmazonFresh account," reads the product page. "Say or scan items into your Dash, and then view the list on your desktop or mobile device to purchase and schedule delivery."

Unfortunately, Amazon Dash can only be acquired by invitation. Even more, AmazonFresh is currently only available in Southern California, the San Francisco area, and the Seattle area. Still, with the launch of both Fire TV and Amazon Dash, it's clear that the company wants customers to stay at home to get their blockbuster movies and popcorn from one online source.

AmazonFresh provides same-day and early morning delivery of a huge selection of items, including fresh grocery and local products. This service provides doorstep delivery and attended delivery, the latter of which means the customer schedules a 1-hour appointment to personally accept the delivery. For Doorstep Delivery, customers choose a three-hour window, but they are not required to be at home.

"We're branching out as fast as we can while being careful not to sacrifice the quality and convenience our customers expect from AmazonFresh," reads the AmazonFresh FAQ.

Finger print biometric recognition is the next big trend in smartphones. Apple popularized it with the iPhone 5S, and now Samsung is hoping to make it a hit with theGalaxy S5, announced on Monday at MWC.

We take a hands-on look at how the scanner works. First of all, it's different from Apple's TouchID in that the user needs to slide his or her finger over the elongated button, rather than just holding a finger over Apple's circular home button.

Samsung demonstrates this feature using the index finger, but we found that it works with the thumb, which aided in usability when holding the phone in the most natural way.

The software supports three finger prints, so you can scan both your left and right hands for those times when you're left shorthanded.  This also grants you the option of giving one of those fingerprint scans to a trusted individual without the need to share any complex lock patterns or passwords.

Once set up, the fingerprint software will allow the user to unlock the device, open private mode, and verify payments.

Check out all our Mobile World Congress 2014 coverage here!

Follow Marcus Yam @MarcusYam. Follow us @tomshardware, on Facebook and on Google+.

We've been talking about the HTC One Max for so long, you'd be forgiven for thinking the device was already announced. However, until today, HTC had yet to make anything official.
  

Today, the company finally unveiled the HTC One Max for real. The device packs a huge 5.9-inch 1920 x 1080 super LCD 3 display, making it more of a phablet than a phone, and the same aluminum chassis that we've seen in HTC's other 'One' handsets.

Under the hood, you're looking at the same Snapdragon 600 CPU we saw in the HTC One, 2 GB of RAM, 32 GB of storage, a large 3,300 mAh battery, a 4-megapixel Ultrapixel camera (same as the HTC One), Android 4.3, and HTC's own Sense 5.5. It should be noted that while the back cover is removable (to allow for storage expansion), the battery itself doesn't come out so that 3,300 mAh battery is not designed to be user replaceable. Most importantly, HTC is following in Apple's footsteps with the addition of a fingerprint scanner for unlocking the device to your favorite application.

The phone is set for availability 'this holiday season' from Verizon and Sprint but HTC wasn't any more specific than that when it came to a release date nor did the company provide any indication as to pricing for the One Max, so we'll have to wait for information from the carriers. Our friends at LAPTOP have already gone hands on with the HTC One Max. Check out the video below and be sure to click through for first impressions.

Follow Jane McEntegart @JaneMcEntegart. Follow us @tomshardware, on Facebook and on Google+.

Apple started with the iPhone 5C (which you can read about here), but, obviously, the main event is the iPhone 5S. The iPhone 5S is available in black, gold and silver and made of high-grade aluminum with cut chamfered edges.

As rumors indicated, the iPhone 5S will run on Apple's A7 SoC. Apple today confirmed that the A7 is actually a 64-bit chip with 2x general purpose registers, 2x floating point registers, over one billion transistors and 102 square mm die size. The result is CPU performance that is 40 times that of the original iPhone and graphics performance that is 56 times that of the original iPhone. You can read all about the A7 here. In addition to the A7, the iPhone 5S also uses a motion co-processor called the M7. The M7 is in charge of motion data, measuring accelerometer, gyroscope and compass data continuously (Apple is eyeing health and fitness apps with this).

The camera has been upgraded, too (again, not a surprise). The iPhone 5S camera packs a 5-element lens with an aperture of f/2.2 on top of a sensor with a surface area that's 15 percent larger than that of the iPhone 5. It also has burst mode, image stabilization, and slow motion (720p at 120 fps).

Apple also confirmed Touch ID, which is fingerprint authentication that can be used to unlock your device and make iTunes purchases. The sensor measures 170 microns thick and boasts a 500 ppi resolution. You don't have to press the home button to scan, just place your finger on top, and the phone can read multiple fingerprints. Additionally, your fingerprints are never stored on Apple's servers or backed up to iCloud.

As far as battery life is concerned, Apple is promising 250 hours of standby, 10 hours of LTE browsing, 40 hours of music playback, 10 hours of talk time on 3G, though the company didn't talk about the battery's size. We're also waiting for more information on the display, RAM, and other aspects of the iPhone 5S hardware.

The iPhone 5S is priced at $199 for the 16 GB model, $299 for the 32 GB model, and $399 for the 64 GB model. iPhone 5S and 5C will be available starting September 20. The 5C will go up for pre-order on September 13 (this coming Friday), and China and Japan are also on the list of launch-day markets. This is a first for China, and Japan's NTT DoCoMo, which is getting the iPhone for the first time. That explains why Apple was streaming the event in Beijing and Tokyo (but doesn't really explain why there was no stream for anyone else).

Check out the hands-on from our sister site LAPTOP:

• iPhone 5S Packs Apple A7 Chip: World's First 64-bit Phone
• Apple Confirms iPhone 5C with 4-inch Retina Display and A6 SoC
• Apple Confirms iOS 7 to Launch September 18

Follow Jane McEntegart @JaneMcEntegart. Follow us @tomshardware, on Facebook and on Google+.

ROM Manager

We’re kicking off this round-up with what might be the most important app there is for your Android-based device. As its name suggests, ROM Manager lets you save and restore your current ROM. But this app also allows you to restart in Recovery mode, install a ROM from the SD card, and partition the card. It's simply a must for anyone who enjoys playing around under the hood of Google's operating system.

What we like: Easy ROM switching, and the fact that you can restore a device’s original ROM.

Titanium Backup

When it comes to modifying certain system settings as root, you risk crashing your smartphone and losing all of your data, forcing you to reinstall Android. Fortunately, Titanium Backup lets you backup and restore all of the information on your device: apps, settings, SD card contents, and so on. This app might just prevent you from making a costly mistake.

What we like: Knowing that our data is safe, even from us!

JS Backup

JS Backup is a clear, simple, and intuitive data storage tool that allows you to save and restore contacts, call histories, text messages, calendars, bookmarks, system settings, shortcuts, alarms, dictionaries, apps, pictures, songs, playlists, and videos. In other words, it lets you save pretty much everything.

What we like: Uses the widely-accepted CSV format.

Android Terminal Emulator

A graphical interface is nice, but everyone knows that nothing beats a good old command-line interface (CLI) when you want to get into the guts of an operating system. Android Terminal Emulator lets you to do just that by providing access to Android’s CLI, where you can type in a host of commands, each one more exciting than the last.

What we like: Full access to Android’s underlying Linux command line interface.

ConnectBot

ConnectBot is an SSH client capable of managing several SSH sessions simultaneously, creating secure tunnels, and copy/pasting to and from other apps.

What we like: The possibility of accessing our home PCs while travelling.

Go Launcher EX

Go Launcher EX allows you to change the Android interface completely. It updates the widgets and icons, and makes folders behave just like those on the iPhone. The launcher, gestures, animations, and screen management receive an overhaul as well, and almost everything is customizable, letting you create a truly unique and personal user interface.

What we like: The ability to customize a device's UI beyond what Android typically facilitates.

System Tuner

System Tuner is one of those do-everything apps. Although it's a task manager at heart, this app also allows you to select what programs start with the phone, set the cache size for SD cards, tweak the CPU frequency, and monitor memory usage. It’s also possible to modify apps that are integrated into the ROM, moving them between the “user” and “system” space, though these functions are only available on a rooted device.

What we like: Being able to customize both a device’s hardware settings and OS internals with a single app.

Spare Parts Plus!

Spare Parts Plus! is a free app that makes it possible to modify a handful of hidden Android parameters. For example, you can enable or disable the “compatible” display mode, adjust haptic feedback, transitions, and animations. This app can also display a history of battery usage on certain devices.

What we like: The multitude of animation controls.

BusyBox

BusyBox enables many standard Unix commands (like dpkg, ls, mount, su, and so on) on a rooted Android device. Essentially, BusyBox is a collection of useful command-line tools conveniently packaged into a single app.

What we like: Being able to use classic Unix command-line tools in Android.

Superuser

Superuser provides access to the protected areas of your rooted Android device, allowing you to choose the apps allowed superuser (also known as “su”) access to the system.

What we like: Being able to customize the exact privileges for any app.

CatLog

CatLog provides a GUI front-end for Android’s logcat command on rooted devices. This app provides an easy-to-read, color-coded version of the Android system log that is searchable and sortable. You can even save system logs to an SD card and send them as email attachments.

What we like: The color-coded column view and system log export.

Battery Spy

Battery Spy provides a visual representation of several key battery usage statistics. It can check the current state and charge level of the power source and track consumption over several days. Notifications can also be set when the battery level or temperature reaches user-defined thresholds. Battery Spy supports profiles, so you can monitor your primary and backup batteries.

What we like: Customizable 2D or 3D graphs representing usage statistics over time.

Cool Tool

Cool Tool allows you to monitor numerous system parameters right from Android’s Action Bar. Some of the subsystems that Cool Tool monitors include CPU load/frequency, memory usage, Wi-Fi/cellular network traffic, SD card capacity, battery level, cellular signal, temperature, and more.

What we like: An always-on, user-configurable system monitor.

Total Commander

Total Commander is the Android version of the famous desktop file manager by the same name. This app allows you to create/rename/copy/move directories, delete files, compress/decompress, search files, send files via Bluetooth, and basically every other primary function of a file manager. LAN, FTP, and WebDAV plugins for Total Commander are also available on Google Play.

What we like: The comprehensive file management options, and robust plug-in system for adding even more functionality.

File Manager HD

File Manager HD is, as its name suggests, a file manager for Android tablets. Like all file managers, File Manager HD can cut, copy, paste, compress, decompress, and search for files and directories. But it also has a built-in text editor. File Manager HD supports networked file systems over LAN or SMB, as well as several major cloud storage solutions, including Box, DropBox, Google Drive, SkyDrive, and SugarSync.

What we like: The folders, sidebar, and navigation bar work well on tablets, and will be immediately familiar to any desktop user.

eFile

eFile is yet another file manager option for Android. On top of the usual capabilities enabled by most file managers, eFile allows file sharing via Wi-Fi and supports drag and drop (via press and hold). The top and bottom toolbars with pop-out make good use of the limited screen real estate available on smartphones.

What we like: Thumbnail view for image files, and drag and drop support.

Astro File Manager / Browser

Astro File Manager / Browser is one of the most popular file manager apps for Android. It allows you to browse or search for files on your Android device, local network, Box, DropBox, Google Drive, Facebook, or SkyDrive accounts. A directory path in the uppermost toolbar makes backward navigation a snap.

What we like: Integration with most major cloud storage services.

3G Watchdog

3G Watchdog measures your mobile data consumption over GPRS, EDGE, 3G, and 4G networks. You can also get a detailed daily, weekly, or monthly report on your usage.

What we like: The notification icon that changes color and historical usage reports.

Call Meter 3G

Call Meter 3G gathers statistics on calls, text messages, and data consumption. This app also allows you to set limits for each type of communication service.

What we like: The simple usage breakdown and alerts.

Quadrant Standard Edition

Quadrant Standard Edition is a simple system (CPU, GPU, and I/O) performance benchmark.

What we like: Easy to compare devices...and brag to your friends about your latest purchase!

Uninstaller