Lenovo’s ThinkPad X1 Carbon Gen 14 (starts at $2,032; $2,217.30 as tested) enters the laptop market with the poise of a veteran. This series has spent years at the top of the ultralight business stack and shows no sign of slowing down. Light and refined as ever with Intel’s “Panther Lake” silicon and equipped with a class-leading keyboard, OLED display, and enterprise-ready tech, it’s every bit the flagship it’s intended to be. That said, consumer models like Dell’s XPS 14 have also raised the bar, so whether the ThinkPad X1 Carbon’s refinements keep it ahead of the pack remains to be seen.

Design of the Lenovo ThinkPad X1 Carbon Gen 14

The ThinkPad X1 Carbon Gen 14 may have received a refresh, but its design principles remain unmistakable. Squared edges, matte black surfaces, and functional lines keep it faithful to the lineage that IBM established long ago. The webcam notch is the quickest visual cue that this is a current-generation model. Lenovo’s only real design flourish is the colorful X1 badge on the lid for this flagship; lesser ThinkPads simply have the brand name.

At just 2.15 pounds, the ThinkPad X1 Carbon feels unrealistically light, to the point that I wondered if the unit I received was hollow. Rest assured, there’s a real computer inside, and the build quality is excellent. The outsides are made mostly of lightweight magnesium, with precise fitment and a premium feel. True to the product name, carbon fiber is used, mainly in the lid cover and its internal frame. Rigidity is just average, though; the palm rest and surrounding deck feel solid, but the chassis flexes more than I expected. The lid, however, is impressively stiff, and the hinges are precisely tuned for one-handed opening.

Measuring 12.3 x 8.49 x 0.6 inches, the ThinkPad X1 Carbon has a similar footprint but is much lighter than the 3.4-pound Apple MacBook Pro 14-inch and the three-pound Dell XPS 14.

Port selection is decent for an ultraportable, with HDMI 2.1, two Thunderbolt 4 (USB-C), and a headphone jack on the left and another Thunderbolt 4 and a USB-A port on the right. Intel wireless networking provides Wi-Fi 7 and Bluetooth 5.4. Cellular WAN is available.

Lenovo ThinkPad X1 Carbon Gen 14 Specifications

Productivity Performance on the Lenovo ThinkPad X1 Carbon Gen 14

We tested the ThinkPad X1 with a Core Ultra 7 355 processor, 32GB of RAM, and a 512GB PCIe 4.0 SSD. This “Panther Lake” processor features four Performance and four Low Power Efficiency (LPE).

For performance comparison, we included the Apple MacBook Pro (14-inch, 10-core M5; $2,349 as tested), Asus Zenbook Duo (Core Ultra X9 388H; $2,299.99), and the Dell XPS 14 (Core Ultra 7 355; $1,699.99). These are all premium systems that you might consider if you’re cross-shopping consumer models. The Apple and Asus use significantly stronger processors, though the ThinkPad X1 Carbon and the XPS 14 can be equipped with the Core Ultra X-series to compete in that tier if you're willing to spend even more.

On the cross-platform Geekbench 6 benchmark, the ThinkPad X1 Carbon’s single-core score of 2,635 points matched the XPS 14, an expected result considering they both use the Core Ultra 7 355, but the ThinkPad trounced it in multi-core, with 11,107 points to Dell’s 7,964. Neither machine proved a match for Apple’s M5, particularly in single-core. With both Apple and Asus scoring over 17,000 points in the multi-core portion, their processors are clearly in a different league.

The ThinkPad X1 Carbon finished our 4K-to-1080p video transcoding test in 6 minutes and 12 seconds, more than 30 seconds quicker than the XPS 14 but not competitive with the Asus (4:22) and MacBook Pro (3:31).

Last, the ThinkPad X1 Carbon completed our 25GB file transfer test with an average transfer rate of 1,256.31 MBps, well behind the Asus’ 1,531.3 MBps and not anywhere close to the MacBook Pro’s leading 1,917.35 MBps.

To stress test productivity laptops, we run 10 loops of Cinebench 2026. The ThinkPad X1 Carbon scored 2,581 points in the first loop and stayed within the 2,500-to-2,600-point range for the remaining loops, demonstrating stable performance. The P-cores ran at an average clock of 3.73 GHz while the LPE-cores ran at 3.46 GHz.

Graphics on the Lenovo ThinkPad X1 Carbon Gen 14

We run 3DMark Steel Nomad to gauge raw graphics horsepower. Scoring 596 points, the Intel Graphics inside the ThinkPad X1 Carbon are adequate for basic use but little else. The Arc B390 integrated chip in the Asus’ Core Ultra X9 proved far more robust, leading the group with 1,483 points, followed by the MacBook Pro with 1,122 points.

Display on the Lenovo ThinkPad X1 Carbon Gen 14

Our ThinkPad X1 Carbon uses the 2880 x 1800 OLED panel, an upgrade over the base 1920 x 1200 IPS screen. It produces memorable visuals no matter what you’re viewing. In Dune: Part One, I noted the brilliant white sands of the desert and the inky depths of space. Disney’s Zootopia showed its deep, vibrant color that kept me watching longer than I intended. The anti-glare surface does a commendable job of reducing glare. This panel doesn’t support touch, but you can option touch versions of both the 1920 x 1200 IPS and OLED panels.

The ThinkPad X1 Carbon’s numbers leave little doubt that it has an excellent display. Its color coverage – 82% of DCI-P3 and 115% of sRGB – were within a percentage point of the MacBook’s, just shy of Asus’ 87% and 123%, and ahead of the Dell’s 71% and 101%. Its 473-nit peak brightness is impressive considering it uses an OLED panel, which are often close to or just shy of 400 nits.

Keyboard and Touchpad on the Lenovo ThinkPad X1 Carbon Gen 14

The ThinkPad X1 Carbon maintains Lenovo's standard for excellent input devices. The keyboard feels more cushioned than my older ThinkPad X13 yet still provides plenty of key travel for engaging feedback – my fingers intuitively knew when I reached the top and bottom of the keystroke. Keystrokes sound slightly rubbery, with less clack than my older ThinkPad, but still pleasant. The keyboard deck is rock solid and doesn't flex. Highly visible white backlighting makes typing in the dark a breeze.

The layout is highly intuitive, with an inverted-T arrow cluster, dedicated Home/Page Up/Page Down/End/Delete keys, and a programmable F12 shortcut key. The power button at top right is also a fingerprint reader. Classic ThinkPad users will notice Lenovo switched to the mainstream left Ctrl and Fn key arrangement, but they can be swapped in the Vantage app.

The ThinkPad X1 Carbon touchpad that we tested is mechanical, but a haptic version is available as a no-cost option on customized models. The pad's top-hinged design works reliably and delivers a firm, satisfying click. The three buttons above the pad couple with the classic TrackPoint pointing solution in the middle of the keyboard. ThinkPads are the only laptops that have them anymore.

Audio on the Lenovo ThinkPad X1 Carbon Gen 14

I listened to my test audio tracks multiple times on the ThinkPad X1 Carbon since the speakers sounded that good – who would have thought a business laptop could so easily double for entertainment?

In William Black & Luka's "Bleed 4 U", I heard crisp vocals, well-separated instruments, and detailed, impactful bass -- this setup didn't skip a beat or distort in any way when the drops hit.

The Atmos Detailed preset in the Dolby Access app (its default setting) evened out the soundstage for a clearer tone. I switched to the Detailed preset to sharpen the drum hits and vocals in a vinyl rip of "Shattered Dreams" by Johnny Hates Jazz, which also didn't show any distortion even at maximum volume. The realistic volume limit is around 80%, after which the setup doesn't get audibly louder. This setup is easily loud enough for personal listening.

Upgradeability of the Lenovo ThinkPad X1 Carbon Gen 14

User-upgradeable components in the ThinkPad X1 Carbon are limited to its M.2 2280 SSD and battery pack; models with WAN can also have that module upgraded. Access through the bottom panel is straightforward – four captive Philips-head screws loosen it, then the clips need to be popped along the back edge, which were snugger than I thought. Lenovo maintains an excellent hardware maintenance manual and even provides a video walkthrough on YouTube.

Lenovo advertised a 9/10 iFixIt score, citing that the keyboard, daughter I/O board, and USB-C ports are also customer replaceable.

Battery Life on the Lenovo ThinkPad X1 Carbon Gen 14

The ThinkPad X1 Carbon lasted 13 minutes and 10 seconds in our battery test, which consists of web browsing, running OpenGL tests, and streaming videos with the screen at 150 nits while connected to Wi-Fi.

Though not a leading time – the Zenbook Duo lasted 14:23 with one screen active while the MacBook Pro went for 18:14 – it’s still an impressive showing considering it uses an OLED panel. Dell’s XPS 14 lasted an astounding 20:41, but that’s with the IPS screen; equipped with an OLED screen and more powerful processor, it lasted only 12:23 (not shown in the charts), and that’s with a 70 Wh battery whereas the ThinkPad X1 Carbon’s is rated for just 58 Wh.

Heat on the Lenovo ThinkPad X1 Carbon Gen 14

During our 10-loop Cinebench 2026 stress test, the ThinkPad X1 Carbon's surface temperatures peaked at 72 degrees Fahrenheit on the touchpad, 94 degrees between the G and H keys, and 106 degrees on the underside. Internally, the Core Ultra 7 355 averaged 91 degrees Celsius across all cores.

Webcam on the Lenovo ThinkPad X1 Carbon Gen 14

This ThinkPad's 1440p webcam produces an impressively sharp and well-lit picture. I could make out the wood grain on a door about 10 feet away and even the individual hairs on my cat who was looking over my shoulder. Bright lights, like windows and lamps, were properly exposed, and didn't cause my person to be underexposed. The webcam includes an IR sensor for facial recognition and the security of a sliding privacy shutter.

Software and Warranty on the Lenovo ThinkPad X1 Carbon Gen 14

Lenovo’s main app is Commercial Vantage, a hub for support access, system updates, and user settings. Those settings include battery smart charging, OLED power settings (taskbar, background, and display dimming), Fn/Ctrl key swap, F12 user shortcut key configuration, and keyboard backlighting brightness. Some settings have moved out of this app into Windows Settings, such as presence detection and display color temperature.

Vantage also provides access to the Aura Edition’s special Smart Modes: Working, Gaming, Creating, Entertainment, Meeting, and Learning. The system automatically switches between the modes depending on the user’s activity. These modes are a combination of settings, some of which are or aren’t enabled for a given mode: auto-prompt VPN (launches a configured VPN client), attention timer (a time limit where notifications are silenced, with a default of 30 minutes), distracting domains (automatically blocks popular social media sites), Wi-Fi fencing (only connects to wireless networks that you designate “safe”), Dolby Audio (sets the mode to Voice, Movie, Music, and so on), and microphone noise canceling mode and listening range. If you move between different environments, like an office to a crowded café, you could tweak the settings and get some usefulness out of this, though I wouldn’t consider it an essential selling point.

As for other apps, Lenovo Smart Share lets you share content from your phone, such as photos, and assists connecting Bluetooth devices.

The ThinkPad X1 Carbon comes with a one-year minimum warranty with mail-in service, though preconfigured models often include three years with onsite coverage. Coverage and service plans are available up to five years with accidental damage protection.

Lenovo ThinkPad X1 Carbon Gen 14 Configurations

We tested a midrange ThinkPad X1 Carbon Gen 14 Aura Edition with a Core Ultra 7 355 processor, Intel Graphics, 32GB of memory, a 512GB SSD, and a non-touch OLED display, retailing for $2,217.30.

The starting model is $2,032 at this writing, though Lenovo’s frequent sales can change the price. CPU options begin with the Core Ultra 5 335 and top out with the vPro-enabled Core Ultra 7 365. Models with the Core Ultra X7 are also supposed to be available but I didn’t see them on Lenovo’s site at the time of writing. Memory is available in 16GB, 32GB, or 64GB configurations depending on the processor. The base 1920 x 1200 IPS display can be upgraded to a 2880 x 1800 OLED panel, with either panel available with touch, though only OLED offers a 120 Hz refresh rate. A privacy screen option only comes in 1920 x 1200 display.

Bottom Line

Lenovo’s ThinkPad X1 Carbon Gen 14 delivers everything that a flagship ultraportable should. Its featherweight chassis, best-in-class keyboard, vibrant OLED display, and competent everyday performance make it an ideal fit for professionals or prosumers who take their work with them. It even doubles as a surprisingly capable entertainment machine thanks to its impressive speakers.

While consumer rivals like Dell’s XPS 14 offer similar performance, the ThinkPad X1 Carbon remains the productivity go-to because of its superior keyboard, broader port selection, and enterprise-ready options like vPro and WAN availability. For those willing to accept a thicker and heavier machine and use macOS, Apple’s 14-inch MacBook Pro remains the performance benchmark in this category.

Overall, the ThinkPad X1 Carbon Gen 14 remains an impeccable blend of portability, polish, and practicality, a mix that earns it a place among our recommendations.

I don't think I need a keyboard with a built-in touchscreen, but I also didn't think I needed a keyboard with a built-in Stream Deck (and I was, apparently, wrong). Turtle Beach's take on the customizable-macro-pad-on-your-keyboard is its recently-launched Command Series, which includes a handful of ultra-customizable peripherals with fancy touchscreen inputs — two keyboards, a modular numberpad, and three mice.

The flagship product is the Turtle Beach Command Series KB7 TKL: a low-profile, wired keyboard with magnetic Hall Effect switches and a built-in "Command Center" touchscreen. It features a modified TKL layout (the touchscreen is where the navigation cluster would normally be), dedicated media and macro keys, a light-diffusing silicon wrist rest, and connection points on either side for the KP7 — a modular, standalone 10-key numberpad with a bunch of extra keys, buttons, and a programmable knob (it's actually so fancy we're going to give it its own review). The KB7 is basically the Corsair Galleon 100 SD, but modular and touch...ier. And it's about as pricey — it's available now, in black, for $199.99 (not including the KP7 — that's a separate $99.99 purchase).

Design and Construction of the KB7

The KB7 has a TKL layout — sort of. It has a full function row and arrow keys, but it has a 4.3-inch touchscreen where the navigation cluster would normally be. It's a slim, well-built keyboard with a brushed aluminum top plate and a plastic chassis. Size-wise, it's pretty standard for a TKL, measuring 15.71 inches (399mm) wide by 6.85 inches (174mm) deep, and it's 1.22 inches (31mm) thick at its thickest point (including keycaps/knobs). It's not particularly heavy — it weighs 1.89 pounds (858g) without the wrist rest, which weighs an additional 9.24 ounces (262g). That's nowhere near as hefty as my current favorite TKL — the 5.25-pound (2,380g) Wobkey Crush 80 Reboot Pro — and it's even a little lighter than Logitech's budget-friendly low-profile wireless G515 TKL, which weighs 1.94 pounds (880g). So, yeah, if you're looking for a keyboard with weight... this isn't it.

The KB7 has a nice spacious setup for a TKL — there's plenty of room above the function row for extra controls, including a dedicated volume knob, two media keys, and four triangular shortcut keys. The keys come with presets programmed in — from left to right, there's a game mode button, a profile switch button, a mic mute button, and a lighting toggle button — but all of these keys, including the clickable volume wheel and the two dedicated media keys, can be remapped in Turtle Beach's Swarm II software. The volume knob is notched and features metal textured edges for a premium feel, but the rest of these extra media/shortcut keys are mediocre at best. Pressing them gives cheap, plasticky feedback — but at least they're easy enough to press thanks to all the extra space around them.

The KB7's standout feature is its 4.3-inch full-color "Command Center" touchscreen, which is located on the left side of the board where you'd normally find the navigation cluster. The touchscreen comes with the navigation cluster programmed in, of course — a full 3 x 3 grid, as well as a link at the top that directs you to download the Swarm II software for full functionality. Once you download Swarm II, you can replace this link with three additional touch-buttons, for a full 12-button grid — similar to the physical LCD key setup of the Corsair Galleon 100 SD.

Of course, the KB7's touchscreen doesn't have the extra knobs that the Galleon 100 SD has, but it is a touchscreen, so you can swipe left and right to switch "pages" of button layouts (most of my dial usage on the Galleon 100 SD is, admittedly, moving between layouts). The touchscreen also features a few non-remappable "buttons," including a home button at the bottom, a profile switch button in the top left, and a settings menu at the top, which drops down to let you change settings such as keyboard and touchscreen brightness, set the actuation point, toggle ReacTap and Rapid Trigger, record macros, and remap keys. (There's also yet another button for lighting adjustment, as well as one for profile switching — functions that are extra-redundant, given that the shortcut keys are also pre-mapped with them.)

The KB7 comes with an accompanying silicon wrist rest that's virtually identical to that of the Roccat Vulcan II Max, which is still one of the prettiest keyboards we've tested to date. The wrist rest is made of soft, smooth translucent silicon with a grooved design that showcases the keyboard's bright, customizable RGB lighting when it's hooked into the bottom of the board. It isn't powered, like the wrist rest that comes with Razer's BlackWidow V4 Pro, it just diffuses the lighting from the board onto the silicon. It's a cool feature that still makes for one of the prettiest keyboard lightshows around. The wrist rest itself isn't particularly plush, but it gives you a nice, gradual slope to the keyboard (which is low-profile, anyway, so you probably don't strictly need it), and it's durable and easy to clean.

Along the top of the keyboard, in the upper left corner, you'll find a USB-C port and a USB-A passthrough port. The passthrough port is nice — and it's also practically necessary, because this keyboard needs a lot of juice. It can connect to your PC through one USB-C port, or it can connect to your PC through two USB-A ports — again, similar to the Roccat Vulcan II Max, which also required two USB-A ports' worth of power to run. I guess that's the price of pretty lighting (and 4.3-inch touchscreens). But at least you get one USB-A port back as a passthrough port, I guess.

The back of the keyboard has two sets of flip-out feet, which raise the typing angle from basically flat to 5 and 7 degrees, respectively. It also features rails on both sides for connecting the KP7 modular keypad — just slide the keypad onto the rails, and you basically have a full-size keyboard (now with a touchscreen in the middle). It's nice that the rails are on both sides, even though I suspect that left-handed numberpad users make up a minuscule percentage of the population.

In the box, the KB7 comes with a few accessories: the silicon wristrest, a 6-foot (1.8m) braided USB-C to USB-C cable, a USB-C to dual-USB-A adapter, a keycap puller, and four shiny, textured keycaps (designed for the W/A/S/D keys, but they can be used on any standard-sized key).

Specs

Typing and Gaming Experience on the KB7

The KB7 is a TKL (-ish) keyboard with Turtle Beach's in-house Titan low-profile linear magnetic Hall Effect switches. They have an adjustable actuation point (0.1 - 3.2mm) and a lifecycle rating of 100 million keystrokes. They're not hot-swappable. Like other magnetic Hall Effect switches, they support all the trendy gaming features that will probably get you banned, including Rapid Trigger, multi-input,controller mode, and ReacTap (SOCD).

I'll admit that I didn't love the way typing felt on the KB7 initially — the linear magnetic low-profile switches combined with the relatively flat, lightweight double-shot PBT keycaps makes for a much flightier typing experience than I'm used to. It took less than an hour for me to get used to it, however, and all in all it's a pretty solid typing experience considering I don't love linear, low-profile, or magnetic switches. The flat, shine-through keycaps make it easy to move your fingers across the board quickly, and the relatively spacious layout means you won't make too many mistakes once you've gotten used to the feel. The board sounds decent — it's not incredible, or anything, but it gives pretty solid thunky feedback for a mainstream gaming keyboard.

I did have some issues that definitely weren't related to my "just getting used to the feel of the board," however — issues that could probably be remedied with a firmware update and software that worked, but, well... I'm not going to hold my breath. I had several issues while I was typing this review, including ghost inputs, missed inputs, and repeat glitches. All of these issues were quickly resolved, but having to stop the review six times because the keyboard was acting up does not give me great confidence in the board's ability to perform under pressure. I didn't have any serious issues while I was gaming, but I also don't generally play a lot of fast-paced first-person shooters where latency and input accuracy is wildly important (I play a handful for testing, but that's it). The KB7 does boast a speedy 8,000 Hz polling rate, but that means nothing if you can't get the right keys to actuate.

The touchscreen performance was decent, though I don't know if I'd trust it in a high-pressure situation. While it registered my inputs quickly and accurately, it's still a touchscreen — hardly the kind of input you want to rely on when you're in the middle of a high-stakes battle for your (digital) life. So, while it's useful for general macro-pad type things, it's not something I'd recommend as part of your ultra-competitive gamer arsenal.

Features and Software of the KB7

The KB7 is customizable via Turtle Beach's Swarm II software, which you can download by pressing the link on the touchscreen. You can use Swarm II to program the keyboard (remapping keys, adjusting actuation, recording macros, etc), the touchscreen, and the keyboard's bright, per-key RGB backlighting.

The keyboard settings are pretty standard for a magnetic switch gaming keyboard — you can turn on and adjust Rapid Trigger, multi-input, and ReacTap, and you can fine-tune the actuation point of every single key individually (a thing that nobody does). You can also remap every single key, except the Windows key and the Fn key, to do what you want — on the primary layer you can only swap around the existing alphanumeric punctuation keys (e.g. if you want to set up an unusual keyboard layout), but on the secondary (Fn) and tertiary (Easy Shift) layers you can set keys to do everything from outputting macros to changing system settings and, of course... launching AI? (Apparently this is such a core functionality that it comes standard in the Swarm II software, now — and it's not just CoPilot. There are options for ChatGPT, Gemini, Deep Seek, and Grok, too.)

But the real customization is, or should be, in the touchscreen. It is, however, unfortunately — but not unexpectedly — limited. There are several native widgets you can swap in, but they're not particularly life-changing: PC performance monitoring (CPU / GPU / RAM), multimedia controls (which there already exist dedicated keys for), opening apps / websites / files / folders, controlling your Turtle Beach devices (by which I mean mouse DPI), executing macros, and a count down timer. There's also a tab for third-party apps, which currently consist of Discord (though you'll need to set it up via the Discord developer portal as an integration), Counter Strike, Steam (which just opens Steam), OBS, and Streamlabs. You can also assign the standard keyboard functions, Windows functions, and shortcuts such as copy / paste, forward / back, launch calculator, etc.

It's a start, but it's definitely nowhere near as robust as the functionality you'll find on Elgato's Stream Decks. Still, this limited functionality is probably workable for most people (after all, you can set up a lot of the functionality you get through third-party apps and plugins by just spending some time creating macros) — if it worked. But the software is, well, let's just say finicky... at best. Setting up the touchscreen seemed to be pretty straightforward, but getting it to stay set up was more of a chore. The software frequently failed in my testing — it would reset, restart, shut down, prompt another six-cycle firmware update (this, at least, I was prepared for — it's pretty standard for Turtle Beach devices to require 14 firmware updates out of the box), etc. When the software was working, the touchscreen worked as expected and could probably replace my Stream Deck. But this was only about 85% of the time, which would probably be decent if my Stream Deck didn't work 100% of the time. And so the touchscreen has ultimately been relegated to a fun gimmick, at least for now.

The Bottom Line

After my surprising love affair with the Corsair Galleon 100 SD, I really wanted to love the KB7 — after all, it's basically the Corsair Galleon 100 SD, but with the option of an actual 10-key numberpad, which was my only real gripe with the latter. But Corsair owns Elgato, which has been making Stream Decks for almost a decade — and has the software and the third-party plugins to back them up. And at the end of the day, a highly customizable macro pad is actually only as customizable as its software allows for. While Turtle Beach currently lacks the integrations to fully compete with Elgato, I do think it has a fighting chance... if it can get the Swarm II software to work, like, at all. I can certainly work without built-in Creative Suite integration, but only if there's something to work with.

I still love the idea of the KB7 and I'm fully ready to hop on the Command Center bandwagon, as soon as Swarm II lets me. But given that this is hardly the first time Swarm II has shot itself in the foot, I'll probably keep this Galleon 100 SD around for a little while longer.

In the enthusiast market, the most shopped-for monitor category is, without a doubt, 4K, specifically in the 27- and 32-inch sizes. The density of 3840x2160 pixels in these two screen formats is, put simply, where you stop seeing the dots. This makes anything look better, be it graphics, photos, documents, videos, or games. And once you’ve experienced it, you can’t go back to QHD or FHD.

The hurdle to resolution nirvana used to be price, but today there are plenty of 4K monitors available for less than $500. But for gamers, the question of frame rates remains. Yes, you can buy a fast 4K monitor, but how much will you have to spend on an equally fast video card? Moving 8.3 million pixels at 200fps is expensive. Luckily, there is a solution – the dual-mode display.

This is a fairly new trend where you get one of the best 4K gaming monitors that can switch to FHD and double its refresh rate with the push of a button. I’ve already reviewed a number of them, and here's AOC’s latest entry, the U27G4XM. It’s a 27-inch IPS screen with 160 Hz, 320 Hz in FHD, Adaptive-Sync, a full-array local-dimming Mini LED backlight with 1,152 zones, wide-gamut color, and DisplayHDR 1000. Let’s take a look.

AOC U27G4XM Specs

The U27G4XM maximizes the nit-to-dollar ratio with a full-array Mini LED IPS panel that has 1,152 local dimming zones. There are three different settings that reduce the black level to increase contrast. This gets the monitor closer to the broad contrast delivered by an OLED at a lower cost, and with higher brightness. DisplayHDR 1000 means a 25% window pattern will exceed 1,000 nits peak, and the U27G4XM achieves this with over 1,100 nits measured from my review sample. Black levels can go all the way down to the backlight being turned off (unmeasurable), but with local dimming on medium, I saw just over 13,000:1.

There’s plenty of color saturation that covers a measured 109% of the DCI-P3 gamut, and that’s without the benefit of Quantum Dots. The U27G4XM is reasonably accurate out of the box, but a quick calibration takes it to reference-level. And there’s a spot-on sRGB mode that’s suitable for color-critical tasks.

The dual-mode feature is a one-button affair that switches between 4K/160 Hz and FHD/320 Hz. It’s great for systems without a high-end video card. To run 160 fps at a game’s max detail level in 4K will require a GeForce RTX 4090. But you can run 320 fps in FHD with a cheaper GeForce RTX 4070 Ti Super. So, what you save on the video card will buy you the U27G4XM with a few bucks left over for a nice mouse and keyboard.

At this writing, that price is $350, which gets you terrific performance and a bright, colorful image. It also includes aiming points, sniper modes, and a frame counter. There are plenty of image modes and HDR support, with four presets for HDR10 content. You even get side USB ports, which I haven’t seen in a long time. The only things missing are LED lights and internal speakers. But you get a 3.5mm headphone jack alongside the HDMI 2.1 and DisplayPort 1.4 inputs, all of which support Display Stream Compression (DSC). On paper, the U27G4XM is an impressive package. Let’s see how it performs.

Assembly and Accessories

The U27G4XM comes in fully recyclable packaging, with molded pulp that protects the contents. The panel, upright, and base assemble without tools into a lightweight setup that fits easily on the average desktop. The accessory bundle includes IEC for the internal power supply, plus nice HDMI and DisplayPort cables.

Product 360

The U27G4XM goes for a chiseled look with flat, angular surfaces making up the back of the panel and an all-screen configuration in the front. Bits of red trim outline the stand’s fulcrum and the offset cable hole in the upright. This is unique in that it’s turned to the side, so you don’t see through it if you’re not using it. The base is cast aluminum with a wide stance and solid construction. The panel snaps on, and though the fulcrum looks large, it’s a little wobbly. It’s perfectly secure, but there is some play when you tilt or rotate the screen. Ergonomics include a 5/23-degree tilt, 20-degree swivel, a 90-degree portrait mode, and a 5.3-inch height range. If you want to use an arm, there’s a 100mm VESA mount with fasteners included.

The front bezel is thin and flush with just a small AOC logo at the bottom. The OSD is controlled by buttons rather than a joystick, a move that probably helped drop the price a tad. A tiny LED indicates power status with a soft glow. The anti-glare layer has a 3H hardness and is free of visible grain.

The input panel is up and under and features two HDMI 2.1 ports and a DisplayPort 1.4 hookup. All ports have DSC and work at full resolution and refresh. There are two USBs here, too, an upstream and a downstream. Three more downstreams grace the right side for easy access. This is a rarity nowadays, and I’m glad to see a nod to convenience from a budget display. There are no internal speakers, but you get a 3.5mm jack for headphones.

OSD Features

The U27G4XM’s OSD is controlled by a row of buttons, which, by modern standards, is a little old-school. But you get one-button access to the refresh mode switch, which is nice.

The dual-resolution feature is up first in the U27G4XM’s Game Setting menu. The easier way to do it is with the second control key, which pops up a little selector. The switch takes a few seconds once you’ve set it up in Nvidia or AMD’s control panel. Also, there are seven picture modes, Standard is the default and best. You also get shadow control to brighten up dark parts of the image, general color saturation, aiming points, sniper modes, a five-level overdrive and a frame counter.

The Picture menu has eight more image modes, which is a bit confusing since you can theoretically combine them with the Game modes for hundreds of combinations. I recommend sticking with Standard/Standard for all content. There are three gamma presets plus fixed color temps and RGB sliders. HDR gets four modes, three of which can be used as emulators for SDR content. HDR10 signals add a mode called Display HDR, which is fairly accurate for color and luminance tracking. Local Dimming gets three levels of aggressiveness. Each further lowers the black level while leaving peak brightness unchanged. Strong turns off the backlight completely under zero signal conditions. You can also choose the sRGB color space here. This mode is very accurate and can be used for critical applications. The U27G4XM also has a PIP/PBP feature for the viewing of two sources at once. The PIP window can be sized and moved.

AOC U27G4XM Calibration Settings

The U27G4XM can be used without calibration in its Standard picture mode. You can use local dimming for SDR and HDR content if you like. The medium setting increases the contrast ratio to around 13,000:1 for HDR and 1,600:1 for SDR content. I improved grayscale tracking to reference level with a few tweaks to the RGB sliders. I also noted that the sRGB mode is spot-on out of the box if you need the smaller gamut. The native color space covers just over 109% of DCI-P3. Local dimming does not change peak brightness, only black levels, so the brightness values below work in all dimming modes.

Gaming and Hands-on

After spending time with the U27G4XM playing games and getting through some work, I found that I only had two minor complaints. The stand, while secure, has a wobbly fulcrum where the panel attaches. This isn’t a big deal except when operating the OSD, which ties into my second nitpick. I am so spoiled by joysticks that on the rare occasion I get control buttons, I am sad. I realize that this is a budget monitor and at that vocation it excels. But pressing the arrow keys while the panel jiggles about is annoying.

These things aside, the U27G4XM is close to flawless. I’ve noticed that budget 4K screens have been getting quicker with each new model I review. This is the fastest one I’ve seen to date with extremely low input lag and smooth panel response from an excellent overdrive. Though 4K tops out at a mere 160fps, it is free of blur or phasing artifacts. In-game movement is precise with instantaneous reaction to control inputs. It seems counterintuitive to pair a $350 monitor with a $1,600 video card, but the U27G4XM qualifies as a competitor’s tool. It’s a frag master for sure.

The dual-refresh feature was seamless in operation once I set it up in the Nvidia Control Panel. I had to set the options for UHD and FHD just once, 160 and 320 Hz, respectively, then it worked every time. I noted that each mode retained its HDR status, on or off. So, if I was in HDR/160 Hz and switched to 320 Hz, it wouldn’t stay in HDR mode if I hadn’t set it that way previously; no complaint, but just to make you aware. Fortunately, I was fine with running at 160 fps in 4K, so I had no real need for 320 Hz. If you have a slower video card, though, it’s a great alternative to backlight strobing, which usually brings phasing artifacts.

The image is gorgeous in both SDR and HDR. The native color gamut covers over 109% of DCI-P3, so there were always vivid hues to enjoy. When I needed the correct color space for SDR, I could reach for the very accurate sRGB option. The U27G4XM is qualified for color-critical apps.

Work tasks were accomplished with a clear and bright picture that made small fonts and graphical details easy to see. I could sit as close as I liked thanks to the U27G4XM’s 163ppi pixel density. The local dimming worked well on its medium setting. It didn’t increase the peak output, which was nice because then there was no need to change the brightness setting. Rather, it lowered the black level, so contrast was increased with no clipping of detail.

Other things of note were the U27G4XM’s USB ports, three of which are on the right side of the panel. Many budget screens forgo USB ports entirely so to find them here at all is a bonus. I also made use of the gaming aids, aiming point, sniper mode and frame counter. The only thing missing is internal speakers. I had to use headphones to hear game sounds.

Takeaway: The U27G4XM is a superlative value. For $350, you get 4K, 160 Hz, dual-refresh flexibility and a bright, colorful picture. It’s accurate out of the box and calibrates to a high standard. Video processing is top-notch with smooth response and super low input lag. Aside from the OSD buttons and the stand wobble, I have no complaints.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

To compare the U27G4XM’s performance, I’ve rounded up dual- and single-refresh 4K 27-inch monitors. The dual models are Titan Army’s P2712V, Acer’s XB273K V5, and ViewSonic’s VX2730D-4K. Single-mode displays include TCL’s 27R94 and Gigabyte’s GS27U.

Pixel Response and Input Lag

Click here to read up on our pixel response and input lag testing procedures.

The monitors in the response time chart are different because it’s unfair to compare LCDs and OLEDs in this test. OLED’s black-to-white transition is much faster, which is why they are smoother at lower refresh rates. On planet LCD, the XB273U F5, a G-Sync Pulsar monitor, shares the speed crown with the ViewSonic VX2730D-4K in its 288 Hz FHD mode. The U27G4XM is a bit behind at 3.58 and 3.74 ms, respectively. Doubling the refresh rate does not halve the response time. This is with the overdrive set to medium, which is the smoothest option that doesn’t have ghosting artifacts.

In the lag test, the U27G4XM takes the win at 320 Hz with a 14ms average of 25 mouse clicks. At 160 Hz, it’s 16.5ms, which is faster than nearly every other screen, including two that are running at 320Hz. So, if you can run in 4K, you’ll have one of the quickest examples available.

Test Takeaway: The U27G4XM has an average panel response for a gaming LCD with an overdrive that removes most motion blur. It’s one of the quickest in my lag test with 14ms at 320 Hz and 16.5ms at 160 Hz. That puts it above all the other 27-inch 4K LCDs I’ve tested except for the ViewSonic VX2730D-4K at 288 Hz. It is certainly the quickest 4K monitor you can buy for $350.

Viewing Angles

The U27G4XM’s off-axis image quality is a little better than most of the IPS panels I’ve photographed. The side view has a slight red tint but no change to brightness or gamma. The top view is also a bit red with reduced brightness and gamma. This is a very shareable monitor.

Screen Uniformity

To learn how we measure screen uniformity, click here.

It’s safe to say that IPS glow is a thing of the past. Today’s panels, like the U27G4XM, show excellent uniformity with no visible issues whatsoever. My sample had no problems in any part of the screen and was uniform in all field patterns, color and neutral.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

To read about our monitor tests in-depth, please check out Display Testing Explained: How We Test PC Monitors. We cover brightness and contrast testing on page two.

Uncalibrated – Maximum Backlight Level

The U27G4XM has more than enough light output for any indoor or outdoor environment. With nearly 500 nits peak available, it could be used at a location shoot or turned down to 50 nits for dark room gameplay. The panel’s native contrast is 923.3:1, a bit below the IPS average of 1,000:1. With local dimming set to medium, it delivers around 1,600:1 for SDR content with no clipping in shadow or highlight areas.

After Calibration to 200 nits

Calibration bumps the contrast up to 989.2:1, a slight improvement. With local dimming on medium, it’s around 1,600:1. If you set it to strong, the backlight is shut off, making measurements impossible. ANSI contrast stays relatively consistent at 922.1:1, which is average for IPS panels with either full-array or edge backlights.

Test Takeaway: The U27G4XM offers effective local dimming for SDR content, which sets it apart from some other screens that only dim for HDR. It has nearly 500 nits peak available, making it good for brightly lit environments. Native contrast is just below 1,000:1, and dynamic contrast is 1,600:1 with local dimming on medium. Image depth and texture are exemplary at this price point.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

The U27G4XM ships in its Standard picture mode. In the OSD’s parlance, that’s the Standard Game Mode and Standard Eco Adjustment. There are gamut selections available and accuracy is high both before and after calibration.

Grayscale and Gamma Tracking

Our grayscale and gamma tests use Calman calibration software from Portrait Displays. We describe our grayscale and gamma tests in detail here.

The U27G4XM’s default grayscale is a tad cool, but errors barely crack the 3dE threshold, the point where they become visible. Most are below the line, so this monitor can be enjoyed right out of the box. Gamma is right on the 2.2 reference line as well, with no visible issues.

Calibration takes grayscale tracking down to a pro-level 0.42dE. It doesn’t get much better than that. Gamma gets a shade light from 50-90%, but the change is barely visible in actual content. This is excellent performance.

The sRGB gamut mode cannot be calibrated, but it has visually perfect grayscale tracking and slightly light gamma that closely tracks the 2.2 reference. It is suitable for color-critical tasks.

Comparisons

The top three screens, including the U27G4XM, are visually identical in the grayscale test. 2.40dE is a solid result. After calibration, there is no visible difference among the monitors. The U27G4XM’s 0.42dE is a standout result. In the gamma test, the range of values is fairly tight at 0.11 from smallest to largest. Overall tracking is a tad light with a 3.18% deviation representing an actual value of 2.13.

Color Gamut Accuracy

Our color gamut and volume testing use Portrait Displays’ Calman software. For details on our color gamut testing and volume calculations, click here.

The U27G4XM is a very colorful monitor with extra punch across the entire gamut triangle. The default is close to perfect with slight oversaturation and hue errors in red and green. The points track linearly, so there is no loss of detail in any part of the image. Calibration only makes a slight difference that will be hard to see in a before-and-after comparison. But it’s worth doing for the improvement in grayscale tracking. The sRGB mode is better than many pro monitors I’ve tested with an average error of just 0.80dE.

Comparisons

As impressive as the U27G4XM’s accuracy is, it is competing with some equally precise displays. 2.27dE is a great score, but others score better. Is that a difference worth paying for? Not in my opinion, especially when you consider gamut volume, which is the U27G4XM’s main advantage. Amazingly, it does not have a Quantum Dot layer, but it does a great impression of that tech with over 109% coverage of DCI-P3. You will see that in the brightest content where primary colors dominate. There is a good deal of extra red, green, and blue. The sRGB result of 97.93% is nearly ideal and qualifies the U27G4XM for critical tasks.

Test Takeaway: The U27G4XM can be enjoyed without calibration, but a few tweaks take it to reference levels. Gamma is a touch light, but color is exemplary and richly saturated. It has more coverage than the competition, which is a difference that can easily be seen. The sRGB mode is very precise with nearly perfect color and grayscale.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

Our HDR benchmarking uses Portrait Displays’ Calman software. To learn about our HDR testing, see our breakdown of how we test PC monitors.

Mini LED displays like the U27G4XM are great for HDR content thanks to their bright backlights and full array zone dimming (FALD). This monitor is rated for VESA DisplayHDR 1000, which is something rarely seen in the OLED category. The signal switch is automatic, and you get four additional picture modes.

HDR Brightness and Contrast

It’s easy to identify which screens are Mini LED in the first chart above. The U27G4XM exceeds its HDR1000 rating, reaching a peak of 1,107 nits. I measured that from a 25% window pattern, and it is extremely bright. Only the TCL is brighter. Local dimming can be used to make the contrast ratio unmeasurable at its highest setting. On medium, the ratio is 13,061.4:1, which is excellent for an IPS panel. I noted that different local dimming settings did not change the peak value, only the black level.

Grayscale, EOTF and Color

In the HDR grayscale test, the U27G4XM is slightly warm in tone from 70% brightness and higher. The error is hard to spot in content, but it would be nice to have RGB sliders available. Of the four HDR modes, Display HDR delivers the best EOTF with values a little below the reference until 60%, where it goes a bit light. These are small variations that will not negatively impact the image. The tone map transition is at 75%, which is correct for the measured white and black levels.

In the color test, the U27G4XM’s large gamut is used to full effect with general oversaturation in all colors. This gives HDR content some extra punch that isn’t strictly accurate, but it is pleasing. The BT.2020 test shows the same response with color running out at 90% red, 80% green and 95% blue. In both charts, hue targets are on point for all six colors.

Test Takeaway: The U27G4XM is a superb HDR display with bright highlights and deep shadows. The Mini LED backlight effectively increases contrast to over 13,000:1 with no penalty. Color is pleasingly oversaturated with vivid hues that deliver great impact to game and video content.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

With OLED commanding the premium market, it has driven down the prices of other premium technologies, namely 4K and Mini LED. It is now possible to buy a 27-inch 4K IPS monitor with wide-gamut color, 160 Hz, premium video processing, accurate imaging, and high brightness, and that is a clear sign of progress. The U27G4XM exemplifies that design goal for $350 and offers performance we’d have paid over $1,000 for just a few years ago. And did I mention the Mini LED backlight with 1,152 dimming zones? Yeah, it has that too. For perspective, Acer’s X27 from 2018 had 384 dimming zones, 4K, and 144 Hz for an eye-watering $2,000. Check out that review here if you’re curious.

The AOC U27G4XM is better than that Acer in every way. It has 1,152 dimming zones. It has 109% coverage of DCI-P3 color. It has 160 Hz and a 320 Hz FHD mode. It has lower input lag and better panel response – and it’s $350. That’s what eight years of progress looks like. My only complaints are that the OSD is controlled by buttons instead of a joystick, and the stand is secure but a little wobbly. But performance and image-wise, there are no flaws. For the money, I can’t fault it at all. It games with the very best and provides a bright, colorful, and accurate picture for work and play alike.

If you’re looking for an affordable 4K monitor that pairs well with budget and premium video cards alike, the AOC U27G4XM is well worth checking out.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

I realize I'm probably in the minority, but I love full-size keyboards because I love 10-key numberpads. For data entry and for everything else (let's face it, I don't do much data entry). Since I often have to test keyboards that aren't full-size, I'm always on the look out for a great standalone numberpad — but the options usually consist of $15 (and under) bargain bin finds or difficult-to-justify $100+ premium picks.

Turtle Beach's Command Series KP7 isn't exactly in the middle — it retails for $99.99, after all — but it's an interesting option nonetheless. This 28-key modular gaming keypad is part of the brand's recently-launched Command Series, which consists of several modular, multi-input peripherals (two keyboards, three mice, and this keypad). It's designed to work with the series' flagship Turtle Beach Command Series KB7 keyboard, which is a premium TKL keyboard with a built-in touchscreen. The KP7 hooks onto either side of the KB7, turning it into a full-size keyboard with a numberpad and a bunch of extra macro keys. The KB7 retails for $199.99, so the full set is just shy of $300. But the KP7 is interesting because it works very well on its own — in fact, it might even perform better as a standalone keypad than it does with the KB7 it's designed to complement.

The KP7 is a modular gaming keypad with 28 keys, one clickable, programmable dial, and one (non-programmable) button for switching between numberpad mode and game mode. It can hook onto the right side of the KB7 to function as a typical numberpad, or it can hook onto the left side to function as a gaming keypad — complete with an extendable thumb bar. It features magnetic Hall Effect linear switch, bright per-key RGB backlighting, and its own mini wrist rest. It's available now, as a separate purchase from the KB7, for $99.99.

Design of the KP7

The KP7 is a standalone macropad that consists of a 10-key numberpad surrounded by a bunch of other keys. At the top, there are three additional Fn keys, as well as a profile switch key — it seems Turtle Beach is really trying to push multiple profiles, because if you use this with the KB7 keyboard, this marks the fourth dedicated way to switch profiles. On the right of the numberpad are two columns of keys: The first column consists of the math function keys that usually surround a numberpad, and the second column has four dedicated macro keys.

In the bottom right corner there's a programmable knob and an additional thumb button, which is set up by default as a space key. In the upper right corner there's a small button that switches between numberpad mode and game mode — game mode turns the center keys into W / A / S / D (and changes the surrounding keys to the corresponding keys on the left half of the keyboard. It essentially becomes a gaming half-keyboard, sort of like Nordic Game Supply's The Shrimp.

For the times you're using it in gaming mode, the KP7's lower right cluster of keys, including the dial, extend down and right so you can use them with your thumb while you've got your left hand on the center keys. The wrist rest is even cut out to accommodate the keys when extended, which is a nice touch.

The KP7 is well-built for a standalone numberpad — it follows the same design philosophy as Turtle Beach's KB7, with a brushed aluminum top plate and a plastic chassis. It has low-profile linear switches, flat, double-shot PBT keycaps with shine-through legends, and bright, customizable RGB backlighting that extends to its silicon wrist rest, which is like a mini version of the wrist rest that comes with the KB7 (and, previously, the Roccat Vulcan II Max). The wrist rest slots into the bottom of the numberpad and is made of misty, translucent silicon with a grooved pattern that acts as a conduit for the lighting in the keypad. The wrist rest only attaches on one side of the keypad, so the lighting effect isn't quite as bright or mesmerizing as is the lighting effect on the keyboard, but it's still pretty and comfortable.

The KP7 has connectors on both sides that are designed to slot into the rails on either side of the Turtle Beach KB7 keyboard. The keypad gets its power from the keyboard when it's connected, so you shouldn't have to plug it in separately via USB-C, which is for when you want to use the keypad alone. However, I found that the keypad's lighting seemed to be inconsistent when it was powered through the keyboard — it flickered and reset itself several times throughout my testing, but this did resolve itself when I plugged it in via its own USB port. Of course, plugging it in via USB port means the KB7 and KP7 keyboard combo takes up either one USB-C and one USB-A port or three USB-A ports, which is a lot for what is essentially a single full-size keyboard.

In the box, the KP7 comes with a 5.9-foot (1.8m) detachable braided USB-C to USB-A cable, a silicon wrist rest, and four glossy, textured keys keys (to replace the center W / A / S / D keys, for quicker tactile recognition), and a keycap puller. The back of the keypad has two sets of flip-out feet that raise the typing angle from flat to 5 - 7 degrees. These match the KB7, so you can raise and lower the entire setup as a single unit.

Specs

Performance of the KP7

It's hard to judge a keypad's performance on its own, but the KP7 has enough keys that it can function as a single-handed "controller." It features the same linear magnetic Hall Effect switches as we saw in the KB7, with the same functionality — individually adjustable actuation point (between 0.1 and 3.2mm), Rapid Trigger, multi-input, and ReacTap (SOCD). It also sports an 8,000 Hz polling rate, so I guess if you're playing a game that only requires the left half of your keyboard, this might be one way to get the ultra-low latency and speed of an 8,000 Hz magnetic switch keyboard without having to buy the whole keyboard (though it does cost exactly half of what the KB7 costs, and I don't think it's quite half the size).

The low-profile switches and flat keycaps make for pretty speedy input, and while the layout is slightly different from standard numberpad layouts, it's not too bad — in numberpad mode, the first column to the right of the numbers has five keys, but the bottom two keys are both Enter and the two keys above that are both + (normally, the Enter and + keys are vertical 2U keys). It's not quite the same, but it worked well enough for me for data input and other numberpad tasks.

The switches themselves are pre-lubed linear switches with a smooth, consistent keypress and no tactile or audible bump. Typing on the KP7 (well, for as much typing as you can do) is a surprisingly pleasant experience — it's not going to wow any hardcore keyboard enthusiasts, but the linear switches, slim, flat keycaps, and speedy, lightweight Hall Effect actuation is a great combo for extra gaming keys. And, as the KP7 is entirely tactile, it's much easier to integrate into your work (or game) flow. The extendable thumb bar (which just pulls away from the keyboard in a diagonal / downward direction) is also comfortable when you're using the KP7 as a gamepad. The keys on the thumb bar are well-placed, but I did find the knob difficult to turn without hitting other keys, as it's flush with the keys around it.

Software

The KP7 is programmable via Turtle Beach's Swarm II software, which originally belonged to Roccat. It's designed to work in tandem with the KB7 keyboard, but you can still program it fully without having to depend on the keyboard at all. You can use Swarm II to change the keypad's actuation, toggle on and adjust Rapid Trigger and ReacTap, and set up multi-input. You can also use it to remap keys up to four layers deep (Fn, Game Mode, and Easy Shift).

You can customize the KP7's bright, per-key RGB lighting in the software. This works pretty well when you're just customizing the KP7 alone, but I had issues getting the keypad to light up seamlessly alongside the KB7 (they were often out of sync, and/or the KP7 seemed to go into hibernation immediately and often when it was connected to the KB7 — even if it was being powered separately via its own USB-C port.

The Bottom Line

The KP7 is an interesting product — it's designed to work with the KB7, but I actually like it better as a standalone keypad. Partly because it works better as a standalone keypad — it was very finicky in my tests when it was connected to the KB7, regardless of whether it was powered separately or through the keyboard. This is probably because of the software, which, while fairly straightforward for the KP7, still wasn't great. For example, I was prompted to update the KP7's firmware everytime I connected it to the KB7 — but not when I used it as a standalone keypad.

The KP7 isn't the only standalone numberpad-slash-gaming-keypad out there, but it's the only one I've seen that manages to switch between the two modes pretty seamlessly. We tested Nordic Game Supply's The Shrimp mini-keyboard a few years ago, but that was definitely designed as a gaming keypad first and a numberpad... not at all. At $99.99, the KP7 isn't cheap, but it's versatile enough that you can definitely make it worth it.

Qualcomm has steadily iterated on its Arm SoC efforts in the Windows space and made further gains with the launch of the Snapdragon X platform.

Now we’re seeing a wave of new systems using Qualcomm’s Snapdragon X2 series of Arm processors, and the HP OmniBook Ultra is the latest to take a bow. Our review unit is powered by a new flagship Snapdragon X2 Elite (X2E-90-100) SoC, paired with a generous 64GB of memory and a capacious 2TB SSD, and features an all-new chassis design.

The OmniBook Ultra proved to be a top contender in performance benchmarks and is a great all-around system, but about that price…

Design of the HP OmniBook Ultra 14

Our OmniBook Ultra 14 review unit arrived with an entirely new design language compared to the Ryzen AI 9 HX 375-equipped system that I tested in 2024. This year’s design features a more angular look and is finished in a color that HP calls “Stone Blue.” The edges of the forged aluminum-alloy chassis are highly polished to a mirror-like finish. It has a decidedly more upscale appearance compared to its predecessor, although the overall quality feels the same (which is to say, excellent).

Upon opening the OmniBook Ultra, you’ll notice “OmniBook Ultra” printed vertically along the forward right palm rest of the keyboard deck. “OmniBook” is printed on the chassis, while the Ultra is actually engraved with the same polished finish as the chassis edges. A massive 5.4 x 3.4-inch trackpad sits at the center of the deck, while the keyboard is directly above, with the keys finished in dark “stone blue” plastic. One thing of note: the fingerprint reader integrated into the power button on the 2024 model is missing this time around.

The 14-inch display has tiny bezels on the left and right, and thicker ones along the top and bottom. The bottom bezel also features “HP” branding in the middle. The display is covered with a glossy glass, which does wonders for improving image vibrancy but tends to increase reflections.

I complained about the fussy USB-A port on the previous-generation OmniBook Ultra, which was partially covered by a spring-loaded door. This time, HP just ditched the USB-A port entirely and replaced it with another USB-C port. As a result, the new OmniBook Ultra has a single USB-C port and a 3.5 mm jack on its left side, and two USB-C ports on its right side. None of the ports support Thunderbolt, although they do feature USB Power Delivery 3.1, DisplayPort 1.4, and a 40 Gbps signaling rate.

The OmniBook Ultra measures 12.25 x 8.49 x 0.42 inches and weighs 2.81 pounds. The Asus Zenbook A16 is just a hair heavier at 2.87 pounds, while the 13-inch MacBook Air is a smidge lighter (2.7 pounds). The previous generation OmniBook Ultra tips the scales at 3.4 pounds.

HP OmniBook Ultra 14 Specifications

Productivity Performance on the HP OmniBook Ultra 14

The OmniBook Ultra is among the first laptops we’ve tested to use the new Snapdragon X2 Elite SoC (up to 5.0 GHz clock speed, 18 cores), with the first being the Asus Zenbook A16. As we learned in our review of the ZenBook A16, the Snapdragon X2 Elite is a strong performer, and it’s made even more potent by the inclusion of 64GB of LPDDR5x-9523 in our review unit, paired with a 2TB PCIe 5.0 SSD.

The Snapdragon X2 Elite also includes an integrated Hexagon NPU, which delivers up to 85 TOPS of compute performance.

The OmniBook Ultra made a strong showing in the Geekbench 6 CPU benchmark, securing a single-core score of 3,942, putting it in second behind the MacBook Air with its M5 SoC (4,168). The ZenBook A16 wasn’t far behind with 3,807. Looking at multi-core performance, the ZenBook A16 jumped up in the lead with 22,733 compared to 20,075 for the new OmniBook Ultra. The new OmniBook Ultra was also well ahead of the Ryzen AI 9 HX 375 model (2,846 single-core, 14,838 multi-core).

HP has opted for a PCIe 5.0 SSD on the OmniBook Ultra, and our review unit came in at a healthy 2TB of capacity. The laptop dominated our 25GB file transfer test, delivering 2,620.91 MBps compared to 1,924.84 MBps for the second-place MacBook Air.

The Snapdragon X2 Elite in the OmniBook Ultra again flexed its muscle in our Handbrake test, which involves transcoding a 4K video file to 1080p. Here, the OmniBook Ultra completed the task in 2 minutes and 11 seconds, putting it just three seconds behind the ZenBook A16. Both the AMD-equipped UltraBook Ultra and the MacBook Air took well over 4 minutes to complete transcoding.

We run Cinebench 2024 through ten loops to stress-test ultrabooks. The first run delivered the highest score at 1,099. The subsequent runs fell short of that high mark, settling between 920 and 950 towards the end of the test. The 12 Prime cores averaged 4.03 GHz, while the 6 Performance cores averaged 3.05 GHz.

Display on the HP OmniBook Ultra 14

HP is using a 14-inch 2880 x 1800 display on the OmniBook Ultra, and it, like many in this segment, is an OLED panel. The panel has a 120 Hz refresh rate and a glossy finish. However, with the brightness cranked to the max, stray reflections were hardly noticeable.

I used the OmniBook Ultra to watch about half an hour of Marvel’s Thunderbolts on Disney Plus, and spent a good chunk of time watching the San Antonio Spurs blow the biggest lead in an NBA championship series via Sling TV. In both instances, the display looked great with accurate colors and excellent contrast.

In our testing, the OmniBook Ultra covered 87% of DCI-P3 and 122%of sRGB, placing it at the top of the rankings among the assembled laptops. Maximum brightness measured crested 400 nits, specifically hitting 414 nits. While impressive for an OLED display, the ZenBook A16 (429.4 nits) and MacBook Air (458.8 nits) were brighter still.

The display on the OmniBook Ultra is multi-touch-capable, and I had no trouble navigating the Windows interface or using swiping gestures. However, reaching out to touch the screen isn't a natural movement for me, and it isn’t very ergonomic. I was perfectly fine using the humongous trackpad.

Keyboard and Touchpad on the HP OmniBook Ultra 14

The OmniBook's keyboard is full-size and backlit, and the lattice-free keys themselves felt good under my fingers. There’s no clickiness to the keys, and you’ll hear a gentle, low “thud” as you type along. There’s plenty of key travel, and I was comfortable typing on the keyboard for an extended period of time. My only complaint concerns the half-height up and down arrows squeezed between the full-size left and right arrows.

Using my trusty keyhero.com typing benchmark tool, I managed 87 words per minute with 97 percent accuracy. This was roughly equivalent to what I scored with the keyboard on the previous-generation model.

Complementing the keyboard is a haptic touchpad measuring a large 5.4 x 3.4 inches. It’s a haptic touchpad that provides a satisfying “click” no matter where you press on its surface. While some top-hinged touchpads require additional force to register a click as you move towards the top of the mousing surface, there are no such limitations on the OmniBook Ultra.

Audio on the HP OmniBook Ultra 14

The OmniBook Ultra 14 features quad downward-firing speakers on the underside near the front of the chassis. The speakers sounded excellent, with little to no distortion at higher volume levels (80% and higher), and were loud enough to fill my home office.

I played a wide variety of music, ranging from Steve Wonder’s keyboard-heavy and soft vocals on Superwoman from 1972 to the ear-splitting screaming and electric guitars of Linkin Park’s Lying from You, from the band's early 2000’s sophomore album.

Interestingly, no sound utilities are installed by default to tune the speakers further.

Upgradeability of the HP OmniBook Ultra 14

Hopping inside the OmniBook Ultra is incredibly easy – there are only four screws to remove from the bottom panel, which is at least half as many as you’ll typically find on a thin laptop.

Once the bottom panel is removed, there’s not much that’s replaceable. The battery is at least easily accessible, with four screws securing it and one cable connecting it to the motherboard. You also have access to the PCIe 5.0 SSD, which sits beneath a heat shield. That’s about it when it comes to upgrades for the OmniBook Ultra.

Battery Life on the HP OmniBook Ultra 14

Our OmniBook Ultra 14 review unit came powered by a 70 WHr battery, which is topped off using a 65-watt GaN USB-C charger. Our battery life test includes web browsing, video streaming, and WebGL tests, with the display set to 150 nits.

We were able to extract 13 hours and 27 minutes out of the OmniBook Ultra, exceeding the runtime of the ZenBook A16 with the same Snapdragon X2 Elite chip by 3 hours. It also outlasted its predecessor by over half an hour. The only laptop that stood in its way was the MacBook Air, which lasted an impressive 15 hours and 28 minutes.

Heat on the HP OmniBook Ultra 14

Heat output on the OmniBook Ultra was measured while running our Cinebench 2024 stress test. The keyboard measured around 95 degrees Fahrenheit between the G and H keys, while the hottest part of the laptop was the bottom rear near the exhaust vents, at 118 F.

The touchpad stayed cool to the touch at 77 degrees, the ambient temperature in the room at the time.

While we usually mention CPU temperatures in this section, HWInfo, in its current iteration, does not provide the specific values we need for reporting.

Webcam on the HP OmniBook Ultra 14

HP has included a 5MP webcam on the OmniBook Ultra, which has IR capabilities to support Windows Hello biometric authentication. Like the previous generation, the image quality was decent, but not segment-leading.

My face and shirt looked slightly blurry, and my skin tones were "off." This performance would be passable for a laptop costing in the $1,000 to $1,500 range. But for a laptop with an MSRP of nearly $2,900, it's a tough pill to swallow. The Lenovo Yoga Slim 7i Aura Edition I recently tested, with an MSRP of $1,700, had a far better onboard webcam.

Software and Warranty on the HP OmniBook Ultra 14

HP loads plenty of its own utilities on the OmniBook Ultra, which seems a bit excessive. There’s an HP app that provides limited system information and access to HP’s online store to purchase accessories. There’s also HP Documentation, HP Energy Star (lets you purchase Energy Star-compliant appliances… why?), HP Privacy Settings, HP System Event Utility, and HP TV+. Finally, there’s Find My HP by Absolute, which allows you to track/locate, remote lock, and remote wipe your laptop.

And of course, there’s also the usual Windows bloat in the form of app links to Booking.com, Dropbox, and various other services. And we can’t forget McAfee, which is always an instant delete when I set up a new PC for friends and family.

The OmniBook Ultra comes with a 1-year manufacturer's warranty.

HP OmniBook Ultra 14 Configurations

There are several prebuilt versions of the OmniBook Ultra with Snapdragon X2 Series chips, the cheapest starting at $1,249.99. It features a Snapdragon X2 Plus processor, a 512GB PCIe 5.0 SSD, 16GB of RAM, and a 14-inch 1920 x 1200 OLED display.

The next prebuilt system comes with a Snapdragon X2 Elite processor, 32GB of RAM, a 1TB SSD, and a 2880 x 1800, 120 Hz OLED display for $2,399. For an extra $300, you get the same specs with a 2TB SSD.

Finally, our review configuration can only be obtained through the “Customize and Buy” option, which gives you a wide pricing range to play with. It has the Snapdragon X2 Elite processor, 64GB of RAM, a 2TB SSD, and a 2880 x 1800 OLED display, priced at $2,879.99.

Bottom Line

HP has built a solid foundation with the OmniBook Ultra and the Snapdragon X2 Elite SoC. Our review unit came fully loaded, with 64GB of memory and a capacious 2TB SSD. The ultrabook excelled in our productivity benchmarks, including a stellar run on our storage test. In addition, its battery endurance was only topped by the 13-inch MacBook Air. The laptop also has a beautiful 120 Hz 3K OLED panel and comes standard with a Wi-Fi 7 radio.

On the other hand, the biggest knock against our particular unit is the price: nearly $2,900 is a tough pill to swallow for an ultrabook, primarily due to its lofty memory and storage configurations. However, if you go light on options, you can dramatically lower that price. If you go light on the options, you can opt for a Snapdragon X2 Elite chip, 32GB of RAM, 512GB SSD, and a 3K OLED display for $1,719.99, or add a 1TB SSD for an additional $120.

At that price, the OmniBook Ultra makes a lot more sense and would help us overlook the lack of Thunderbolt ports and the so-so webcam.

The latest CPU cooler air cooler from Taiwanese manufacturer Montech PC is the NX600, a budget air cooler that incorporates six copper heatpipes to transfer heat from the CPU to the fin plates.

What sets the NX600 apart from the competition is the inclusion of two high-performance, thick E28 fans. And the price is right, at less than $30, which makes the purchase arguably worth it even if you’re only interested in the fans (and not the heatsink).

Let's take a look at the specifications and features of the cooler, then we’ll go over thermal and noise benchmarks so you can decide if the NX-600 deserves to make our list of the best CPU coolers.

Cooler specifications

Features of Montech’s NX600 air cooler

● Dual-tower heatsink with six heatpipes

Six copper heatpipes transfer heat from the CPU contact plate to the fins of the heatsink. The heatsink features serrated edges – an engineering choice which generally results in lower noise from turbulence as air enters and exits the fin stack. There are interlocking “zipper” tabs on the sides of the fins, which improve structural rigidity and prevent the plates from squishing together to ensure proper airflow.

● Dual 120mm E28 performance fans

There’s more to a cooler than just the heatsink. The included fans directly impact on noise levels and cooling performance. The E28 fans are especially powerful, as we discovered in our review of Cooler Master’s Hyper 212 3DHP – dropping temperatures by over 3 degrees C compared to the default fan included with the 3DHP air cooler!

● Color scheme options

The color scheme of the NX600 model included in this review is black, grey, and silver, which might appeal to users who prefer old-school designs. There’s also a white option, shown below. ARGB and non-ARGB fan options are available with both heatsink designs, for about $5 more.

The top covers of the heatsink are mainly decorative, and can be removed if you so desire – as shown in the picture below.

● RAM Clearance

Standard-height DDR5 sticks 40mm tall fit well under the NX600 with the E28 fans installed, but taller sticks won’t fit perfectly underneath. Our current CPU cooler test bench incorporates TeamGroup’s Sakura Rose T-Force Xtreem DDR5-7200 sticks, 48.8mm (1.92 inches) tall.

If you also use taller RAM DIMMs, you’ll have to raise the intake fan’s placement by a few millimeters for things to fit properly. This might result in slightly lower cooling performance in certain scenarios, particularly those where the fans’ speeds are limited to ensure lower noise levels.

● Packaging

The heatsink towers are protected by soft coverings and foam inserts, with the fans and accessories packaged in cardboard. Included in the package are the dual-tower heatsink, two high-performance E28 fans, mounting hardware for Intel and AMD platforms, fan clips, and a small tube of thermal paste.

● AM5 Installation

You’ll need to first remove the default AM4/5 retention mechanism and then place the mounting studs around the exposed screw holes.

The next step is to place the mounting bars on top of the studs, securing them with the included screws.

Then, apply Montech’s thermal paste. If you’re not sure how to do that, we have a hadny thermal paste application guide that covers the different methods you can use. Afterwards, place the heatsink tower against the CPU and mounting bars, and use a screwdriver to secure it.

Once the heatsink is secured, attach the high-performance E28 fans to the cooling tower with the included clips. Lastly, connect the PWM and (optional) ARGB cables of the fans to the corresponding headers of your motherboard.

Real-world testing configuration – AMD AM5 platform:

We’ve tested coolers with both the Ryzen 9950X3D and its non-V-Cache sibling, the 9950X. There are some differences in how the 9950X and 9950X3D CPUs are impacted by thermal events. While the heat output of the CCDs of AMD’s 9950X3D is relatively balanced, the 9950X I used has one CCD that runs much hotter than the other, with a difference of over 10 degrees Celsius in some scenarios, shown below.

We’ve since returned to using a 9950X3D for cooler testing, as it has a more balanced heat profile, and is almost certainly a more widely adopted CPU.

The benchmark results shared in these reviews may differ from others because I emphasize results that are comparable to real-world use. This means I generally test CPU coolers inside of a closed desktop case, which increases cooling difficulty compared to other testing methods.

Many reviewers test coolers on an open test benches, which have a combination of lesser airflow needs and lowered ambient temperatures. This results in making weak coolers appear stronger than they really are. Some publications have also used generic thermal plates to test cooling solutions. I reject both of these methods because they don’t accurately reflect real-world cooler conditions.

Our latest testing setup uses the FLOVA F50 computer case from Tryx.

This case features a unique “crossflow” fan that pulls air from the side, which the company claims is more effective than traditional intake fans. For air cooling tests, we’ve added a single Noctua NF-A12 G2 intake fan.

PBO Performance and maximum noise levels

We’re going to start this review’s benchmark section by focusing on a traditional maximum performance test, with the CPU cooler’s fans allowed to reach their fastest speeds, for the best cooling possible.

Turning on PBO allows AMD’s Ryzen 9 9950X3D to stretch its legs to an extent, and all air coolers I have tested with PBO enabled using MSI’s X870E Carbon Wifi motherboard reach the maximum CPU temperature of 95 degrees C (203 F) and thermally throttle to some extent.

Results from the liquid coolers we’ve recently tested aren’t shown above because they are able to keep AMD’s Ryzen 9 9950X3D under TJ Max – and as such, power consumption figures aren’t quite comparable.

Montech’s NX600 does fairly well in this test. While it is technically beaten by Sudokoo’s SK700V & DeepCool’s AK620 G2, it only falls behind by a few watts. This means benchmarks will basically be the same, as any scaling gained from raising power consumption to 251W (compared to 248W) would be so small as to be irrelevant.

To give a wider variety of comparison examples, I’ve included a chart of the same tests from our last AMD Ryzen 9 9950X3D cooling test bench. But keep in mind these results aren’t 100% comparable, due to a different case and fans being used. Our newest test bench used for this and other recent reviews incorporates a Tryx crossblade intake fan as well as a single Noctua NF-A12 G2 intake fan, which results in better air cooling performance.

Montech’s NX600 performs well, but a large part of that comes from the inclusion of its flagship E28 fans, which are a bit overpowered compared to the included dual-tower heatsink. As such, at full fan speeds, this cooler does not run at all quietly – with a measurement of 50dBA, it is the loudest air cooler I’ve ever tested.

200W thermal benchmarks

For the next thermal test, I leave the motherboard settings at their defaults, which results in a power limit of 200W when running Cinebench R23.

The result here was curious, as with a “stock” power profile, its thermal performance was not as strong as our previous test.

150W + GPU thermal results, noise levels

Our next test runs Cinebench on the CPU with a 150W power limit, while also running Furmark on MSI’s RTX 4070 Ti Super Ventus 3x OC. This causes the GPU to consume ~295W of power. This test is designed to emulate the thermals of games, which primarily stress the CPU and GPU.

Montech’s NX600 did particularly well with this combined CPU and GPU thermal test; the included E28 fans seemed especially effective in helping to push the GPU’s heat outside of the case.

But again, the cooling potential provided by the E28 fans comes at the cost of noise – with the NX600 averaging 44.6 dBA. Between our current (shown above) and former (shown below) testing setup, this result is the third-loudest we’ve seen from any air cooler on this test.

Noise-normalized testing

Most testing is performed with the cooler tied to the default fan curve of our MSI X870E Carbon motherboard, but some of y’all prefer to see tests when the noise levels of coolers are equalized. This is especially important to those of you who prefer silent computers. This next test has the CPU cooler noise-normalized to 38.9 dBA, with PBO enabled for the Ryzen 9 9950X3D CPU.

The results here were not impressive for the Montech. Perhaps the thermal transfer from the NX600’s CPU coldplate and copper heatpipes is not ideal, as performance should be better with these flagship E28 fans.

Another factor to consider when evaluating this benchmark is that my test bench utilizes DDR5 DIMMs that are 49mm tall – which requires lifting the front intake fan a few millimeters to fit. The particularly poor noise-normalized performance may be the result of reduced performance when the fans are set to lower noise levels and rotation speeds.

Karhu DDR5 RAM thermals testing!

Your CPU cooler does not operate in isolation. It has an impact on not just your CPU’s temperatures, but also the other components in your build, like your RAM and GPU. To that end, I’ve run the Karhu RAM stress test. This places a load of ~153W on the CPU and ensures system RAM (DDR5 in my case) is fully stable. In this type of scenario, most AIOs tend to produce worse results than air coolers.

The DDR5 temperature recordings from this test seem to be within expectations compared to other dual-tower air coolers, with Montech’s NX600 performing only 1.2 degrees C behind the best dual-tower air cooler in our setup.

Conclusion

Montech’s NX600 generally performs well, and has a very reasonable price of only $29.90, which makes purchasing this cooler worth it even if you’re “only” interested in the fans.

On the flipside, the included E28 fans seem overpowered compared to the heatsink, resulting in high noise levels during moderate to intensive workloads. If you’re looking for a quietly running system, you should consider alternatives like SAMA’s E360 air cooler instead.

Asus’ B850-Creator Wifi Neo motherboard is designed to serve the creator community (think 3D modeling and rendering, animation, and other media production). This community tends to use a lot of bandwidth, particularly for storage when you’re dealing with large files, either via physical connections to external drives (think USB-C) or via a NAS over Ethernet. With that, Creator boards tend to focus on faster networking, storage, and USB ports to speed up workflows.

The B850-Creator Neo is a ProArt-class board for creators that tries to serve that niche, offering dual 5 GbE ports and two PCIe 5.0 x4 M.2 sockets, neither of which is often found on a $299.99 motherboard. However, it doesn’t have any USB4/40 Gbps ports for connecting external storage (unlike the more expensive X870E-Creator Wifi), so any work that requires speed from your storage will have to come from the internal connections.

Outside of that, the X850-Creator Wifi Neo has enough ports and sockets to keep most budget users happy. On top of the 160 MHz Wi-Fi 7 and dual 5 GbE ports, you get eight total USB ports, though only six Type-A ports will leave some wanting. You also get three M.2 sockets (two PCIe 5.0) and a solid last-gen flagship-class audio codec. If you need more USB ports, you can use one of the two extra PCIe expansion slots for more, and even faster ports. Aside from the hardware specifications, the ProArt uses a clean, professional design with a black PCB and black heatsinks, along with metallic gold accents that spell out the ProArt branding on the VRM heatsinks.

Performance testing went as expected and was quite similar to most other boards we tested. Overall, it performed well across most of our benchmark suite. It was competent at gaming and also above average at most tests, including rendering (as we’ll see in detail later), so there's nothing to worry about in terms of performance, no matter how you want to use this motherboard.

Below, we’ll examine the B850-Creator Neo’s performance and other features, although as you have probably guessed from the score above, it won't make it to our list of the best motherboards. But before we share test results and discuss details, here are the specifications from Asus’ website:

Specifications of the Creator Wifi Neo

Inside the Box

Asus includes a few accessories to get you started, including the typical collection of cables, antennas, and guides, but not much else, as is typical for this class of motherboard. We’ve listed the accessories inside the box below:

• (4) SATA 6Gb/s cables
• Thermal pad for M.2 22110
• ASUS WiFi Q-Antenna
• Q-connector
• (3) M.2 Q-Slide package
• (5) M.2 backplate rubber packages
• ROG stickers
• ROG screwdriver
• Quick start guide

Design of the Creator Neo

The B850-Creator Wifi Neo is decidedly a ProArt board. With its all-black, 8-layer server-grade PCB design, gold accents on the heatsinks spelling out Pro Art in some form (Pro on the chipset heatsinks and Art on the left VRM heatsink. The M.2 heatsinks cover the PCIe 5.0-capable sockets, but I wouldn't call them sufficient for extended use of hot-running 5.0 drives without throttling. Outside of that, there isn’t much bling, like RGB lighting. So if that’s a part of your vision, it would need to come from the case and fans through the onboard headers.

Overall, I like the professional appearance. And what you see, a lot of motherboard showing, is typical for the price point. I just wish at least one of the PCIe 5.0 x4 sockets included a heavier heatsink.

Starting with the top-left corner, we get a better look at the two 8-pin ProCool EPS connectors (one required) to feed the processor. Next, we get a better look at the gold-accented ProArt design on the matte-finish VRM heatsinks. Moving past the socket area, the next feature we encounter is the four DRAM slots, with the locking mechanism located only on the top. Asus lists support for DDR5-8400 with 9000-series processors and higher, and DDR5-8600 for 8000-series APUs. We didn’t have any issues running our DDR5-6000 and DDR5-7200 kits with the 9900X. But, as is typical with our test hardware, we needed to break out the APU for our DDR5-8000 kit to work. If you want to get the most out of your RAM, you can use Asus’ AEMP (Asus Enhanced Memory Profile) functionality, which detects the memory ICs in the kit and provides optimized frequency, timing, and voltage profiles that you can easily apply to increase performance.

Above the DRAM slots are the first three (of seven) 4-pin fan headers. Each header supports PWM and DC-controlled devices. Each header outputs up to 3A/36W, which is plenty of output for any pump-and-fan configuration. You can even run a custom loop off the headers without worrying about power. Control of any attached devices is handled via the BIOS or Fan Expert 4 in the Armory Crate software suite. Or you can use ProArt Creator Hub to monitor PC status, switch fan profiles, and adjust performance modes.

Continuing right and moving down the long edge, the first thing we run into is the 4-LED debug header, which lights up during the POST process. If something goes wrong during this critical time, one of the LEDs corresponding to the problem (CPU, DRAM, VGA, Boot) remains lit, indicating where the hiccup is. Moving down the same edge, we then see the first two (of three) 4-pin ARGB connectors. The ProArt Creator hub or Armory Crate controls each. Next, we see the 24-pin ATX connector for board power, a front-panel USB 3.2 Gen 2x2 (20 Gbps) connector that also supports 30W PD/QC4+ charging, and a 19-pin front-panel USB 3.2 Gen 1 (5 Gbps) header.

Power delivery on the B850-Creator Wifi Neo is plenty capable, sporting 17 phases, with 14 dedicated to Vcore. Power heads from the 8-pin EPS connectors and onto the Digi+ EPU (ASP2205) voltage regulator. From there, it moves to the 80A DrMOS MOSFETs. The 1,120 Amps available isn’t a ton, but it will handle any CPU you throw at it, including the Ryzen 9 9950X or the Ryzen 9 9950X3D2.

At the bottom of the board, on the left, are the last-gen flagship Realtek ALC1220P codec and a couple of dedicated audio capacitors. This audio solution is typical for the price point, and most users won’t have any issues with it.

Next are four PCIe slots in the middle of the board. The top full-length slot uses reinforcement, connects to the CPU, supports PCIe 5.0 x16 speeds, and uses the PCIe Slot Q-Release to simplify removing a PCIe card from the motherboard. The other reinforced full-length slot also connects via the CPU and supports PCIe 5.0 x8 speeds. If both slots are in use, they run at x8/x8. The bottom full-length slot (not reinforced) connects via the chipset and supports PCIe 4.0 x4 (but it shares lanes with M.2_3). The tiny x1 slot also connects through the chipset and runs up to PCIe 4.0 x1 speeds.

There are three M.2 sockets, with the top two (under heatsinks) connecting to the processor and supporting PCIe 5.0 x4 (128 Gbps) speeds. The last M.2 socket, M.2_3 on the right, connects through the chipset and runs up to PCIe 4.0 x4 (64 Gbps) speeds. This socket shares bandwidth with the PCIe_3 slot. If either one is populated, the other gets disabled. So keep that in mind if you need to use that slot, as it will take away an M.2 socket.

Along the right edge, we find another 4-pin fan header and two (of four) SATA ports. If you need to run RAID, the SATA ports and M.2 sockets support RAID0/1/5/10 modes.

We also included images of several chips used on the board. Asus uses mainly Realtek (audio, networking), and it's own PWM ICs.

At the bottom are several headers, including front-panel audio and USB ports. A complete list of connectivity is listed below (from L to R):

• Front panel audio
• 2-pin thermal sensor header
• COM port
• 2-pin Clear CMOS
• 3-pin ARGB header
• (3) 4-pin fan headers
• TB/USB4 header
• (2) USB 2.0 headers
• (2) SATA ports
• Front panel
• Battery

The rear I/O on the B850-Creator Neo has all the basics, but lacks native 40 Gbps USB ports, which could be a concern for creators whose files are on an external drive. Outside of that, it does come with two USB-C ports, both running at 10 Gbps (USB 3.2 Gen 1), and one supporting DP Alt Mode (for video). You get two 10 Gbps and four 480 Mbps Type-A ports for a total of six, which can be limiting for some users (including me). Additionally, there are DisplayPort and HDMI for video output, a BIOS Flashback button, dual 5 GbE ports, quick-connect Wi-Fi 7, and three 3.5mm audio jacks (no SPDIF).

Note: this board uses a ‘slower’ Wi-Fi 7 that’s limited to 2.9 Gbps. It’s still plenty fast, and faster than Wi-Fi 6, but it's worth mentioning in case you’re one of the few who actually need the fastest Wi-Fi around and have the hardware to use it.

MORE: Best Motherboards

MORE: Best Motherboard Deals

MORE: Best RAM Combo Deals

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Firmware

Asus updated the BIOS layout and for the Creator Neo, using the familiar white/tan/gold that matches the ProArt theme and the mostly black board. EZ Mode is mostly informative, with system information and temperatures on the left, and quick settings for EZ Flash, Aura RGB, the Driver Hub for easy driver installation, and more in the middle. Fan control and access to full Q-Fan functionality are below. The right block displays DRAM and storage status, and Boot Priority is self-explanatory.

Advanced mode still has the headers across the top, but the shortcuts to Q-Fan and AI OC have been moved to quick settings and other locations. The right panel holds system information, including CPU frequency and temperature, in a colorful display. The main body has all the editable functionality.

Overall, I like the aesthetic update for the BIOS. It’s easy to read, and, as usual with Asus' BIOS, frequently used items are easy to find and not buried too deeply in their menus. Anything that’s a pain to access can be added to the customizable favorites screen.

Software

Armoury Crate has Several applications exist for various functions, ranging from RGB lighting control and audio to system monitoring and overclocking. It's also worth mentioning the included software. When purchasing this Asus motherboard, you get a one-year AIDA64 license – a helpful application for stress and performance testing; Asus’ Driver Hub (get your updated drivers here); Dolby Atmos (for audio); and a custom version of Hwinfo for real-time monitoring — all are helpful applications. We’ve captured a few screenshots of the applications below.

Test System / Comparison Products

We’ve updated our test system to Windows 11 (23H2) 64-bit OS with all updates applied as of late September 2024 (this includes the Branch Prediction Optimizations for AMD). Hardware-wise, we’ve updated the RAM kits (matching our Intel test system), cooling, storage, and video card. Unless otherwise noted, we use the latest publicly available non-beta motherboard BIOS. The hardware we used is as follows:

TEST SYSTEM COMPONENTS

• CPU - AMD Ryzen 9 9900X
• Cooling - Arctic Liquid Freezer II 420
• Storage - Crucial 2TB T705 M.2 PCIe 5.0 NVMe SSD
• RAM - Kingston Fury Beast DDR5-6000 CL36 (KF560C36BBEAK2-32)
• RAM - Teamgroup T-Froce Delta DDR5-7200 CL34 (FF3D518G7200HC34ABK)
• RAM - Klevv Cras XR5 RGB DDR5-8000 (KD5AGUA80-80R380S)
• GPU - Asus TUF RTX 4080 16G
• PSU - EVGA Supernova 850W P6
• Windows 11 64-bit (24H2)
• NVIDIA Driver 561.09

Benchmark Settings

MORE: Best Motherboards

MORE: Best Motherboard Deals

MORE: Best RAM Combo Deals

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Benchmark Results

Our standard benchmarks and power tests are performed using the CPU’s stock frequencies (including any default boost/turbo) with all power-saving features enabled. We set optimized defaults in the BIOS and the memory by enabling the XMP profile. For this baseline testing, the Windows power scheme is set to Balanced (default) so the PC idles appropriately.

Synthetic Benchmarks

Synthetics offer a valuable method for evaluating a board's performance, as identical settings are expected to yield similar results. Turbo boost wattage and advanced memory timings are areas where motherboard manufacturers can still optimize for stability or performance, though, and these settings can impact specific testing scenarios.

Starting with our synthetic benchmarks, the Creator Neo was average to above average among the other tested boards. In some tests, such as 7-Zip compression, it was among the fastest. It was average elsewhere and rarely below. This is what we see from most boards—nothing to worry about so far.

Timed Applications

Timed applications also went well. It delivered the slower of the two times in LAME (9.0 seconds) but faster in Corona (41 seconds). Handbrake results were also good, running above average across both sub-tests. Nothing out of the ordinary here, either.

3D Games and 3DMark

Starting with the launch of Zen 5, we’ve updated our game tests. We’re keeping the F1 racing game but have upgraded to F1 24. We also dropped Far Cry 6 in favor of an even more popular game in Cyberpunk 2077. We run both games at 1920x1080 resolution using the Ultra preset (details listed above). Cyberpunk 2077 uses DLSS, while we left F1 24 to native resolution scaling.

The goal with these settings is to determine if there are differences in performance at the most commonly used (and CPU/system-bound) resolution with settings most people use or strive for (Ultra). We expect the difference between boards in these tests to be minor, with most falling within the margin of error. We’ve also added a minimum FPS setting, which can affect your gameplay and immersion.

In our 3DMark and game tests, the Creator Neo was average overall. It scored slightly lower on the 3DMark Speedway test but slightly faster than average on Steel Nomad. In our games, Cyberpunk 2077 results were right around average, while F1 24 was slightly above average at 124/175 FPS (minimum/average, respectively). To summarize, this board has no issues running games or any productivity/creator-related activities.

Overclocking

Over the past few CPU generations, overclocking headroom has been shrinking on both sides of the fence, while the out-of-the-box potential has increased. For overclockers, this means there’s less fun to have. For the average consumer, you’re getting the most out of the processor without manual tweaking. Today’s motherboards are more robust than ever, and they easily support power-hungry flagship-class processors; We know the hardware can handle them. There are multiple ways to extract even more performance from these processors: enabling a canned PBO setting, manually tweaking the PBO settings, or just going for an all-core overclock. Results will vary and depend on the cooling as well. In other words, your mileage may vary. Considering all the above, we will not be overclocking the CPU. However, we will try out our different memory kits to ensure they meet the specifications.

For memory testing, we start with our fastest non-clock driver kit: Klevv 32GB (2x16) DDR5-8000. Per usual on this platform, it booted to Windows but wouldn’t pass a stress test with our 9900X. The Team Group DDR5-7200 kit worked without issue. Those speeds are well past the ‘sweet spot’ for the AMD platform, and with today’s RAM prices, we doubt many people are looking at these higher speeds in the first place.

Dropping in the Ryzen 5 8600G APU, we were able to run our Klevv DDR5-8000 kit without issue, which is a trend we’ve seen since we started using the APU (the IMCs on the APUs are generally better than those on desktop processors). Again, AMD’s sweet spot is around 6000-6400 MT/s, and this board (really any X870E board we’ve tested) can handle that and then some. Asus also offers AI Overclocking for the CPU if you don’t want to adjust settings manually.

Power Consumption / VRM Temperatures

We used AIDA64’s System Stability Test with Stress CPU, FPU, cache, and Memory enabled for power testing, using the processor's peak power consumption value. The wattage reading is obtained from the wall via a Kill-A-Watt meter, capturing the entire PC (excluding the monitor). The only variable that changes is the motherboard; all other parts remain the same. We've moved to using only the stock power use/VRM temperature charts, as this section aims to ensure the power delivery can handle flagship-class processors.

Stress testing the Creator with our DDR5-7200 kit showed it to be less power-hungry than most boards. At idle, it sat around 81W and peaked at 256W. This averages out to 169W, which is again slightly lower than the average B850/X870/X870E-based boards.

VRM temperatures are good, peaking at just under 48 degrees Celsius on our sensor (internal sensor readings were off and not listed). Between the capable power delivery and large heatsinks, you won’t have any trouble running flagship-class processors or even overclocking.

Bottom Line

Asus’ B850-Creator Wifi Neo is an OK budget motherboard for Creators. Priced at $299.99, you get dual 5 GbE ports, three display output options, and several of Asus’ EZ DIY and AI Features that all help to make this a viable option for the budget-minded creator. That said, you are lacking in a couple of areas compared to the X870E version of the board, which offers faster USB (40 Gbps USB4 ports) and more USB ports. If you’re hung up on the port count, the next Asus Creator option costs almost $250 more ($549.99), and at that point, there are plenty of other options that would work for less.

One example, ASRock’s X870 Taichi Creator ($319.99), is arguably a better option in the creator space and a direct competitor to the Asus B850 Creator. For $20 more, you get 10GbE and 5 GbE, dual USB4 (40 Gbps) ports, and more USB Type-A ports, along with even more robust power delivery. If this isn't the board for you, check out our best motherboards article to see if one of our best picks are right for you.

MORE: Best Motherboards

MORE: Best Motherboard Deals

MORE: Best RAM Combo Deals

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Bambu Lab has finally updated its popular A1 bedslinger with an increased build volume, a few technical tweaks, and additional capabilities. The A2L has all the room of a more expensive H2 Series, but at a fraction of the cost, at $569 for a four-color combo. Its open frame design makes this primarily a machine for PLA, PETG and TPU, which happen to be the most popular filaments for its target audience: family enthusiasts.

Bambu Lab has categorized this machine as a family-friendly craft center. When combined with its gold standard AMS Lite and the H2 Series blade cutter attachment, the A2L can whip up practical prints, toys, and vinyl iron-ons for apparel. The launch coincided with Bambu’s new PLA Pure, a new non-toxic PLA formula with low VOCs, and a pastel matte finish that hides layer lines.

Its one flaw is the insane vibrations generated by the A2L at speed. Normally, larger machines will be slowed down to avoid this, but Bambu Lab decided to toss granulated vibration-absorbing pellets in the frame and throw caution out the door. This makes the machine nearly as fast as the original A1 while still printing error-free. It also makes it a bit of a table snob, as most cheap tables can not hold it. I put the A2L on an IKEA Lack coffee table, a pretty standard surface for budget-minded makers, and the AMS Lite vibrated right onto the floor after 45 minutes of printing. If the table had been any higher (like a card table or wobbly kitchen table), it would have taken the whole printer onto the floor with it. Bambu has included a “steady” printing profile that quells the shaking at the expense of speed. A “high-quality” profile also runs at a less frantic pace.

The A2L retails at $469 for the standalone printer, $569 for a combo, with an additional $69 for a cutting upgrade kit. The printer is also compatible out of the box with the AMS2 Pro ($269), which would give you the ability to dry filament without needing additional equipment. Bambu Lab has really perfected its formula for quality, speed, and ease of use, making this one of the best 3D printers we’ve seen this year.

Specifications: Bambu Lab A2L

Included in the box: Bambu Lab A2L

Our A2L unit came with an AMS Lite and an optional blade cutter/pen plotter attachment. A cardboard toolbox holds all the tools needed to assemble and maintain the printer, plus a blade for making your own scraper. There was also a small coil for running a test print (enough to make that scraper).

Assembling the Bambu Lab A2L

The Bambu Lab A2L requires a small amount of assembly. First you need to attach the heatbed to the Y-carriage, then place the base into the gantry frame, which is a pretty beefy triangular unit. There’s a handful of screws to hold everything together, and a few wires to plug in. The instructions are very easy to follow.

Do apply lubricant to the Y-Axis while you have it exposed. It does not come pre-greased, and our test unit immediately requested grease after only a few prints.

The AMS unit snaps together quickly before being secured onto its stand.

Leveling the Bambu Lab A2L

The Bambu Lab A2L completely auto-levels itself during initial calibration, then can recheck the level before each print. It uses the nozzle as the probe, so it can also set the z height for you. I had no problems with the settings and no need to make adjustments.

Though calibration errors are rare, Bambu Lab has a section in its Wiki to guide you through troubleshooting advice.

Loading Filament in the Bambu Lab A2L

Like the Bambu Lab A1, the A2L can work from a single spool holder mounted to the gantry, or from four spools on the Lite AMS unit. In addition, the A2L is also out of the box compatible with the box style AMS and AMS 2 Pro.

For a single spool you must push filament through the Bowden tube to the toolhead, then tap “load filament” on the main menu. The machine will take over from there and display a checklist on the screen so you can follow along.

For the AMS Lite, simply push spools onto the motorized spindles and poke the filament into the intake. The machine will detect the filament and draw it in.

The AMS has an RFID reader, which allows it to read tags on Bambu Lab branded filament. These tags will tell the machine what type and color of filament is on each spindle. If you’re using untagged filament, or the single spool holder, you’ll need to enter this information manually.

Design of the Bambu Lab A2L

The Bambu Lab A2L is a plus sized Cartesian, or as we call them, bed slinger. The A-Series sleek white and brushed aluminum style has been copied by most of the competition, so Bambu Lab seems to be onto something.

The printer is extremely easy to use with a touch screen interface that contains guides to show you how to run the machine. It has the same build size as the H2 Series, but does not come with an option for an enclosure.

It has metal rails, linear bearings, and dual Z rods, all tucked inside a smooth frame and base that makes it look much more polished than other printers. To make up for its large size, it has a pair of hidden “granular dampers” designed to absorb vibration and stabilize the frame. You can’t see them, but you can hear them if you give the printer a bit of a shake.

The A2L has a new feature called “adaptive vibration compensation,” which uses a real-time algorithm to retune calibration as your print gets taller. This is to eliminate vibrations that can affect printers the taller – or heavier – they get.

When equipped with an AMS unit, the A2L takes up noticeably more space. Bambu Lab offered a clip that I could print myself that would attach the AMS to the X gantry, but I’m still not convinced that is the best place for it.

The extruder motor has been upgraded to a PMSM closed-loop servo to actively detect filament grinding in the gears, clogs and air printing. Combined with the new nozzle clumping detector on the purge sweep, the A2L should be able stop a lot of annoying small problems before they become huge spaghetti monsters.

One odd thing is the A2L uses the same quick swap steel nozzles as the A1 and A1 Mini. These are not the upgraded nozzles used on the H2 Series, but H-style nozzles are backward compatible with the A-Series. These nozzles have made a huge impact on the industry, with the competition scrambling to copy them as well. There’s no screws, wiring or messy thermal paste involved here.

The A2L is compatible with a cutting and plotting attachment, which is available as a separate purchase. The add-on is easy to install: simply remove the front cover and the spinner, then snap on the cutting attachment and plug a cable into the toolhead. It is advised to remove the bowden tubes for the filament, but there’s little need to do so. The hotend remains in place and is kept off by the machine.

The attachment can hold either a Bambu Lab made drag knife or any standard size pen or marker. I was able to use a normal ball point pen in mine by adding a few wraps of tape to thicken the pen body.

This is the same attachment you can get on the H2 Series and uses Bambu Suite, a companion app, to “slice” files. There is no laser add-on, since the A2L is open frame. The A2L also lacks a top down camera for positioning, so you will need to use the cutting function with the help of a phone camera equipped with Bambu Handy. This system is a bit janky, but it works, especially if you take care lining up the photos.

Bambu Lab Automatic Material System (AMS) Lite

We received the A2L Combo, which includes an AMS Lite module. The AMS Lite is made specifically for the A-Series and is not compatible with the Core XY machines. It’s not as clean looking as the boxy AMS, with four Bowden tubes stretching from each spool to the tool head. However, this simplified version is much easier to unjam should brittle filament break in the tube.

Each spool is mounted on a spring-loaded spindle, which provides the necessary retraction. Since each color has its own Bowden tube, the A2L only needs to pull the filament a few inches until it's clear of the tool head, not all the way back to the AMS Unit, which makes it a bit faster during color swaps than the original system. Each spool also gets a motor to push filament to the tool head. This new system solves the problem of cardboard spools with crushed edges, spools that are too wide or too tall, and spools that are too light.

There are notable drawbacks to using either of Bambu Lab’s AMS systems. The printer wastes an amazing amount of material when it cleans out the nozzle between colors. The waste can be reduced with tuning, but takes time and patience. The slicer will tell you exactly how much it will use before you start a print.

The second drawback is loss of speed. The printer needs to pause between layers to purge filament, then do a nozzle wipe on the purge tower. The Flexi Factory Panda takes 2 hours and 40 minutes to print in one color using default settings, and 6 hours and 46 minutes to print in three colors.

Preparing Files / Software

Bambu Lab has its own custom slicer, Bambu Studio, which is a fork of PrusaSlicer. The slicer is also the primary means of transferring files via Cloud service or local LAN to the printer. You can also control and monitor your printer remotely with Bambu Handy, a mobile app.

Bambu Studio is also the “paint” program you need to assign color to your prints. The device tab allows you to talk directly to the printer to define the filaments in the AMS, heat up the printer and control the speed.

It has a wide range of presets for filament from both Bambu Lab and other manufacturers. I’ve found these presets to be incredibly helpful and accurate.

You also use Bambu Studio to access MakerWorld, a one stop shop for the Bambu Ecosystem. Here you’ll find free files for your printer, MakerLab with AI powered design tools and Maker’s Supply, a store front will model kits, hardware, tools, LED lights and CyberBricks so you can level up your 3D printing hobby.

Printing on the Bambu Lab A2L

The printer arrived with several prints preloaded. You’ll definitely want to shop for filament before bringing this machine home, especially if you plan on doing a lot of AMS printing. We have a guide with our favorite filaments for 3D printing here.

To show what the Bambu Lab A1 can do under normal conditions, I printed three Ice Cream Narwhales using default settings of 200-300mm/s and a .2mm layer height, taking 10 hours and 44 minutes to print. The models came pre-painted by the designer.

It turned out beautifully smooth with no visible layer lines or ringing. There was a little color bleed from the gold cone to the top of the narwhale’s head. This was printed in a sparkle blue from an unmarked PLA spool I picked up from SMRRF a couple years ago, and now deeply regret not labeling it better, plus a basic PLA Pink and Black from Bambu Lab and gold from Polymaker’s Panchroma Dual Silk Sunset.

For a PETG test, I attempted to print a wind spinner for the garden, which I created in Tinkercad. The print ultimately failed due to the skinny base, but it got about 95% there, which is still impressive. When it fell over, it looped the rest of the layers around the print like a coil. If you look closely at the coil, you can see little bubbles caused by moisture. The print still looks fantastic and I’ll give it another shot later after I make the base wider. This is printed in Bambu Lab PETG Translucent 6 hours and 31 minutes using a .2 layer height and default settings. This print is smooth, without stringing or noticeable layer lines. The two flat bases printed separately – without supports– then snapped together.

For TPU testing, I printed a prototype phone bag that I created in Tinkercad, using CookieCad Pale Blue Elixir. This is a regular 95A TPU, so I needed to use the single spool holder on top and drop it directly into the extruder. The A2L did a great job, with smooth layers and only a couple stray bits from retraction errors at the very top. The print took 7 hours and 9 minutes using the printer’s defaults.

Reminder – the AMS can print special “TPU for AMS” which is a stiffer formula available from Bambu Lab.

I tested out the A2L’s cutting module by making a few branded price tags for items I sell at a local boutique. For this I used a simple shape I whipped up in Canva and my jpg logo. I dropped both into Bambu Suite, the “slicer” software that runs Bambu’s 2D cutter attachment and also the laser module for a H2 Series machine. (No, there’s not a laser for the A2L.)

I swapped the normal build plate for a purpose built sticky plate and very easily snapped the cutter module onto the tool head, then plugged it in. Because the A2L doesn’t have an overhead “birdeye” camera like an H2 Series, I had to take an overhead photo of the build plate with the Bambu Handy app. It’s surprisingly difficult to get the entire A2L bed into an overhead shot when you’re only 5’2” without getting a ladder. This might be why the registration is a little bit off and the inked logos are a little bit lower than they should be.

This required two passes. First the ink layer, which used a normal ball point pen that I wrapped tape around to fit in the pen holder. Then I swapped the pen for the cutting blade and finished it.

The cutter can also handle vinyl, and with the A2L bed size, you could really get a nice sized shirt made. Bambu didn’t send any vinyl to try out, but they do sell it on their website. Because this is a manual cutter and not a laser melting material, you are free to use any material that is under .5mm thick.

Bottom Line

The Bambu Lab A2L is the supersized evolution of the A1 bedslinger that does double duty as a family craft hub. It’s budget-friendly with an easy-to-use interface that lets beginners have a little extra room to play without having to move up to a price H2 Series.

Like the first A1, this is much easier to use and maintain than Bambu’s larger machines, as the AMS Lite isn’t prone to jamming, and the new nozzle clumping detector sweeps printer purge safely away.

At $569 for the combo and $469 for a single color machine, the A2L is a great choice for makers who want Bambu quality and access to the ecosystem without the price of a Core XY machine.

If you’re interested in a larger color bedslinger, the Anycubic Kobra 3 Max has a 420x420x500mm build volume and is currently on sale for $601.60. The Bambu Lab H2S has the same build size in a Core XY enclosed unit for $1349.

AMD answered the demands of gamers and re-released the Ryzen 7 5800X3D, though not without compromise. Although the return of Zen 3 X3D has been a good idea for months, given the limited time we saw those chips on the market, this re-release comes with a surprisingly high price, considering the silicon and how it compares to the best CPUs for gaming.

Price is the biggest issue for the Ryzen 7 5800X3D. AMD shaved $100 off the original MSRP for the 10th Anniversary Edition re-release, but that puts it in very competitive waters, even considering current RAM prices. The CPU is flanked on one side by the Core i7-14700K that also supports DDR4 memory, and on the other by the Ryzen 5 7600X3D, which offers superior gaming performance and a lower price to offset the cost of a DDR5 platform.

The chip mainly appeals to those who already have an AM4 motherboard and memory to go with it, and who were unfortunate enough to miss the small window when you could buy the Ryzen 7 5800X3D a few years ago. In that situation, just about any price is a deal compared to the competition.

Otherwise, the Ryzen 7 5800X3D is about $70 to $100 too expensive, and even that lower price would be questionable if DDR5 prices weren’t out of control. Although the chip has earned its legendary status among gamers, revisiting it in 2026 shows clearly that it maxes out what DDR4 platforms are capable of in games, and it falls far too short of the DDR4 competition in applications.

The island of AM4 users stranded without a clear upgrade path will love the 5800X3D re-release. But the chip is not nearly as impressive as it once was if you have to buy a motherboard and/or RAM alongside your CPU, however.

Some notes on this re-review

We don’t normally re-review products here at Tom’s Hardware, much less update existing reviews outside of some extraordinary circumstance. We will follow up reviews with additional coverage as needed, but our reviews are as much buying advice at the time they’re written as they are historical context years down the road. Reviews exist in the context in which they’re written.

That’s important because, especially with PC hardware, some good products can become worse over time and bad products can become good over time. Even in this past generation, AMD had several stumbles with Zen 5, which it addressed post-launch through a combination of firmware updates and exposing additional settings in the BIOS. Intel had some major regressions in performance with Arrow Lake, which it partially addressed after release with Core 200S Boost.

These products are better now than they were at launch, but it’s still important to know that they had issues at launch. That’s the function of our reviews. They’re a snapshot of how a particular component performs and compares to the rest of the market at a certain point in time. Our list of the best CPUs for gaming and CPU benchmark hierarchy pagesare where you’ll find the consistent updates on which chips are best at any given time.

That preamble is to say that this re-review of the Ryzen 7 5800X3D does not replace our original review, which is why this is a separate piece of content and not merely an update. We’re re-reviewing the chip because AMD is re-releasing it, and we need to compare the chip to the current market it exists in.

That market includes high memory prices, which is a driving force behind the re-release of the Ryzen 7 5800X3D in the first place. We’re paying especially close attention to memory in this review, both in terms of price and performance. However, we’ve also brought some price-competitive DDR5 chips into the mix, including some of AMD’s own CPUs.

Finally, we’re reviewing the original Ryzen 7 5800X3D here. AMD says that the new 10th Anniversary Edition should be identical to the original model, but it’s using a slightly different bonding process, which could have a minor impact on power and thermals, in particular. We’ll be getting a 10th Anniversary Edition into the lab in order to find out, but we don’t expect major performance differences between the original and re-release versions.

AMD Ryzen 7 5800X3D specifications and pricing

It’s difficult to evaluate the specs of the Ryzen 7 5800X3D given the current market, so this is a refresher of what the processor offers and how it compares to some of the current options featured in our test suite. It’s an eight-core / 16-thread chip sporting AMD’s Zen 3 architecture, and it boosts up to 4.5 GHz, with a base clock of 3.4 GHz.

The chip is fabricated on TSMC’s 7nm FinFET process, with GlobalFoundries stepping in to fab the I/O die on its 12nm process. Of course, the main draw of the CPU is the 64MB chunk of SRAM that’s bonded to the compute die, giving the processor access to a total of 96MB of L3 cache.

In recent years, we’ve seen both AMD and Intel increase cache sizes broadly, not just on X3D CPUs. For instance, the Ryzen 7 9700X has the same 32MB of on-board L3 that we can see all the way back to Zen 3, but it has double the L2 cache. Intel has traditionally split L2 and L3 more evenly, and we’ve seen an increase in both with Arrow Lake and Arrow Lake Refresh.

Still, the huge boost in L3 helps a lot here. It comes with some thermal trade-offs, however. Although the Ryzen 7 5800X3D is a very efficient CPU, it also has careful power management. The SRAM sits on top of the compute die, insulating the cores from the IHS. This thermal design means the Ryzen 7 5800X3D has relatively low peak clock speeds out of the box, and it doesn’t officially support AMD’s Precision Boost Overdrive.

AMD has addressed that issue in newer X3D chips, riding the efficiency of Zen 4 with the Ryzen 7 7800X3D and moving to a new bonding process that situates the SRAM below the compute die with the Ryzen 7 9800X3D.

• MORE: Best CPU for gaming
• MORE: CPU Benchmark Hierarchy
• MORE: Intel vs AMD
• MORE: How to Overclock a CPU

Although the Ryzen 7 5800X3D was a revelation when it first released, it’s important to remember that it wasn’t leagues faster than Intel’s competing Alder Lake chips, at least not on the level of the 30%+ delta we see today with Arrow Lake and the Ryzen 7 9800X3D. The pedigree that 3D V-Cache has built comes in part from the newer X3D chips, and that’s clear when looking back at the Ryzen 7 5800X3D.

Even more clear is the split between DDR4 and DDR5. Now that we have Raptor Lake (and Refresh) as a comparison point, the Ryzen 7 5800X3D positions itself as the peak of what DDR4 platforms are capable of in games. It’s marginally faster than the Core i7-13700K and Core i7-14700K with DDR4 memory, 17% ahead of the Core i7-12700K with DDR4, and even 4.5% ahead of the Core i7-12700K with DDR5.

This wall that you can see, around 145 fps in our geomean, directly translates into a handful of the games we tested. Especially among the newer titles in our suite, simply moving to DDR5 memory results in more than a 31% increase in performance on the same CPU. That led to a handful of situations where both Raptor Lake CPUs perform worse than the Ryzen 7 5800X3D with DDR4, but offer double-digit improvements with DDR5.

Based on our RAM price tracker, a 32GB kit of DDR5-6000 runs between $400 and $450 currently, while a 32GB kit of DDR4-3200 will run you between $200 and $250. There are plenty of exceptions, for better and worse, but we’re going to call the price difference between DDR4 and DDR5, right now, about $200.

Establishing that number is important because of one CPU: AMD’s own Ryzen 5 7600X3D. It’s on the AM5 platform and requires DDR5, but it’s also $230, $120 less than what AMD is re-releasing the Ryzen 7 5800X3D at. Assuming our lowest RAM prices, that means the Ryzen 5 7600X3D is around 14% more expensive than the Ryzen 7 5800X3D when memory is brought into the price. But the Ryzen 5 7600X3D is also 18% more performant.

That’s the biggest hurdle standing in the way of the Ryzen 7 5800X3D. Even if you already have DDR4 memory, the Ryzen 5 7600X3D and 16GB of DDR5-6000 is only around $80 more expensive, and much more performant. Plus, it gets you on an AM5 platform, setting up cheaper future upgrades (AMD says AM5 will receive support through at least 2029).

In Intel’s camp, the two Raptor Lake chips with DDR4 run up against a similar wall as the Ryzen 7 5800X3D, but offer around a 15% jump with DDR5. The Ryzen 7 5800X3D still makes sense if you already have an AM4 board. However, if you have to buy a motherboard, the Raptor Lake chips offer similar gaming performance with DDR4 and much better application performance, which we’ll get to next.

Elsewhere, there aren’t a lot of surprises. The Ryzen 7 5800X3D is as efficient as ever, drawing just 77.5W on average in our testing. AMD has pushed efficiency even further now, but it’s remarkable to see the Ryzen 7 5800X3D offering similar performance as the Core i7-14700K with DDR4, while consuming half the power.

We also have our value geomean here, which is deceptive. Obviously the price of memory is a huge influence here, not only on total platform cost, but also on performance. The value geomean here just represents a true CPU-to-CPU comparison of value, devoid of RAM context.

007 First Light Benchmarks

The first game in our suite is the newest, which is 007 First Light. The Ryzen 7 5800X3D surprisingly struggles in this title, which is strange given how well IO Interactive’s previous title, Hitman 3, took to 3D V-Cache CPUs. Still, the Raptor Lake chips are ahead here, even with DDR4, and they claim top slots with DDR5.

Baldur’s Gate 3 Benchmarks

Baldur’s Gate 3 favors X3D chips, which is clear based on the fact that the Ryzen 7 5800X3D is just a touch behind the much newer Ryzen 7 9700X. The 5800X3D is also around 11% faster than the Raptor Lake chips with DDR4 memory. However, the Raptor Lake chips with DDR5 are about 20% faster than the Ryzen 7 5800X3D, while the Ryzen 5 7600X3D is around 35% faster.

Crimson Desert Benchmarks

In Crimson Desert, the 5800X3D outpaces the Raptor Lake competition by a few frames, but this game clearly favors faster memory. It also scales oddly well on Raptor Lake chips, as evidenced by the fact that the Core i7-14700K outclasses the Core Ultra 7 270K Plus.

Counter-Strike 2 Benchmarks

Counter-Strike 2 is more competitive, with only the Ryzen 7 7800X3D offering a clear lead above the rest of the pack. The Ryzen 7 5800X3D is marginally ahead of the Raptor Lake chips with DDR4 here, and even moving to DDR5 doesn’t offer a significant improvement.

Cyberpunk 2077 Benchmarks

Cyberpunk 2077 is another clear example of the memory divide that’s growing between DDR4 and DDR5. The Ryzen 7 5800X3D tops the DDR4 rankings by a few frames, but moving to DDR5 on Raptor Lake offers roughly a 15% performance jump.

Doom: The Dark Ages Benchmarks

When we brought up a 31% gap in DDR4 and DDR5 performance earlier, we were referring to Doom: The Dark Ages. This is a fairly recent game that’s clearly designed with DDR5 in mind. The Ryzen 7 5800X3D does surprisingly well, though, offering a 10.8% jump over the Raptor Lake competition with DDR4.

F1 24 Benchmarks

F1 2024 slants heavily toward AMD processors, and it’s a game that scales well with 3D V-Cache. Here, the Ryzen 7 5800X3D is around 5% faster than the Core i7-14700K with DDR5, and 13% faster with DDR4. The Ryzen 5 7600X3D spoils the fun a bit, though, offering a solid 10% jump over the 5800X3D.

Far Cry 6 Benchmarks

Final Fantasy XIV Benchmarks

Final Fantasy XIV scales well with X3D chips, with the three 3D V-Cache CPUs in our test pool topping the rankings. The 5800X3D takes a clear backseat to the DDR5 CPUs, but it still manages a solid 6.6% improvement over the Core i7-14700K with DDR5 and a 16% jump with DDR4.

Flight Simulator 2024 Benchmarks

Flight Simulator 24 leans back toward Intel, with the 5800X3D only managing to outclass the Core i7-12700K with DDR4. In this game, the Ryzen 5 7600X3D is around 18% faster, while the Core i7-14700K is 11% faster, even with DDR4.

Hogwarts Legacy Benchmarks

We can see a similar situation in Hogwarts Legacy, with the Core i7-14700K paired with DDR4 outpacing the Ryzen 7 5800X3D by about 12%. The Ryzen 5 7600X3D, meanwhile, is more than 20% ahead, while the Core i7-14700K with DDR5 marks a 25% lead.

Marvel Rivals Benchmarks

Marvel Rivals is an Unreal Engine 5 game, and it’s mostly bound by the GPU, which explains the stair-step pattern you can see in our data. We can see a clear divide between DDR4 and DDR5 platforms here, with even the Ryzen 5 7600X offering superior performance.

Minecraft RTX Benchmarks

Intel struggles in Minecraft, while X3D chips run away with performance. The Ryzen 7 5800X3D ends up in the middle of the pack, offering a clear buff of around 18% over the Raptor Lake competition but falling short of even AMD’s weaker DDR5 options.

Spider-Man 2 Benchmarks

Starfield Benchmarks

The Last of Us Part One Benchmarks

• MORE: Best CPU for gaming
• MORE: CPU Benchmark Hierarchy
• MORE: Intel vs AMD
• MORE: How to Overclock a CPU

The Ryzen 7 5800X3D was a weak CPU for productivity when it was released, and it hasn’t aged particularly well since then on that front. The clock speed is limited compared to non-X3D Zen 3 chips. Add to that the limitations of DDR4 platforms more broadly in non-gaming applications, and it's clear the 5800X3D was never destined for high marks here.

Even compared to the Ryzen 5 7600X3D, the 5800X3D is only 4.6% faster, which is bad considering the latter CPU has access to 33% more threads. Intel’s hybrid architecture boosts core counts, allowing Team Blue to dominate the top of our multithreaded performance rankings.

The CPU to call out here is the Core i7-14700K, particularly paired with DDR4 memory. It’s more than twice as fast as the 5800X3D in our multithreaded geomean, while offering similar gaming performance. Again, we can see this dilemma for the 5800X3D, where the 14700K is a better all-around CPU with DDR4, while the Ryzen 5 7600X3D is a superior DDR5 CPU for not much more money (even considering the price of DDR5).

Turning to single-threaded performance, the playing field levels out, short of the 5800X3D, which even loses out to the Ryzen 7 5700X due to its limited clock speed (the 5700X boosts 100 MHz higher).

Rendering Benchmarks

Rendering apps factor heavily in our overall multithreaded geomean, so the individual results here largely mirror what you can see in the geomean above. In Cinebench 2024, the Core i7-14700K with DDR4 still offers more than double the performance of the Ryzen 7 5800X3D, and it pushes higher with DDR5. We can see a similar situation in the newer Cinebench 2026.

Single-core results are especially weak considering the Ryzen 7 5800X3D’s limited boost potential. DDR5 isn’t much of a factor here. Although we can see some scaling on the Raptor Lake platforms when moving to DDR5, the Ryzen 7 5800X3D is limited in rendering apps, even by DDR4 standards.

Blender shows similar disparities, with the six-core Ryzen 5 7600X3D largely matching the Ryzen 7 5800X3D, despite sporting a 33% reduction in thread count.

The Ryzen 7 5800X3D is a gaming CPU, pure and simple, so we never expected any miracles here. However, the Ryzen 5 7600X3D is offering comparable performance on a DDR5 platform for less money, while shooting ahead in games. And the Core i7-14700K offers much better productivity performance and comparable gaming performance, even when paired with DDR4.

Encoding Benchmarks

Encoding is similarly a major factor in our geomean, though with a more even split between lightly- and heavily-threaded workloads. Starting with the latter, Handbrake is an all-out workload for any consumer CPU that leans on high core counts, power limits, and clock speeds. With an x265 10-bit encode, even the Core i7-12700K with DDR4 is 39% ahead of the 5800X3D, while the Core i7-14700K with DDR4 shoots ahead, scoring 82% lead. AV1 and X264 show similar gaps.

The LAME audio encoder provides a closer look at single-threaded encoding workloads, and once again, the Ryzen 7 5800X3D struggles. It was the worst performer in the test pool running a standard LAME encode, and the gap between it and the more performant options in our test pool only grows in the extended run.

Creator App Benchmarks

Perhaps the biggest area of crosstalk with gaming is creator applications. The Ryzen 7 5800X3D is a bit more impressive here compared to our rendering and encoding workloads, often matching the Core i7-12700K with DDR4 memory. Still, it ends up at the bottom of our rankings, with newer Raptor Lake and DDR5 CPUs offering better performance.

Photoshop is an app that leans toward AMD chips, though that’s mainly with newer Zen 5 chips. For the 5800X3D, the Core i7-14700K offers a 10% performance jump with DDR4. Things are a bit more competitive in Premiere Pro, with the limitation of DDR4 platforms becoming abundantly clear in our data.

Outside of the Adobe suite, we have DaVinci Resolve, which isn’t as lopsided as Photoshop with recent AMD CPUs. Here, even DDR4 Raptor Lake options are near the top of the chart. The 5800X3D secures a minor win over the 7600X3D and 5700X, but not by a meaningful margin.

Finally, After Effects offers a look at VFX performance. AMD’s 5800X3D especially struggles here, with the Core i7-12700K with DDR4 offering a 13% boost in overall performance. With DDR5, that lead jumps to 22.9%.

Web and Office Benchmarks

In lighter web and office workloads, the Ryzen 7 5800X3D still struggles. Most basic productivity and web applications are lightly-threaded, so it makes sense to see the 5800X3D would fall below the marks we can see in creator apps. Across all of the workloads here, the 5800X3D is either at the bottom of the pile or just narrowly outclasses the 5700X.

That doesn’t mean you’ll run into major performance issues. The workloads here are relatively light, and they’re suitable for far weaker CPUs than the Ryzen 7 5800X3D. We’re mainly looking at general application performance for browsers and lightly-threaded apps here, which is an area where the 5800X3D struggles.

Chess Engines, Compilation, Compression, AVX, and Other Benchmarks

Finally, we have a range of benchmarks examining code compilation, chess engines, database workloads, and far more. The 5800X3D is a gaming CPU, and these workloads are highly targeted at specific applications. There are some workloads where the 5800X3D shines, such as the ebizzy web server test, SQLite database test, and Linpack. However, it mostly lands at the bottom of the pile when looking at workloads more broadly.

We’ve certainly seen a benefit from a larger L3 cache for specialized workloads (see our Ryzen 9 9950X3D2 review for more), but that’s not the draw here. The Ryzen 7 5800X3D is weak in most productivity apps, so unless you plan on running a server and somehow gaming on the same machine, I wouldn’t weigh these results too heavily.

• MORE: Best CPU for gaming
• MORE: CPU Benchmark Hierarchy
• MORE: Intel vs AMD
• MORE: How to Overclock a CPU

AMD’s X3D CPUs are known for their efficiency, and although those efficiency gains mainly show up with the Ryzen 7 7800X3D, the 5800X3D is no slouch. Power is carefully limited on the chip, with it peaking at just 119W in our y-cruncher test. In the same test, the Core i7-14700K drew 335W.

Y-cruncher leverages AVX instructions, pushing the Ryzen 7 5800X3D to consume more power compared to the 7800X3D. In an all-out, non-AVX workload like Cinebench 2024, the two CPUs are in lockstep. Again, we can see the Raptor Lake competition dancing with 300W across configurations.

Blender and Handbrake tell a similar story, but efficiency is what’s important here, given how broad the performance window is for the chips in our test pool (particularly in applications). The 5800X3D is much more efficient than the 13700K and 14700K. The margins are also narrower than raw power consumption.

The efficiency picture is clearer with a scatterplot, where we can see a clear separation between the AMD and Intel chips in our test pool. Short of the Core Ultra 270K Plus, there’s a compromise. AMD’s options are much less performant but consume far less power, while Intel's options are generally the opposite.

One upside of the 5800X3D, in particular, is its low idle power consumption. With Zen 4 and Zen 5, we saw a massive increase in power consumption in both idle and active idle (YouTube playback) scenarios. With both the Zen 3 chips in our test pool, we can see a return to single-digit power draw under idle circumstances.

Test Setup

We try to minimize the differences between our test platforms to keep our results consistent, which is especially important here given that we used so many different platforms. Even with big differences, we used the same OS image that’s specifically tailored for testing and frozen to avoid updates skewing our results as we test.

In addition to standardising our OS, we standardise our BIOS settings. We test with XMP/EXPO enabled on memory kits that we’ve validated for stability on the platforms we use. We also disable VBS in the BIOS and turn on ReBAR.

Modern AMD and Intel CPUs come with sophisticated boosting algorithms, but they aren’t always covered by warranty. AMD doesn’t cover PBO, for instance, and Intel doesn’t warranty the “Extreme” power profile that’s common on motherboards. Because of the lack of warranty coverage, we test with PBO disabled and Intel’s power profile set at its default settings.

• MORE: Best CPU for gaming
• MORE: CPU Benchmark Hierarchy
• MORE: Intel vs AMD
• MORE: How to Overclock a CPU

At $350, the Ryzen 7 5800X3D really only makes sense if you already have an AM4 motherboard and DDR4 memory to go along with it. That’s not a small audience, especially considering the relatively short time the original Ryzen 7 5800X3D was available on the market. But if you need to buy a motherboard and/or memory for your upgrade, there are better options at this price.

The 5800X3D re-release is a victim of poor pricing. The chip taps out DDR4 platforms in gaming performance, but we’re closing in on half a decade with mainstream DDR5 platforms, and we’ve seen much more powerful CPUs in that time. And yet, the Ryzen 7 5800X3D’s price hasn’t moved significantly away from its original $450 MSRP.

If RAM prices weren’t out of control, a $350 price tag on the Ryzen 7 5800X3D would look insane given the current options around that price. As I’m writing this, the Ryzen 7 7800X3D is on sale for less than $350. And even considering current RAM prices, the Core i7-14700K and Ryzen 5 7600X3D are compelling alternatives.

On the productivity front, the Core i7-14700K is a clear winner, even with DDR4 memory. There are some titles where the Ryzen 7 5800X3D is faster than the Core i7-14700K is when paired with DDR4. But overall, they offer similar gaming performance. Especially now that AMD has turned its attention to AM5, the Core i7-14700K is a better all-around option if you don’t already have an AM4 motherboard.

Although AMD almost always comes out on top with platform longevity, Intel is surprisingly ahead in that regard in the Ryzen 7 5800X3D versus Core i7-14700K matchup. Intel is reportedly set to launch Raptor Lake Next on the LGA 1700 socket next year, though I suspect any performance benefits that come along with it will require an upgrade to DDR5.

On the gaming front, the Ryzen 5 7600X3D does a lot of work. It requires DDR5, but that’s partially offset by the fact that it’s $120 cheaper than the 5800X3D. Factoring in RAM or not, spending a bit more on the Ryzen 5 7600X3D not only nets you much higher performance in games, it also gives you an AM5 platform that you can easily upgrade in the future.

That doesn’t discredit the island that AM4 users are currently on. The Ryzen 7 5800X3D left the market fairly quickly, and given the popularity of AM4, it stands to reason that there’s a large group of people for whom the 5800X3D is a significant upgrade. For that group, it’s the CPU of legend you’ve heard so much about. But if you’re planning on buying a motherboard and/or RAM with your new CPU, shop around a bit more.

• MORE: Best CPU for gaming
• MORE: CPU Benchmark Hierarchy
• MORE: Intel vs AMD
• MORE: How to Overclock a CPU

Everyone wants the freedom of a great wireless gaming headset, but it's hard to find one that's budget-friendly without compromising on... well, a lot. HyperX's new Cloud Stinger 3 Wireless is a budget-friendly (ish) wireless headset that doesn't compromise on much — it's got large, 50mm dynamic drivers with a frequency response range of 10 - 50,000 Hz, an ultra lightweight, comfortable frame with reinforced metal yokes for durability, and it gets up to 80 hours of battery life on a single charge, which is a number we usually only see in much pricier products. The Cloud Stinger 3 Wireless comes in at just under $100, which isn't nothing, but is significantly cheaper than the $200 - $300 price range of headsets with similar specs.

Design and Comfort of the Cloud Stinger 3 Wireless

The Cloud Stinger 3 Wireless is an over-ear (circumaural) wireless gaming headset with a lightweight plastic frame and a non-detachable flip-to-mute boom mic. It's mostly made of plastic, with a matte black finish and some machined texturing around the outer edges of the large, oval earcups, but it does have stainless steel sliders in the headband for durability. The earcups are independently height-adjustable, and they also tilt and swivel 90 degrees for flat storage (though the headset doesn't come with a travel bag or case). The earpads and the underside of the headband are padded with soft, leatherette-covered memory foam, which does an excellent job of passive noise cancellation.

It looks like a gaming headset, mostly because it has a non-detachable boom mic. HyperX's HX logo is glossily debossed on both earcups, while the HyperX text logo adorns the top of the headband. It's not unattractive, but it does look a little cheap — I blame the matte black plastic; it's hard to have this particular finish without looking like you're cutting corners at least somewhat. It feels relatively sturdy for its weight, and the flexible frame makes me think it would hold up decently well if you take care of it. The leatherette-covered memory foam padding feels more premium than the rest of the headset, so that's a nice touch.

The Cloud Stinger 3 Wireless is expectedly lightweight, weighing in at just 10.1 ounces (286.5g). It's almost two ounces lighter than the extremely comfortable HyperX Cloud III S Wireless (12.05oz / 341.5g), and it's half an ounce lighter than the Turtle Beach Atlas Air (10.61oz / 301g), which is one of the lightest wireless headsets we've tested. It's still heavier than wired lightweight headsets, however, such as the SteelSeries Arctis Nova 1 (8.3oz / 235g).

Weight isn't the only factor when it comes to comfort, but it certainly helps — and the Cloud Stinger 3 Wireless is a very comfortable and adjustable headset. The headband measures around nine inches (228.6mm) with the earcups fully retracted, and each earcup can be extended around 1.5 inches (38mm) for a total headband length of just under 12 inches (304.8mm). While it wasn't so comfortable that I forgot I was wearing it, I had no problems wearing the Cloud Stinger 3 Wireless for hours at a time.

It's a wireless headset, so it's got several on-ear controls. On the left earcup, from back to front, there's a power button and a USB-C charging port. On the right earcup, from back to front, there's a multi-function button, a volume wheel, and a mode switch for switching between 2.4GHz wireless and Bluetooth.

In the box, the Cloud Stinger 3 Wireless comes with a short, 1.6-foot (0.5m) USB-C to USB-C charging cable, a 2.4GHz wireless USB-C dongle, and a USB-C to USB-A adapter.

Audio Performance of the Cloud Stinger 3 Wireless

The Cloud Stinger 3 Wireless has large 50mm dynamic drivers with a listed frequency response range of 10 - 50,000 Hz, which is pretty far outside the normal range of 20 - 20,000 Hz (though we often see headsets outside this range these days).

Music on the Cloud Stinger 3 Wireless sounded good — a fairly neutral profile with no immediately noticeable distortion at either end of the spectrum. The bass in songs such as Kaskade's POW POW POW was powerful-feeling without being muddy, and I could hear the full reign of the kick drum at the beginning of Lorde's Royals. It wasn't the absolute clearest bass I've heard, but I was impressed with the balance between power and restraint. The mid- and high-range also sounded good, though perhaps a little too neutral for my taste. The high strings in Britney Spears' Toxic were well-curbed, but the dynamics in songs such as Farruko's Pepas and David Guetta's Titanium (ft. Sia) felt a little... lacking in excitement.

Games also sounded good on the Cloud Stinger 3 Wireless. While it doesn't have a particularly wide soundstage, the headset did a good job of picking up detail in dense, layered audio environments such as the cities of Baldur's Gate 3 and Where Winds Meet, or the jungle of Uncharted 4. In noisy environments these details did tend to blend together in a way that was more cacophonous than artistically rendered (but I suppose that's pretty realistic to how audio works in the real world, too). Directional sound was accurate, if not as finely-tuned as it is in premium, performance-oriented headsets such as the Razer BlackShark V3 Pro.

The headset features dual-wireless connectivity, though not simultaneous — a switch on the right earcup lets you quickly switch between the headset's 2.4GHz wireless and Bluetooth 2 connections. Audio sounded a little compressed over the headset's Bluetooth connection, but it was fine for taking calls or scrolling through TikTok.

Specs

Microphone of the Cloud Stinger 3 Wireless

The Cloud Stinger 3 Wireless has a non-detachable flip-to-mute boom microphone that sits on the end of a 3.5-inch (89mm) flexible gooseneck arm. The mic is a 6mm back electret condenser microphone with a unidirectional polar pattern, and it does a decent job of keeping background noise at bay, even without any software-based noise cancellation active.

As for quality, the mic isn't anything to write home about — it's a pretty typical headset mic that picks up vocals loud and clear, but won't make you sound like you're a professional podcaster (or anywhere near that level). My voice sounded decently full over both voice chat and in recordings, but it still had the thin, slightly fuzzy quality that easily gave away the fact that this was a headset mic and not a standalone USB gaming mic. But as long as you're not creating content, this mic is perfectly capable for communicating with coworkers, friends, and teammates.

Features and Software of the Cloud Stinger 3 Wireless

The Cloud Stinger 3 Wireless is supposed to be configurable via HyperX's Ngenuity software, but I downloaded both the new Ngenuity (Ngenuity 3?) and the legacy version of Ngenuity, and neither was able to recognize the headset when it was connected to my PC (wired or wireless).

Luckily, the Cloud Stinger 3 works pretty flawlessly out of the box, so you shouldn't need the software (HyperX's Ngenuity software generally tends to be lacking on features, anyway) to use the headset. The headset comes configured by default with voice prompts that tell you when you're switching between connections and when your mic is muted/unmuted (though this is fairly obvious given that it's a flip-to-mute mic), and tapping the power button once gives you the headset's current remaining battery life. There aren't any settings I desperately wanted to change, though I did want to see if tweaking the EQ could pull out footstep details, and HyperX does boast about its software-based spatial audio on the Cloud Stinger 3 Wireless' product page.

Battery Life of the Cloud Stinger 3 Wireless

HyperX rates the Cloud Stinger 3 Wireless' battery life at "up to 80 hours" over 2.4GHz wireless, which is pretty impressive given how light the headset is. The BlackShark V3 Pro, which weighs 12.96 ounces (367g), gets up to 70 hours of battery life, while the 1.23-pound (560g) Audeze Maxwell 2 gets over 80 hours. Of course, HyperX's other headsets also have battery life that far outclasses their peers — the HyperX Cloud III S Wireless, which weighs 12.05 ounces (341.5g) and retails for $180 gets up to 120 hours over its 2.4GHz wireless connection.

Bottom Line

The Cloud Stinger 3 Wireless is a lightweight, comfortable wireless gaming headset that offers impressive performance and battery life for its $99.99 retail price. It has large dynamic drivers with a wide frequency response range and a fairly neutral (if a little boring) sound profile and up to 80 hours of battery life. It's extremely lightweight and comfortable, though it does look and feel a little cheap overall, and its flip-to-mute boom mic is perfectly adequate for most typical gaming situations. At $100, it's not quite what I'd call budget-friendly (more like the low end of mid-range), but it's definitely worth the price. My only real complaint is that it's a little boring (in both sound profile and looks) — for something a little more fun, I'd still recommend the $120 Corsair Void Wireless v2.

It’s hard not to be impressed by what manufacturers have achieved with the numerous options that are now available among the best PC gaming handhelds. What was once a domain left largely to Valve with the Steam Deck has evolved into a broader market, with the bulk of the systems available running Windows 11.

These devices pair sleek exteriors with high-refresh IPS or OLED displays and pack powerful (for a handheld) computing hardware inside. However, even the gaming handheld market has been squeezed by rising component prices, leading us to the MSI Claw 8 EX AI+, the subject of this review. It’s by far the most powerful gaming handheld we’ve ever tested, thanks to its Intel Arc G3 Extreme SoC, but it’s also the most expensive ($1,799).

At every turn, the Claw 8 EX AI+ impressed with its performance, but the staggering price tag incessantly looms over the experience.

Design of the MSI Claw 8 EX AI+

We hope you like purple, as that’s the only color currently available for the Claw 8 EX AI+. Purple isn’t my favorite color in the world (that distinction belongs to blue), and it gives me Joker vibes from the 1989 Batman starring Michael Keaton and Jack Nicholson. Not only is the top of the chassis molded in purple plastic, but it also has a sparkly finish that makes it look a bit jewel-like. The lower half of the system is constructed of traditional black plastic.

The handgrips on the Claw 8 EX AI+ are well-spaced, perfectly sized for my hands, and evenly distribute weight, which helps mask the system’s 785-gram heft. Speaking of the handgrips, there’s a laser-etched dot texture molded where your palms and index fingers rest on the MSI Claw 8 EX AI+ — this helps keep it in place instead of sliding around in your hands (especially if they perspire during long gaming sessions).

Controls-wise, the Claw 8 EX AI+ doesn’t break any new ground, which is generally a good thing. If you’re familiar with an Xbox controller, there are backlit ABXY buttons on the right, with a joystick just below, along with another joystick on the left side, with a D-pad (which features a metal dome) below it. Both joysticks sport an RGB ring at their bases. There are also four buttons on the face of the handheld, flanking the display, with the bottom-left button assigned to bringing up the MSI Quick Settings overlay for the Xbox Game Bar. The bottom right button launches MSI Center M, for launching and configuring games.

You’ll find the usual allotment of bumper and triggers at the top of the Claw 8 EX AI+. While the triggers feel nice, the bumpers are a smidge “wiggly,” but that’s a nitpick on my part. If you’re a fan of macro buttons, you may be disappointed to find that there are only two on the back of the Claw 8 EX AI+, versus the three or four you’ll find on competing systems like the Steam Deck.

Unlike Lenovo’s Legion Go series of handhelds, you won’t find a touchpad on the Claw 8 EX AI+. So all of your screen navigation will be limited to touching the screen with your finger, or using the joysticks in the Xbox overlay or MSI Center M.

All of the I/O ports are located at the top of the unit; here you’ll find a power button that doubles as a fingerprint reader, two Thunderbolt 4 ports, a microSD reader, a 3.5mm audio jack, and volume buttons.

There are two intakes (one for each cooling fan) located beside the macro buttons, while vents below the IO panel exhaust heat.

Although I’ll talk about the screen in detail later on, I must mention that the display doesn’t fit neatly within the confines of the center mass of the Claw 8 EX AI+. Instead, the bottom portion of the screen extends roughly half an inch below the body. It looks a little wonky at first, but it seems to be the only way that MSI could fit the 8-inch display, at least without making the whole device bigger.

The Claw 8 EX AI+ measures 12.6 x 5.12 x 1.98 inches, compared to 11.42 x 4.76 x 2.00 inches for the Asus ROG Xbox Ally X, 11.64 x 5.38 x 1.66 inches for the Lenovo Legion Go 2, and 11.73 x 4.6 x 1.97 inches for the Steam Deck OLED.

MSI Claw 8 EX AI+ Specifications and Components

Gaming and Graphics Performance on the MSI Claw 8 EX AI+

The Claw 8 EX AI+ features the new Intel Arc G3 Extreme SoC, which includes a 14-core GPU (4.7 GHz max clock) and a 12-core Arc B390 GPU that clocks up to 2.3 GHz.

The default power mode is MSI's AI Engine, which typically runs games at 25 watts, but can vary between 15 watts and 30 watts depending on the workload. There's also an Endurance Mode, which leverages the Intel Endurance Gaming Efficiency Preset. This limits the chip to 15 watts and targets 30 frames per second (FPS) while gaming. Finally, Manual Mode allows you to run PL1 Max at 35 watts.

For our testing, we used the default AI Engine Mode when on battery, and Manual Mode when plugged in (35 watts PL1 Max, 45 watts PL2 Max). We ran games at 1280 x 800 and 1920 x 1200 resolutions on the handheld, leveraging the Xbox Full Screen Experience to limit resource consumption from additional Windows software. Please note, however, that the ROG Xbox Ally X benchmarks were run at 720p/1080p due to its 16:9 aspect ratio display, while the Steam Deck is limited to 1280 x 800 resolution.

While playing Lego Batman: Legacy of the Dark Knight, I was able to easily achieve 150-160 FPS on medium graphics settings at 800p with XeSS Balanced and the AI Engine power mode engaged. Battlefield 6 saw performance in the 70 fps range at 800p, with Auto detail settings and XeSS Balanced enabled.

Spoiler alert: the Arc 3 Extreme powering the Claw 8 EX AI+ is an absolute beast (in the handheld segment), delivering a 20 to 30+ FPS advantage over its peers across all the games in our benchmark suite. Starting with Shadow of the Tomb Raider benchmark (Medium, DX12), the Claw 8 EX AI+hit 98 FPS at 800p on battery power and 112 fps when plugged in (a 30+ FPS advantage over the ROG Xbox Ally X. Bumping the resolution to 1200p, the Claw 8 EX AI+ was faster at 1200p on battery power (67 FPS) than the ROG Ally X and Legion Go 2 were at 720p/800p on battery power. The system also impressed, reaching 72 FPS at 1200p when plugged in.

When benchmarking Cyberpunk 2077 (Steam Deck preset), the Claw 8 EX AI+ again showed its performance might, crushing all rivals. It delivered 78 FPS on battery power and 85 FPS when plugged in at 800p. At 1200p, those figures fell to 48 FPS and 52 FPS, respectively. We're at least getting a semblance of playability at 1200p resolution in the handheld space.

Red Dead Redemption 2 (Favor performance, Vulkan) saw the Claw 8 EX AI+ run the tables again, hitting 115 FPS at 800p and 87 FPS at 1200p while plugged in to the wall. Of course, those numbers fell slightly on battery power, but were still above anything that the ROG Xbox Ally X or Legion Go 2 could muster.

It was more of the same in Borderlands 3 (Medium, DX11), which saw the largest performance variance between the Claw 8 EX AI+ on battery power and when plugged in. The handheld hit 94 FPS at 800p while plugged in, a full 18 FPS faster than with battery power. At 1200p while plugged in, it still managed to pull 78 FPS at 1200p.

Forza Horizon 6 is the newest addition to our benchmark suite, so we only have Steam Deck numbers to compare with the Claw 8 EX AI+. The Claw 8 EX AI+ hovered around the 100 FPS mark at 800p and managed 72-76 FPS at 1200p, depending on whether it was running on battery power.

For stress testing, we ran Metro Exodus 15 times at 800p at Medium quality settings to simulate roughly 30 minutes of gameplay. The Claw 8 EX AI+ hit an average frame rate of nearly 70 FPS on the benchmark. For comparison, the ROG Xbox Ally X managed 63.44 FPS using the same settings.

The Claw 8 EX AI+’s two performance cores averaged 4.06 GHz, the eight efficiency cores averaged 3.18 GHz, and the four low-power efficiency cores averaged 3.0 GHz.

Windows 11 and MSI Center M on the MSI Claw 8 EX AI+

The Claw 8 EX AI+ supports the Xbox Full Screen Experience, just like the ROG Xbox Ally X. However, it doesn't have its own dedicated Xbox app. You can access the interface by opening the Xbox app and selecting Full Screen from the top-right corner. You'll then be able to reboot the system directly into the Full Screen Experience, bypassing unnecessary Windows 11 components that could affect gaming performance.

All the benefits and limitations of the Full Screen Experience, as discussed in our ROG Xbox Ally X review, apply here, so I won't belabor those points. I would like to elaborate on the MSI Center M, which provides its own dedicated, gaming-centric interface. The full-screen interface can be navigated using the thumbpads, eliminating the need to tap the screen.

MSI Center M pulls in all of your games from multiple sources, including the Xbox Store, GOG Galaxy, Epic Games, and Steam (among others). It pulled games from all those sources, and most of them displayed a nice, high-resolution thumbnail. One exception was Cyberpunk 2077, which only displayed a low-resolution Start Menu icon that was blown up to a blurry mess.

You can customize navigation within MSI Center M to use the joysticks in Gamepad Mode (for gaming) or Desktop Mode (for using the standard Windows 11 interface). You can tweak sensitivity for the joysticks, and the Desktop Mode provides key mapping so that you can see which UI shortcut each button corresponds to.

Another component of the software package is MSI Quick Settings, an overlay that appears in the Xbox Game Bar. It can be recalled using the MSI Quick Settings button beside the left joystick. MSI Quick Settings features a sleek, easy-to-use interface that provides access to power profiles, screen brightness, gamepad control modes, display refresh rate, and screenshot capture (among other things).

Display on the MSI Claw 8 EX AI+

The Claw 8 EX AI+ features an 8-inch IPS display with a 120 Hz refresh rate and a 1920 x 1200 resolution. Given the $1,799 price tag, it would have been more fitting for MSI to include an OLED display, as we saw with the Legion Go 2. I think that’s a big miss by MSI, and one that I hope will be rectified in future versions.

Unfortunately, due to time constraints, we were unable to perform detailed, instrumented tests on the Claw 8 EX AI+’s display. So you’ll have to trust my eyeballs and the manufacturer’s specs for the time being. MSI claims that the display on the Claw 8 EX AI+ covers 100% of the sRGB color space and offers a peak brightness of 500 nits.

Although I didn’t have a Legion Go 2 on hand for a side-by-side comparison, I do have a Legion Go and a Legion Go S on my testing bench. We measured the former at 476.7 nits, and the Claw 8 EX AI+ seemed at least as bright (at maximum brightness) as the Legion Go, if not brighter, across multiple games and while using the Windows 11 interface.

Games generally looked good on the Claw 8 EX AI+, with good color balance. I racked up plenty of hours in Battlefield 6, trying my best to soak in the lush reds, oranges, and greens of the city architecture of the Saints Quarter map without getting my head blown off. Forza Horizon 6 looked magnificent in the colorful Japanese landscape, with vibrant city centers and majestic outdoor environments reflected in the vehicles' finishes. Everything looked good, but color vibrancy and overall visual “pop” would have been even better with an OLED panel, but I digress.

Battery Life on the MSI Claw 8 EX AI+

I’ve been spending quite a bit of time recently playing more laid-back games; ones that won’t make my blood boil because of tense online matches. In particular, I’ve spent time playing Lego Batman: Legacy of the Dark Knight.

I played the game at 800p with medium settings, Intel XeSS Balanced, power mode set to AI Engine, and display set to 50 percent. After playing for an hour (with an average of around 150 FPS), the battery dropped from 100 percent to 52 percent. Switching the power mode to Endurance, locked at 30 FPS, I still had 66 percent battery after an hour (when starting at 100 percent).

If I were to drop the brightness a bit, I could see battery life extending past 3 hours in Endurance mode.

Audio on the MSI Claw 8 EX AI+

The Claw 8 EX AI+ has a pair of 2-watt speakers, which are fine. They’re perfectly serviceable for most games. Given that I spent plenty of time playing Lego Batman: Legacy of the Dark Knight, I noticed that the character voices sounded pretty good, and the plastic “clicking” sound of Lego stacking together during building exercises came through loud and clear.

Sound output was a bit more muddled in Battlefield 6, although I prefer to play games like that with headphones on anyway. In fact, given the mobility of handheld gaming PCs, most people will likely default to headphones for the best possible audio experience with the Claw 8 EX AI+.

Heat on the MSI Claw 8 EX AI+

While running the Metro Exodus stress test, the MSI Claw 8 EX AI+ registered around 90 degrees Fahrenheit on the back of the chassis, close to dead-center. Moving towards the top of the unit, heat levels rose sharply to around 109 F near the cooling fans' exhausts.

The Arc G3 Extreme chip averaged 77.3 degrees Celsius.

Upgradeability on the MSI Claw 8 EX AI+

Cracking open the Claw 8 EX AI+ is incredibly easy – only six Phillips-head screws hold the back panel in place. Once the screws are removed, you can use a well-placed plastic pry tool to insert it between the black and purple halves of the chassis where they meet. Once you pry open a small section, the rest pops off easily.

Once inside, you’ll see the battery in the lower portion of the chassis and a full-length 2280 PCIe 4.0 SSD nestled between the two cooling fans (secured with one screw). In our review unit, the SSD was a 1TB Micron 2500 with QLC NAND.

MSI Claw 8 EX AI+ Configurations

As of now, there’s only one configuration available of the Claw 8 EX AI+ with an Arc G3 Extreme SoC. Our system came with the aforementioned Arc G3 Extreme chip, 32GB of RAM, a 1TB SSD, and an 8-inch 120Hz 1200p IPS touchscreen, for a whopping $1,799.99 at Best Buy.

Bottom Line

The MSI Claw 8 EX AI+ is an impressive machine that offers a solidly-built chassis, good ergonomics, and good battery life. However, the most impressive aspect of the handheld is its performance. The Intel Arc G3 Extreme chip allowed the Claw 8 EX AI+ to absolutely dominate our gaming benchmarks, delivering anywhere from a 20 to over 30 FPS advantage at 1200p or 800p resolution.

But that performance comes at a steep price: $1,799. That’s more than just expensive; it’s MacBook Pro pricing for a handheld, which is shocking, to say the least. For comparison, the Lenovo Legion Go 2 with 32GB of RAM, a 1TB SSD, and an OLED display retails for $1,349.99, and we already had extreme reservations concerning its price tag.

The Claw 8 EX AI+ is $450 more expensive and doesn’t even have the OLED display of the Legion Go 2. The saving grace for the Claw 8 EX AI+ is the Arc G3 Extreme chip, which gives unmatched performance across the board. However, it remains to be seen if the price premium is enough of a draw to more casual gamers.

For a long time, the best gaming PCs were relegated to desks. Under the TV has long been the realm of the console, even if some, myself included, have hooked midtowers up to their living room screens. Valve's Steam Machine is an attempt to bridge that gap, letting people who play games on their rigs and on their handhelds also play comfortably on the couch.

The hardware isn't brand new. Like the Steam Deck, Valve has turned to AMD for semi-custom chips using some older technologies. In the case of newer, intensive games, this makes the Steam Machine a 1080p or 1440p computer, though it can support 4K on older games and in some cases with FSR.

But the Steam Machine is pricier than many had hoped, coming in at $1,049 for the 512GB version and, in our review unit, a $1,428, 2TB bundle that includes two faceplates and a Steam Controller. That's largely a result of the current state of the component market, but it will leave a lot more people asking if the Steam Machine (can or should) fill their needs, given the cost.

Design of the Steam Machine

The Steam Machine really looks less like a gaming PC than a mini PC. It's a black box that, at 5.98 x 6.14 x 6.39 inches including the system's feet, can fit discreetly on a TV stand or a desk.

The one part that really stands out is the integrated LED strip with 17 addressable RGB LEDs, which can share the Machine's system status or be customized to your liking. For instance, you can see the strip appear like a light bar when you download updates, and you can choose from solid colors, rainbows, or animations, like breathing. You can even control each of the 17 lights individually for a truly chaotic look. My preference was mostly to keep it off entirely for a minimalist effect.

The front of the Steam Machine is effectively a faceplate, which pops on and off with magnets. Valve ships two extras with the 2TB version: a fuzzy cloth-like red plate, and one with dark wood, which went well with my furniture. The company has also committed to releasing files for people to 3D print their own. (They have a good track record of this, having recently released CAD files for the Steam Controller and its puck.) It doesn't, however, have plans to sell the wooden and red plates separately.

Located at the base of the system are the front ports: a pair of USB Type-A 3.2 Gen 2 ports, a microSD card slot, and the power button.

The rest of the ports are on the rear: DisplayPort 1.4 and HDMI 2.0, the AC power connector, an Ethernet jack, two USB-A 2.0 ports, and a USB-C 3.2 Gen 2 port. (Despite not being officially labeled HDMI 2.1, the HDMI port does support 4K at 120 Hz, and has some other niceties, like HDMI-CEC to turn on televisions).

Also on the rear is the exhaust for the 120 mm fan attached to the heatsink that cools the APU. It's much more obvious than the intake, which is behind the front panel and draws air in from the sides. That fan is truly whisper-quiet. Even while benchmarking, I barely even heard it, and I had to pay attention and move my head near the system to notice anything at all.

Steam Machine Specifications

You can decide whether you believe the Steam Machine is a PC or a console. In Valve's eyes, it's a PC, and the spec list certainly looks like one. On paper, it's easy enough to see the significant jump from what Valve uses in its other gaming system, the Steam Deck, simply by nature of moving from Zen 2 to Zen 4 and RDNA 2 to RDNA 3.

The processor is a semi-custom AMD Zen 4 chip with six cores and 12 threads, going up to 4.8 GHz with a 30W TDP. Meanwhile, the integrated graphics are also semi-custom, using AMD's RDNA 3 with 28 compute units, going up to a maximum sustained clock speed of 2.45 GHz and a 110W TDP. The big number people are thinking about here is the 8GB GDDR6 RAM, which many enthusiasts feel is no longer enough to play some games above 1080p, let alone future-proof a system.

The system is powered by a 300W power supply, smaller than both the one in the Xbox Series X and the PlayStation 5.

Like the Steam Deck OLED, Valve has integrated a discrete Bluetooth antenna alongside the Wi-Fi 6E connection, which should help with latency. There's also a built-in antenna for the Steam Controller.

The Steam Machine starts with a 512GB SSD, but a more expensive option (the one we're testing) comes with 2TB. For further storage, you can add a microSD card (or swap out the SSD entirely).

Gaming and Graphics on the Steam Machine

If you're coming from the Steam Deck, the Steam Machine is a powerful upgrade. If you compare it to other gaming PCs on the market, you'll see that its GPU's aging technology is far from the most powerful option on the market.

First, let's put this GPU into context. Based on testing, we found that the Machine's graphics card would land somewhere towards the bottom of our GPU benchmarks hierarchy. To figure this out, we put together a Linux machine running Bazzite, with an AMD Ryzen 5 7600X and 16GB of DDR5-5600, memory along with both the Radeon RX 6600 — the bottom GPU on our list — and the RX 7600, which is the next AMD-branded step up.

In the Unigine Superposition (1080p Extreme) and GravityMark benchmarks, both of which run natively on Linux, the Steam Machine's graphics ran in between those two Radeons. Using our Cyberpunk 2077 configuration for raster testing desktop graphics cards, the same happened, with the Steam Machine producing 79.98 frames per second, behind the 7600X at 85.48 FPS. This is capable gaming performance, but bottom-rung compared to modern desktop GPUs.

I spent some time playing Resident Evil Requiem on the system. With the resolution set to 2560 x 1440 without any upscaling or advanced features like hair strands, the game ran largely smoothly through the Cedarbook Apartments section, as Leon sneaks past zombies, takes on a violent boss, and escapes through the other side of the building, though there were a few hiccups as he first entered the dark building. The game typically ran between 60 and 70 FPS, though there were some drops to around 20 FPS during the environmental transition, which were extremely noticeable. Here are my recordings from MangoHud, showing how the game ran:

On SoulCalibur 6, the game ran great at 4K, hitting the game's 60 FPS frame limit with maximum graphics settings as I progressed through Arcade mode as Siegfried. Granted, that game came out in 2018 and isn't super intensive, but people have all kinds of games like that in their Steam libraries, and they should play well.

Games that barely run on the Steam Deck, like Black Myth: Wukong, can be made to easily run on the Steam Machine. It's just clear that Valve isn't aiming for people looking for the highest-end performance on every game.

In my time playing around on the Machine, I did notice some crashes and slowdowns, often (but not always!) related to changing settings. One time, this led to the entire Steam Machine crashing and leaving artifacting on-screen when it booted back up. (Another reboot fixed this.)

Some of this may be due to the fact that some games see the Steam Machine as a Steam Deck. You can turn off that auto-detection, which helped to a degree. Valve says it is updating its APIs ahead of availability to avoid these kinds of problems.

One way we tested the Steam Machine was in comparison to the Steam Deck. On paper alone, it's no surprise that the Machine blows the Deck away, but we wanted to see exactly what kind of gains you could get when moving a game from the handheld to the desktop. Here, we tested at our typical handheld settings, though we ran the Steam Deck at native 800p while the Steam Machine was tested at 1080p, 1440p, and 4K.

When I tested, I found that most games would only run at 1080p, unless I went into game settings and changed the maximum display resolution to 4K. Valve reps told me that "1080p is the system default game resolution on Steam Machine to ensure a good gameplay experience out of the box," but you can change it on a global level in Settings > Display, or, like I did, on a per-game basis.

What this reveals is a vision of SteamOS that is significantly stronger than we've ever seen, playing most of our test games at 4K better than the Steam Deck can at 800p, including Forza Horizon 6 and Red Dead Redemption 2. But again — that's at settings designed for the Deck. And it also proved that not all games can run at 4K on the Steam Machine, including Cyberpunk 2077 on the Steam Deck preset. If you were someone plugging your Steam Deck into a dock and outputting that to your TV, you would get a better experience on the same settings.

You'll see some things missing. Shadow of the Tomb Raider, one of our go-to systems-testing games, wouldn't allow the game to run above 60 FPS, even with V-Sync off. That game was tested exclusively at higher settings, where that wasn't an issue.

When comparing to prebuilt PCs, we chose the CyberPowerPC Gamer Xtreme and Acer Nitro 60 that we tested last year. These were two of the last sanely-priced systems we saw before the component crisis got really bad, priced at $1,099.99 and $1,599.99, respectively. The CyberPowerPC boasted an Intel Core Ultra 5 225F and Nvidia GeForce RTX 5060, while the Acer had a Core i7-14700F and RTX 5070. Each offered 32GB of RAM. Notably, you can't find these systems at these prices now, which simply highlights the type of problem Valve had in pricing the Steam Machine. (The newer Acer Nitro 65 is over $2,000.)

For the most part, those larger boxes with desktop-class GPUs significantly outperformed the Steam Machine without any upscaling, FSR, DLSS, or similar technologies. Most importantly, the highest-end settings were playable on those machines. But those boxes are also significantly larger and have room to fit power-hungry components – and they cost a lot more now.

Valve definitely has size on its side. If you want something smaller than a mini-ITX build that comes with SteamOS installed, this is for you. But on paper, if you have nearly any GPU from the last three to four years, you already have a faster machine. And given that the Steam Machine starts at $1,049, that matters a lot.

When testing using our prebuilt desktop methodologies, which include some aspirational settings, it is clear why Valve says you need FSR to get 4K at 60 FPS. Based on the aging hardware alone, it should be clear that you won't be playing games at their top settings. But FSR can certainly help the Steam Machine along.

For example, on Red Dead Redemption 2 at medium settings, the Machine played the game at 20 FPS at 4K. But with FSR 2.0 in Performance mode, it reached 60 FPS.

On Forza Horizon 6's Ultra settings, the game ran at 30 FPS at 4K, but turning on FSR 3.1.5 Performance nabbed an extra 10 FPS.

Still, Cyberpunk 2077 was unplayable on Ray Tracing Ultra even at 1080p. Here, FSR 3.0 performance made it technically playable (up to 41 FPS from 15 FPS), but given the latency that could introduce, I wouldn't try it. (You can play this game on the Machine though — see the Steam Deck comparison above.)

If 60 FPS is your goal, the Steam Machine isn't a 4K machine, and I'm not sure Valve should have advertised it as one. It's much more suited for 1080p or 1440p gaming with appropriately middling-to-high settings, depending on what you're playing.

Upgradeability of the Steam Machine

The only exposed screws on the Steam Machine are on the rear. The two captive Torx T9 screws are in the top corners of the machine, so at least you don't have to worry about losing them. From there, a small pry tool pushed into in two purposeful-looking indents on the bottom lifts the back cover right off.

From in there, you'll see some studs coming in from the bottom. If you look closely at the feet, you'll see they have the same Torx indents in the center of the rubber, and that they're actually screws. This is way better than how some devices require you to remove adhesive to take off screws that are under feet. It's a neat trick that shows Valve had repairability in mind.

Back inside, two more T9 screws hold the fan assembly to the chassis. With these out, you can remove the internals in one massive piece.

From here, you'll be able to see all the ports on small daughterboards, as well as the antennas for the Wi-Fi and Bluetooth. But the real jewel is at the bottom: an easily accessible M.2 SSD slot mounted below the power supply. Ours came with a 2TB drive, which is fairly roomy, but this may become a must-have upgrade for 512GB Steam Machine owners if storage prices ever come down. And this drive is also held in with the same Torx screw, so you can use one screwdriver to make that swap.

Getting to the memory is far more involved and is more of a repairability compromise than we typically like to see. You need to remove the giant heatsink, which cools all of the components with the single fan, in order to get to the DDR5 SO-DIMMS. Given how tightly packed the Steam Machine is, with many cables and ribbon cables to daughterboards throughout the outside of the heatsink and PSU, that's a complex order that takes time and more risk than I think many Steam Machine owners may want to take. But given that the daughterboards are there, you should be able to replace broken ports, even if you have to do it in groups. Valve tells me it will partner with iFixit on repair manuals, similar to the Steam Deck.

Productivity Performance on the Steam Machine

The semi-custom, six-core/12-thread Zen 4 chip in the Steam Machine can hold its own against some current mobile chips.

The closest modern chip we had a record for is the AMD Ryzen AI 7 445, which has the same core count (with four Zen 5c cores and two Zen 5 cores), with a max boost clock of 4.5 GHz and a configurable TDP of 15-54 W. Valve's chip has 30W, but the GPU is discrete and isn't included here.

The Ryzen AI 7 445 ("Gorgon Point") in the Acer Swift Go 16 AI was marginally faster in single-core performance, but significantly faster in multi-core performance. On Handbrake, the Gorgon Point chip was 23 seconds faster than the Steam Machine, which completed the task in 6:33.

In our charts, you can also see comparisons to Intel's Panther Lake Core Ultra 7 355, a weaker chip than Valve's, and the Core Ultra X7 388H, which was stronger (but in far more expensive systems). Apple's M5, under air in the MacBook Pro, was the fastest of the bunch on both tests.

SteamOS and KDE Plasma Desktop

If you've used a Steam Deck before, everything on the Steam Machine will feel familiar. SteamOS 3 is the same here as it is on the handheld, just running on more powerful hardware. If you haven't used a Steam Deck before, but have used Steam's Big Picture Mode on a PC, you'll still be mostly at home, as the interface is very similar.

SteamOS continues to be Valve's primary advantage over the largely Windows-based ecosystem of gaming PCs. It's easily handled entirely with a controller. If you've used SteamOS on the Steam Deck, you might want to consider the Steam Controller, as you'll have all of the same buttons to navigate the operating system (and that's before you get into the fact that gameplay will feel similar).

Valve has adopted the Verified program from the Steam Deck to the Steam Machine. In Valve's documentation, it states that you need to hit 30 FPS at 1080p to be verified, which is pretty low stakes. Games that already run on Steam Deck should be shoe-ins, while the stronger hardware should enable more games to run on the Machine and earn the badge.

If you want a more typical desktop PC experience — perhaps you're playing at a desk – you can use the KDE Plasma desktop. While I suspect most people will never enter the desktop mode, Valve has added some significant updates here over the years, and I appreciate that you can use your computer as a computer. If you like to tinker and install extra software that isn't available through Steam, it's a great option.

Still, not all games run on Steam. While you can add most games to Steam through the "Add a Non-Steam Game to My Library" flow, not all work well. Some launchers have unofficial versions you can run through Linux, like the open-source Heroic Games launcher that will run Epic Games and GOG.

I wish that Valve offered a way to dual-boot Windows and SteamOS on both the Machine and the Steam Deck for these edge cases. The company said it would back when it announced the first Deck. With a 2TB drive, there is plenty of room.

"While Steam Deck is fully capable of dual-boot, the SteamOS installer that provides a dual-boot wizard isn't ready yet," Valve's page on Windows resources reads. "This will ship alongside SteamOS 3 once it's complete."

The Steam Machine is part of an ecosystem

There is a point in using the Steam Machine where I saw it as more of a platform. It was no surprise that the Steam Deck was built around playing games on a Valve platform, even if you can install other OSes. But with the Steam Machine in play, there's a fuller picture: playing your Steam games on the go, uploading the save to Steam Cloud, plopping yourself on the couch, turning on your Steam Machine, and resuming the same game, running locally, with the same controls thanks to the Steam Controller.

Perhaps one of the coolest things you can do is move your SD card from device to device. If you have an SD card in your Steam Deck, you can move it to your Steam Machine, and the games will be immediately playable. (Or, if you prefer, you could quickly move the games to the internal SSD.)

There are plenty of parts you can sub in there: You can play Steam on any handheld, or come home to your own custom-built rig, or use another controller. Despite its hardware, Steam still supports a ton of devices and ways to play.

But if you do have Valve's hardware, it starts to feel like an ecosystem on the level with Apple's, just focused exclusively on gaming. You get some benefits there — sleep and wake work just as well on this desktop as it does on Valve's handhelds. SteamOS is the best version of Steam's Big Picture mode out there. And this is way smaller than most DIY PCs. If you're all in on Valve, this is the way to go. But if you want more power and future-proofing, subbing in a more powerful PC will last you longer in the long run.

One thing that has been notable about the Steam Deck is Valve's commitment to updates. There have been a ton, adding features, squashing bugs, and making it more stable. In fact, that history is the one thing that makes me feel reasonably confident that the bugs I have seen will eventually be fixed.

Of course, Steam doesn't have every single game. Some won't run on SteamOS because of anti-cheat issues with Linux. Others simply have compatibility problems. Valve does have a method for running non-Steam games through Steam, but some, notably Epic Games' Fortnite, don't play well with it. You can install Windows or other launchers via Linux, but you will lose some of the ease the ecosystem offers. Valve offers minimal support for Windows, but at least it's something.

Steam Machine Configurations and Warranty

There are four configurations of the Steam Machine. First, there are two models of the computer; Both of them are identical with the exception of the storage. We reviewed the more expensive $1,349 version with a 2TB NVMe SSD and two extra faceplates, and bundling it with the Steam Controller brought it to $1,428.

The base model is a cheaper $1,049 option with a 512GB SSD. Bundling that with a Controller brings you to $1,128.

512GB isn't huge for a gaming system. Valve's spec sheet highlights that no matter which option you get, it comes with a high-speed microSD card slot. Luckily, the SSD is easy to access (See upgradeability above).

The $1,049 starting price is higher than consoles, including the more powerful PlayStation 5 Pro ($899) with 2TB of storage. A base PS5 Digital Edition is $599 with 825GB of storage. An all-digital Xbox Series X starts at $599.99. If you're looking for a living room solution to play games and don't care specifically about settings and your Steam library, those consoles are a better value.

In Asia, the Steam Machine will be sold through Valve's partner, Komodo, which also sells the Steam Deck. It will be available in Japan, Taiwan, and Hong Kong, but will not be sold in South Korea.

If you're buying this for a Steam library, you could also put Steam on any other computer and run it in Big Picture Mode. And given that supply is tight, that may be a better option for those willing to consider alternatives.

When we put together a parts list to estimate what a custom build looks like to match the Steam Machine, including a Ryzen 5 7600X, Radeon RX 7600, 16GB of DDR5-5600 RAM, a Gigabyte B650M Gaming Plus WiFi Micro ATX AM5 Motherboard, a 650W PSU, a case, cooler, and 1TB of storage, we hit $1,048.83. While our build gets you double the storage of the base Steam Machine, it doesn't get you the small case, dedicated Bluetooth or Steam Controller antennas, or super quiet operation. So if you're only comparing the Machine to other PCs, the price isn't terrible — it’s just the market, in general, that is.

Valve sells the Steam Machine with a one-year warranty.

Bottom Line

Valve's Steam Machine is a complicated little box. It was clearly designed for a simpler time, when components were plentiful, and it would be a somewhat affordable desktop that could be a more powerful option for Steam Deck owners to play their Steam games at home.

But it's not a simple time. The Steam Machine is still cute, still has a good selection of ports, still has an easily upgradeable SSD, and, most importantly, still runs SteamOS and gets all of the benefits that come with it. If you were docking your Steam Deck to the TV and wanted more performance, this will get you there, once Valve irons out the last of the bugs.

If you're just looking to get into gaming, a base-level PlayStation 5 or Xbox Series X is a better deal. You can buy one and a Nintendo Switch 2 and spend less than the Steam Machine.

You can also get many of the benefits of the Steam Machine on other devices. If you have an effective gaming PC or laptop, Steam Big Picture Mode will do most of the work there. Valve is also working to bring SteamOS to more machines, though currently it's only working on Radeon GPUs.

But if you want something small for your living room that plays years of Steam titles and maybe even has a cute little wooden faceplate, the Steam Machine is for you, but you should go in understanding its limitations.

The Kobra 4 Combo is an update of 2024’s Kobra 3 Combo, or perhaps a smaller version of the Kobra 3 Max, depending on how you look at it. It takes style points from the competition, with a brushed aluminum frame and pale plastic base that screams “A1 clone.” Though the printing experience is marked improvement over its predecessor, most of the technical advancement lies in the multimaterial handler, the ACE Pro 2.

Anycubic is clearly aiming the Kobra 4 at budget-minded beginners, with a current sale price of $379 as a combo and $279 as a single color machine. Which, oddly enough, is the same price as the four-color Kobra X we reviewed in February. The Kobra X was a true leap forward, introducing a brilliant multi-material filament switcher built right into the tool head that cut down on time and filament waste.

It’s not that the Kobra 4 is a bad printer. It's just a bit unwieldy when compared to the clean filament paths and slender footprint of the Kobra X. Both machines have excellent print quality and share a quick swap hotend introduced by the Kobra X. It has excellent auto bed leveling and vibration compensation, which is mandatory for modern 3D printers. The ACE Pro 2 can double as a filament drier, which comes in handy for printing moisture-loving PETG.

The Kobra 4 Combo doesn’t have what it takes to knock the Kobra X off its spot as our “best budget 3D printer”, so we’re not including it on the Best 3D Printers of 2026 list.

Specifications

Anycubic Kobra 4 Combo: Included in the Box

The Anycubic Kobra 4 was a beta unit and arrived almost completely assembled with the ACE Pro 2 in the same box. Included are two power cords, one for the printer and one for the ACE Pro 2, a signal cable, and four bowden tubes for connecting the printer to the ACE Pro. The included tool kit has hex keys, grease, and a nozzle cleaner.

Our printer was an early production unit that did not include a single spool holder, but we were able to borrow one from our Anycubic Kobra X to run TPU.

Assembling the Anycubic Kobra 4 Combo

The Anycubic Kobra 4 ships in one large piece. The only assembly is removing the packing screws and attaching the toolhead and purge wiper.

The brackets holding the bed in place require a very long hex key, which was included. Fortunately, the hex keys are magnetized because the screws are so far under the bed that they could be difficult to retrieve if they fell off while removing them. Once the brackets are removed, the included trim pieces hide the holes in the base nicely.

Attaching the printer to the ACE 2 Pro is simple with four Bowden tubes running from the tool head to the rear of the ACE 2 Pro, and a signal wire is run from the right side of the printer as well. Both the printer and the ACE require power cords.

Leveling the Anycubic Kobra 4 ­

Anycubic mastered auto bed leveling and Z offset with the Kobra 3 lineup, and this continues with the Kobra 4. The initial calibration includes PID tuning, resonance testing, leveling, and noise cancellation. The printer also offers bed leveling at the start of every print.

Loading Filament in the Anycubic Kobra 4

The Anycubic ACE 2 Pro makes filament loading simple. Just enter the filament in the feeder, and it gets slurped right up. This is a vast improvement over the earlier ACE generation, which often required multiple attempts.

Anycubic brand filament has RFID tags, which are read by the ACE 2 Pro and automatically identify the filament type and color. When using third-party filament, you need to enter the filament information from the printer’s screen and sync it to the slicer.

Design of the Anycubic Kobra 4 Combo

The Anycubic Kobra 4 shares its styling, motion system, and hotend with the Anycubic Kobra X, which is a definite upgrade over the Kobra 3. The Kobra 4 uses a new quick-release nozzle that uses a clip to hold it in place. Swapping a nozzle without tools takes no time at all. With the nozzle and the cutter being easily removable, clearing a filament jam also takes no time at all.

The Anycubic Kobra 4 has a 720p camera, which is good for monitoring from both PC and the Anycubic App, but not so much for recording timelapses. It suffers, the way nearly all bed slingers do, with the camera angle caused by its placement on the Z axis. The camera is also used for AI spaghetti detection, which does work, and is thankfully not oversensitive.

With the original ACE PRO, inserting filament was sometimes a chore, as it would often start feeding the filament next to the one you were trying to load. When trying to load filaments one after another, you had to wait for the first filament to be completely finished before loading the next, or misfeeds would happen. The ACE 2 Pro now operates like the Bambu Lab AMS with individual feeders for each filament. Currently, the feeding process is slow, and filament swapping takes a long time. I suspect that this will be sped up with a future firmware update.

The ACE 2 Pro has guides for the filament spools on the lid and on the back of the unit. Most of the spools we tried were compatible, including Prusament spools. Smaller spools, coils, and damaged or fragile cardboard spools can be run only on the external spool holder.

While jamming in the ACE PRO was rare, clearing the jams could require extensive disassembly. The ACE 2 Pro solves this with the filament tubes running externally on the bottom of the unit with disconnects at the buffer.

My favorite feature of the ACE is its ability to dry filament and dry while printing. This feature might be wasted on an open frame bed slinger, but it could come in handy with PETG.

Sadly, the Kobra 4 can dry TPU with the ACE 2 Pro, but it can’t run it. Even when I used “made for AMS,” it refused to run through the ACE. The solution is to use the single spool holder (which I had to borrow from the Kobra X) and run TPU directly into the toolhead. Anycubic promised to add a printable “single hub” to the printer’s memory to use with TPU, but for now, I was able to simply pop the whole top off and shove it directly into the extruder. I tested TPU down to 83A shore hardness, and the Kobra 4 did great with all of it.

The Anycubic Kobra 4 Combo is very quiet, with fan noise being primarily what you hear. The ACE 2 Pro is almost completely silent unless it is drying filament and even then the noise is minimal.

Preparing Files / Software

The Anycubic Kobra 4 requires a download of Anycubic Slicer Next which is based on OrcaSlicer, which is based on Bambu Studio, which in turn has PrusaSlicer at the foundation. If you are familiar with Orca or Bambu Studio you should know what to do in Anycubic Slicer Next.

From the home page of the slicer you can see your most recently opened files and access Makeronline, Anycubic’s file sharing library.

Anycubic Slicer Next can be run offline with files transferred manually via USB. I prefer to have access to my printer from my computer and the Anycubic app if only to stop a failed print from wherever I happen to be. From the workbench page of the slicer or the task details page on the Anycubic App, you can stop a print, skip a part, and adjust everything from print temps to print speed. You can print pre-sliced files directly from the Anycubic App but the ability to change the print settings is very limited.

Printing on the Anycubic Kobra 4 Combo

The review unit came with an adorable (though mostly useless) vacuum sealed pack with four coils of PLA. To fill up that ACE 2 Pro, you’ll want to check out our guide to the best filaments for 3D printing for more suggestions.

The Kobra 4 comes with a number of presliced files on the machine’s memory, which include useful tools, test prints, and a few fun models. It also had this “four color cube” test, which is a complete waste of your time, filament and sanity. Don’t print this.

I ran a lot of PLA while testing the Kobra 4. This pair of Geckos is a good example of the results. They printed in 5 hours and 9 minutes at .2mm layer height using 3 walls, 10% infill, and default speeds. Because of the size and shape of the model, the print speed is more like 160mm/s. It’s printed in Anycubic’s Cyan, Yellow, and Black PLA.

I did not change the default flushing volumes and ended up losing about 40 grams of filament in the filament purge. Due to a software bug, Anycubic’s slicer vastly underestimated the waste and even suggested it would flush negative values of yellow filament. This bug has been around since the Kobra X, and I really hope they have fixed it by now.

For PETG, I printed a plate of PentaClick balls from YosaNatural on Makerworld. With a .2mm layer height, four walls, and an average speed of 90 mm/s, they finished in 4 hours 38 min. The print looks great in Prusament Yellow Gold PETG, and the print-in-place hinges worked great with no hint of overextrusion. The model prints flat, then is folded and snapped into a ball shape. A small rubber band around the middle provides a spring action, so when you press the ball flat, it pops back into shape.

For TPU, I ran both hard 68D and soft 83A to make parts for our RC Deathracer, designed by Michael Baddeley.

While the smooth bumpers printed flawlessly, these spiked hubcaps, just based on their design, led to a fair amount of stringing. I’m confident this can be fixed by running with a bit more retraction. Using a .2 layer height and capping the speed at 50mm/s, the hubcaps took 5 hours and two minutes to print. This was printed using Microcenter’s TPU for MFS in black.

For softer TPU, I printed some tires I made in Tinkercad out of 83A TPE from Esun. This required removing the four bowden hub from the tool head and forcing the soft material directly into the extruder. It's annoying, but worth the effort as it ran the TPU beautifully. I used a .2mm layer height, two walls, and 10% concentric infill, with an average print speed of 50 mm/s. This took 1 hour and 35 minutes to complete. The tires are super soft and grippy and nearly perfect, with no stringing at all.

Bottom Line

The Anycubic Kobra 4 Combo is a step rather than a leap ahead of the Kobra 3 Combo, but it is a good value for an entry level 3d printer. It borrows the clean aesthetics of the popular Bambu Lab A1 and greatly improves the multimaterial handler from its first iteration. It’s a highly capable out-of-the-box experience for beginners.

However, it can’t quite compare to the Kobra X in value or space saving. I would only recommend the Kobra 4 over a Kobra X if you were seriously feeling the need for an all-in-one filament dryer solution.

The Kobra 4 Combo is only $20 cheaper than the Bambu Lab A1, which has a more established ecosystem, and $10 less than Creality’s SparkX i7 Combo, our pick for best beginner 3D printer.

Note 6/25/2026: After publication, Creality stated that the review unit may have been a beta or pre-production unit that did not fully reflect the intended retail configuration. Creality also stated that it has been unable to reproduce the rotary attachment issue internally and believes the attachment may have been damaged during shipping.

The Creality Falcon T1 is a galvanometer laser engraver that can swap between different laser modules for specific applications. With diode and fiber lasers for general engraving, a UV laser for creating 3D shapes in crystal, and a MOPA laser for adding color, the Falcon T1 is well-equipped on paper for most engraving tasks. The 175 x 175 mm engraving area and optional conveyor belt allow users to engrave small objects one at a time or in a large batch, and the optional rotary attachment can be used for tumblers, rings, and other cylindrical objects.

The Falcon T1 is at the higher end of Creality’s Falcon line of laser engravers, with the base model equipped with the 20W diode laser listed at $2,249 and the additional modules ranging between $989 and $3,059 each. As of the writing of this review, buying the Falcon T1 with the 20W diode laser module and adding all the additional modules would add up to $9,895 without any accessories such as the air filter, conveyor belt, or rotary attachment.

Creality shipped the Falcon T1 with three laser modules for testing: the 40W diode, the 20W fiber, and the 5W UV. The MOPA laser was not available, and I wasn’t able to use it during testing. Creality indicated the UV module will have a retail launch in late July or early August, with the MOPA laser launching approximately one month after the UV module. This staggered rollout makes the Falcon T1 a risky day-one purchase if either the UV or MOPA module is critical to your workflow. The Falcon T1 has clear potential, but the overall experience I had during testing was that of a beta tester.

I ran into software issues, waited for parts to be delivered, and even had to schedule a 10 PM video call with Creality’s China-based support team due to the lack of US-based technical support. This level of support may be acceptable on a lower-cost hobby machine, but it’s harder to overlook considering a fully loaded Falcon T1 can cost over five figures.

Specifications of Creality Falcon T1

Creality Falcon T1: Included in the box

The Creality Falcon T1 includes more accessories than can fit in a single picture, including calibration boards, USB adapters, a brush, locking fixtures for the laser table, and much more. In addition to the included accessories, Creality also sent over the optional conveyor belt and rotary attachment kits, both application-specific extras that enhance the functionality of the machine.

The packaging is what you would expect from a laser that retails at $2,249, with foam blocks and hook-and-loop straps securing the laser during shipping. Aside from sliding in the laser module (more on that later), there is virtually no assembly required other than plugging in the display and power supply. The Falcon T1 is ready to go right out of the box and doesn’t require advanced calibration or a lengthy setup.

Design of the Creality Falcon T1

The Creality Falcon T1 is a more refined product that stands out in the otherwise generally hobbyist-focused line of Creality Falcon laser engravers. The translucent black cover looks like something you’d find on a piece of lab equipment, and the handle on the front makes it easy to lift without touching the glossy surface. The tall aspect ratio of the Falcon T1 is also a departure from the otherwise flat laser engravers you’d typically see from Creality, and the square 175 x 175 mm engraving area is also an indication that this isn’t a typical diode laser cutter.

Creality heavily promotes the Falcon T1’s “5-in-1” functionality, which refers to the swappable modules that can be inserted and removed to change the type of laser. These include a 20W blue diode laser, a 40W blue diode laser, a 20W fiber laser, a 5W UV laser, and a 60W MOPA (Master Oscillator Power Amplifier) laser. These all have specific applications and are designed to be easily swapped by the end user.

There is an emergency stop (E-stop) button on the side of the Falcon T1 as well as an arcade-style button which is used to fully extend the laser tower to replace modules. I tested the E-stop and found that it immediately cut power to the unit and required a manual reset to turn back on. Above the E-stop is the LCD screen, which offers basic controls, history, and network information. Additional ports for power, rotary and conveyor accessories, and the exhaust vent are all on the back of the unit.

The design of the screen feels like an afterthought: the cable comes out from the left side of the unit which makes it awkward to hold with one hand and it’s sideways when mounted to the magnetic standoff on the engraver. Having the interface be vertical or putting the cord on the top of the unit would solve this, but with the current setup I found myself accidentally pressing buttons when I went to grab the LCD. After connecting to Wi-Fi, I used my laptop to start and monitor builds so the LCD didn’t see much use.

The Falcon T1 has an engraving area of 175 x 175 mm, which is in line with other galvo lasers like the WeCreat Lumos and the xTool F2 Ultra but smaller than most gantry-driven lasers. The engraving area has threaded inserts that allow workholding fixtures to be screwed in, something that’s critical when making very fine engravings where the object must be held still. The exhaust fan sits in the rear of the unit and did a good job during testing of pulling smoke and odor through the fan and into the air purifier unit.

The Falcon T1 is capable of focusing the laser automatically with no manual calibration required, a useful addition that when combined with the integrated camera means setting up a build can be as simple as only a few clicks. The automatic calibration only takes a few seconds, and the workspace in the Falcon Design Space software will display material inside the machine to assist with laying out builds.

Unlike a gantry-driven laser like the WeCreat Vision Pro, the galvanometer system of the Falcon T1 means the laser beam always originates from the top center of the chamber, and as a result it creates angled cuts due to the conical shape of the laser work area. This means that cuts become progressively more angled as you move further away from the center of the chamber, something to consider if making an assembly or a design that has any interlocking parts.

Safety Features of the Creality Falcon T1

Like all lasers, the Creality Falcon T1 can be dangerous if misused or left unattended. Flammable material can ignite during cutting, and engraving produces dust and smoke that can quickly fill a room. When operated with the lid closed the Falcon T1 is a Class 1 laser, which means the user is shielded from the laser during operation. There is a large E-stop on the side of the Falcon T1 which will immediately deactivate the unit when pressed, an important addition for any high-powered laser.

Opening the lid during operation will throw an error, and the interlock will immediately stop the laser and display a warning in the Falcon Design Space. Using the conveyor belt or rotary engraver requires disabling the lid interlock, something that users should take very seriously. The risks of using an unshielded laser are much higher than when the lid is closed, so goggles and ventilation should be used when the lid is opened.

Smoke Purifier for the Creality Falcon T1

The Creality Falcon AP1 Mini is a small desktop air purifier that is a good match for the Creality Falcon T1, and the 174 m³/h airflow is more than enough to pull smoke and dust from the unit during operation. Other air purifiers like the WeCreat Fume Extractor offer much higher airflow volume for laser cutters like the WeCreat Vista and WeCreat Vision, which are designed to cut through thick sheets of material and generate lots of smoke.

The Falcon AP1 Mini is a good match for the Falcon T1, and I didn’t notice any smoke leaking from the enclosure during engraving jobs. If you’re planning on using the Falcon T1 primarily for engraving on metal, wood, and doing some 3D engraving, the AP1 Mini will be a great fit. If you are more interested in using the Falcon T1 for cutting through plywood, acrylic, or other material, you may want to consider a unit that provides more airflow.

Software for Creality Falcon T1

Creality offers the first-party Falcon Design Space software for the Falcon T1 as well as advertising compatibility with LightBurn, another popular laser engraver software. Falcon Design Space offers control over the laser speed and power while also giving users editing tools for vector graphics. I’ve had good experiences with Falcon Design Space in the past and the current version of the software (v1.9.3) keeps all the good stuff while also adding in a new set of generative AI-enabled tools.

Between the modular laser attachments and the additional accessories, the Falcon T1 offers a total of eight different modes for processing a build. Without any accessories attached, the standard three modes are flat surface engraving (a typical laser engraver experience), 3D relief (designed for deeper cuts on material), and color engraving for the MOPA laser. Each mode offers a slightly different experience in the software, and the included video tutorials are well-made and easy to follow.

The software includes a mix of functional tools and creative ones, with the ability to generate 3D depth maps, pictures from generative AI prompts and filters, and general editing tools. Like with many generative AI tools, the output from these tools can vary wildly from surprisingly good to shockingly bad, and the overall experience will heavily depend on the user, inputs, and level of post-processing. I was impressed with the “3D Cutting” tool, which is similar to the Autodesk Slicer software that converts a 3D model into a series of parts for assembly, like a Benchy made of plywood.

Falcon Design Space also allows users to search through a repository of models and import them directly into the canvas, making it easy to find templates or starter projects for beginners. This is a fun way for new users to see what’s possible with a laser engraver, and the ability to modify or customize them lowers the barrier to entry.

One of my favorite functional tools in Falcon Design Space is the Material Test Array, which allows users to quickly create a matrix that shows the impact of adjusting the speed and power of a laser and how they influence the quality and properties of the engraved material. Using the 40W diode laser and the recommended settings, I engraved a sheet of basswood plywood and was able to easily see the impact of adjusting the parameters as well as selecting a configuration that worked for me.

Engraving / Cutting with the Creality Falcon T1

The 40W diode laser module was the most approachable for me to use as an initial test, so I loaded it in and set up a simple engraving / cutting job. Using the Benchy T-Shirt design by InfernoMoon, I used the 40W diode laser to engrave the design on some 3mm basswood plywood and also cut it out of the sheet. Adding layers in Falcon Design Space is simple and intuitive, so it was easy to set up the job so the engraving would be done before cutting the part out of the material.

At 800 mm/s, the coaster build was completed in just 1 minute and 44 seconds. Seeing the rapid movement of the galvo laser on the engraving is impressive, and compared to gantry-based lasers like Creality Falcon A1 Pro it’s no contest when it comes to speed. Based on the results of my initial material test array, I wound up making two versions: one at 800 mm/s and one at 650 mm/s, both of which looked good to me.

Switching to the 20W fiber laser, I engraved the JR-East Suica penguin logo on a dog tag to put on my luggage, and this was the moment that the galvo laser clicked for me. The software went from “Sending Job” to “Completed” in a total of three seconds, of which the laser was only activated for about one. I thought for sure that there was an issue with the design, but to my surprise, the engraving had taken only a single second to complete.

The default speed of 5,000 mm/s is blisteringly fast, and the engraving quality doesn’t suffer at all from this rapid movement of the galvo-driven laser beam. This is one of the most compelling features of the Falcon T1: the ability to make high-quality engravings on metal surfaces in only a few seconds with absolutely no prep work required. The 20W fiber laser is also capable of deeper engravings on metal, stone, and other materials, but the quick engraving is a very compelling feature.

Using the Falcon Design Space default settings, I engraved a dual-layer PU patch for a two-year work anniversary, which took just under 5 minutes. This dual-layer material looks great after a single pass, and the text is clear and legible against the darker top layer. The overall speed of the galvo system is a refreshing change from the lower-cost gantry-driven diode lasers, which operate at a fraction of the speed.

3D Engraving on Slate with the 20W Fiber Laser

The Creality Falcon T1 offers a 3D Relief mode that can create models typically associated with subtractive manufacturing, such as CNC milling. This mode will engrave thousands of layers away from an object to create a model with a deep relief that has a 3D shape. The Falcon Design Suite software has an integrated tool that allows users to upload an image and convert it to a depth map suitable for an engraving. Using one of the provided images, I went through the process to create a stone coaster with a 3D relief using the 20W fiber laser.

Once the image has been prepared, the 3D Relief Editor mode gives users more granular control over the engraving. The direction of the relief (emboss vs. deboss), the shape of the frame, and image adjustments are all fast and responsive, and the 3D preview mode instantly updates to show the impact of these changes. This was a surprisingly simple process, and I wound up using the default settings for my first 3D relief.

In 3D Relief mode, the Processing Preview in Falcon Design Space will look familiar to anyone who has used a 3D printer, with a series of contours on each layer stacked vertically to form the final model. This preview shows the path of the laser on each layer as well as an estimated processing time. For the 3D relief on this slate coaster, the original estimate was 1:53:30 but the final processing time wound up being 2:35:41.

The 3D relief looked incredible, and even the finest detail resolved on the slate coaster. Doing a deep engraving on a slate coaster creates a lot of dust, so I ran the AP1 Mini at max speed for this job. Even with the purifier running at the maximum setting, there was still quite a bit of dust to clean up inside the Falcon T1.

3D Engraving in Crystal with the 5W UV Laser

Creality has advertised the AI features of the Falcon Design Space heavily, but using them is a bit of a mixed bag. Using a headshot of myself with a clean background as a prompt, the resulting 3D model had me inexplicably in a toga and generally looked unimpressive. However, it did create a watertight 3D model, which is what’s required for the UV laser to engrave into a crystal. When using Falcon Design Space, I was only able to scale the model in increments of 10% (for example: 1.0, 0.8, 1.5), which was a frustrating experience when trying to align the model with the crystal. Using the special UV engraving mode in the software, I prepared the build and then set the Falcon T1 up for crystal engraving.

Installing the UV laser module is a bit more complicated than switching between the fiber and diode modules. Most of the Falcon T1 documentation is on a publicly accessible wiki, but the calibration for the UV module is spread across three Word documents that total an eyebrow-raising 57 pages. This calibration documentation is helpful, but much of it was watermarked with a stamp from around the day I had requested it, which made it appear unfinished and rushed. Considering the price of the UV module (over $2,000), I would have expected this calibration documentation to be a bit more presentable.

Switching from the fiber laser module to the UV module involves replacing the module, replacing the field lens with the F-Theta field lens, calibrating the laser, and calibrating the F-Theta field lens. Creality sent two lenses for the UV module: one for internal 3D engravings and one for flat engravings. The process to switch out the parts and calibrate took about 30 minutes and required two sheets of calibration paper to be used, something to consider if you plan on swapping between modes frequently.

Overall, the crystal looked as expected and would be perfect for presenting to a sales manager who just surpassed a quarterly quota. The engraving time was accurate (about 12 minutes), and the default settings worked well. I had some issues with alignment, and the model was positioned much lower in the volume than I was expecting based on the layout in Falcon Design Space. Engraving other objects had the same issue, which made me wish for a way to generate a 3D bounding box to visualize where the engraving will land in the crystal.

Conveyor Belt / Rotary Attachment Engraving with the Creality Falcon T1

Creality offers a conveyor belt attachment for the Falcon T1 which can be used for engraving parts longer than 175mm or batch engraving many small parts sequentially. The $459 optional accessory attaches to the work area using a few bolts and connects directly to the T1 for communication and power. When using the conveyor belt, the lid needs to be kept slightly open which can allow smoke and light to escape. With the lid open, the Falcon T1 should be treated as an exposed Class 4 laser system, so safety goggles, ventilation, and general caution should be exercised when using the conveyor belt.

Falcon Design Space has a specific workflow for the conveyor belt that involves selecting a design, aligning it on a workpiece, and then entering batch processing mode. I recently purchased a set of bookshelves for my wife’s ever-growing collection of books, and creating genre plaques for the shelves was a perfect batch-processing test. The batch processing mode has a few restrictions: only one design can be engraved, only one material profile per job, and settings can’t change between parts.

Each of the bookshelf plaques took about 23 seconds to engrave with the 20W fiber laser in a single pass at the default speed of 5,000 mm/s and 100% power. After each part was completed, the conveyor belt would advance a few inches, pause, and take a picture, and then scan for another part to engrave. As the parts completed, they would roll off the edge of the conveyor and hit the table with a satisfying click sound. My only criticism of the belt is the symmetrical design: having a shorter edge to the right of the unit would allow for more belt area for parts to process on the left of the unit without impacting the cost or size of the accessory.

Creality also sent a rotary attachment for the Falcon T1, but pricing and a launch date had not been announced as of this review. The optional rotary accessory includes two modes: chuck and roller. The chuck rotary uses an adjustable chuck mechanism for engraving things like rings, mugs, and other objects that are gripped from the inside or outside. The roller mode is for cylindrical objects, and has an adjustable width for accommodating various size objects.

During testing, I was not able to get the Falcon T1 to recognize the chuck rotary and always got the same error message when trying to connect. The motor used for the chuck and roller mode is the same, so this was an unusual problem and Creality was not able to fix it for this review. Creality does not have US-based support for this engraver, so I had to schedule a call with their China-based technical team at 10 PM. The issue was not resolved on the call, and Creality wasn’t able to fix it despite updating the software and spending several days on back-and-forth troubleshooting. For a small business that has a backlog of orders, this lack of urgent support is going to be a critical issue.

Bottom Line

The Creality Falcon T1 has clear potential to be an “all-in-one” laser for prosumer workshops, but seeing messages like “Oops… Calibration Completed” on the LCD gave me the impression that I was beta-testing a nearly-finished piece of hardware instead of a retail-ready prosumer manufacturing tool. The 175 x 175 mm engraving area of the Falcon T1 didn’t feel cramped at all while I was using it, and the high-speed galvo system meant most 2D engravings were completed in under a few minutes when using the 40W diode or 20W fiber laser.

The Falcon T1 is an impressive machine, but it suffers from an incomplete and confusing product rollout. Originally announced at CES 2026, the Falcon T1 is becoming available for order as of the writing of this review (May 2026), with the UV and MOPA modules following a staggered rollout instead of launching alongside the Falcon T1. Creality claimed the unit sent to me was a production unit, but the lack of a shipping date for additional modules, issues with customer support, and limited accessories available on the site are going to be a concern to serious power users who will rely on Creality for support.

The modular nature of the Falcon T1 lends itself to users who have varying interests or entrepreneurs with flexible business needs. I’ve frequently seen kiosks in shopping malls that engrave crystals with an image while you wait, and wondered about the commercial viability of the business model due to the high cost of the industrial laser required. The prosumer Falcon T1 is a laser that is capable of making these engravings without being locked into that single application, and can easily pivot to engraving dog tags, slate coasters, and other small products.

The Falcon T1 is not a cheap laser: buying the unit, all accessories, and starter materials can easily exceed $10,000. If you want a fast galvo laser to start making products today, the 20W fiber and 40W diode lasers are impressive and should serve that need well. If your decision hinges on the UV or MOPA modules, the Falcon T1 may be a laser engraver worth watching instead of buying immediately.

If you’re looking for a laser to get started with making custom products, the less expensive WeCreat Lumos offers a similar galvo laser setup but has a smaller volume and less powerful laser. The xTool F2 Ultra (one of the best laser engravers) offers a similar experience and has a MOPA laser for color engraving, something Creality advertises for the Falcon T1 but has yet to ship.

I’ve been reviewing video tech long enough to remember the advent of 4K displays. In 2013, my first experience was with an Asus PQ321Q. It cost around $3,500, delivered 3840x2160 pixels at 60 Hz, and only worked with a handful of graphics cards that had a DisplayPort 1.2 output.

As prices dropped and choices multiplied, I saw that one of the most common searches on Tom’s Hardware was “budget 4K”. Even though you still need a powerful video card to game at 8.3 megapixels, most users want that high pixel density on their desktops.

The good news is that 4K monitors are no longer $3,500. The ViewSonic VX2730D-4K I’m reviewing here isn’t even $350. It’s $280 and works with many video cards. And it runs at 144 Hz with a dual-mode switch that lets you reach 288 Hz at 1080p resolution. And you get wide-gamut color and HDR on a 27-inch IPS panel with high contrast and high performance. Let’s take a look.

ViewSonic VX2730D-4K Specs

The VX2730D-4K has an impressive feature set wrapped up in an IPS panel that delivers an impressive 1,500:1 contrast ratio. That’s claimed by ViewSonic and verified by me. In fact, I measured over 1,600:1 after calibration, which isn’t necessary because it’s nearly pro-level accurate out of the box. There’s plenty of color coverage too, over 98% of DCI-P3. It also supports HDR10 signals, and there I found a couple of issues that I’ll get into later.

A 144 Hz refresh rate anchors gaming performance at the full 3840x2160 pixels. You can run that signal over DisplayPort or the two HDMI 2.1 inputs. A dual-mode switch changes the resolution to FHD 1920x1080 and 288 Hz. In both cases, you get G-Sync with Nvidia certification and FreeSync Premium. It’ll also do VRR with consoles. There’s a three-level overdrive to smooth out motion blur, and you get MPRT backlight strobing to use when speeds drop below 120fps.

A few unimportant things have been sacrificed on the altar of value. There’s no LED lighting, nor are there USB ports. But you do get internal speakers and a 3.5mm headphone jack. A decent stand and cable bundle come in the box as well. The VX2730D-4K is more than prepared to take on your favorite games, be they twitchy shooters or lavish adventures.

Assembly and Accessories

The VX2730D-4K comes in the cardboard version of a plain brown wrapper with nothing but the words “ViewSonic Gaming” to signal the intent of what’s inside. Crumbly foam disgorges the three main bits, panel, upright, and base, which assemble without tools. Cables are DisplayPort and IEC for the internal power supply.

Product 360

The VX2730D-4K’s styling is minimalist industrial with nothing in front but the screen and its effective anti-glare coating, a ViewSonic logo, and a thin flush bezel. The IPS variant is proclaimed as “Super Clear,” which refers to its higher-than-average contrast. More on that later.

The stand is solid and stable and includes a 4.7-inch height range, 90-degree portrait mode and 5/20 degrees tilt. There is no mechanical swivel, but the metal plate base slides easily on the desktop if you want to turn the panel to either side. The package has decent heft and is very solid in operation. Build quality is well above the norm for a budget display.

In the back, there is purposeful grillwork that provides cooling airflow to the hardworking internals. A larger ViewSonic logo lets others know where your monitor comes from. And there’s a tiny OSD joystick reachable from the front bottom right that controls all functions. The side view shows a relatively slim panel with a component bulge. There’s a small cable hook on the back of the stand.

There are three video inputs in total, one DisplayPort 1.4 and two HDMI 2.1. All support 4K at 144 Hz and FHD at 288 Hz with Adaptive-Sync and HDR. You also get a 3.5mm headphone jack should you wish to forgo the internal speakers, which play politely in the upper mid frequency band.

OSD Features

The VX2730D-4K’s OSD is operated by a joystick, which I found to be a bit different than others. You’d expect that pressing it would bring up the menu, but instead, it’s the dual-mode switch so the screen goes blank for a few seconds, then changes resolutions and refresh rates. There’s no provision to change this, but the dual-mode switch is super quick and convenient.

There are nine picture modes aimed at different game types and uses. Standard is the default and best choice. It’s within a whisker of perfect; the VX2730D-4K does not require calibration. But tweaks are available in the Color Adjust menu where you can change the color temp, adjust saturation and hue and panel luminance. HDR is simply Auto or Off, there are no adjustments there. The sRGB choice in the color temp menu does not change the gamut size, so there is in effect, no sRGB picture mode. You get full color all the time, covering just over 98% of DCI-P3.

You can switch resolution/refresh modes in the OSD using the Dual-Mode field if you’d rather. Like any dual-refresh monitor, you must set up the refresh rates in the Nvidia Control Panel once initially before you can toggle back and forth.

There are three overdrive levels, of which Ultra Fast is the smoothest. I saw no artifacts on any setting. It is reasonably effective, but 144 Hz still has some blur. In the Setup menu, you can turn off Adaptive-Sync in favor of MPRT, which is a backlight strobe. At 144 Hz, the phasing artifact is severe enough that I couldn’t use it. It’s better at 288 Hz where you can only see it slightly. Adaptive-Sync, along with a video card that can maintain max frame rates is the best option. The VX2730D-4K includes PIP and PBP, which is called Multi-Picture. It allows the viewing of two video sources at once.

ViewSonic VX2730D-4K Calibration Settings

The VX2730D-4K measured extremely well out of the box in its Standard picture mode. Grayscale, gamma and gamut tracking were without visual flaws. I found a tiny improvement with a couple of changes to the RGB sliders in the color temp menu. I noted that though there is a choice labeled sRGB, it is not a color space selector. You will see the full native gamut in all scenarios, covering just over 98% of DCI-P3. My recommended SDR settings are below. In HDR mode, there are no options and I noted some issues, which you can read about on page five.

Gaming and Hands-on

The VX2730D-4K’s gameplay experience was excellent in every important way. It is a seriously quick monitor with extremely low input lag on par with many premium screens. Movement and aiming are very precise. Motion blur is well below the level of distraction, even at a relatively pedestrian 144fps when playing in 4K mode. Detail is tack-sharp thanks to that 163ppi pixel density. The panel’s high contrast and saturated color made fine textures pop in both the foreground and the background.

The dual-mode switch worked perfectly once I had set up both resolutions in the Nvidia Control Panel. I found I could switch when paused in Doom Eternal without exiting to the desktop. Playing in FHD/288 Hz mode showed even less motion blur. There is measurably less input lag, but I could not perceive the difference of just 3.1ms. Ultimately, I preferred the sharper detail of 4K even though the frame rate was lower. The overdrive is good enough to keep moving objects sharp, so I could aim precisely at a distance. Lobbing grenades across the icy arena was child’s play as they almost always fell on target.

The only flaw I found was in the VX2730D-4K’s HDR presentation. There is a blue cast to the image, and it is less bright than SDR. You’ll see this confirmed in the tests on page five. The color saturation is there but I ultimately preferred playing in SDR mode because it had more impact.

I’ve noted in the past that dual-mode monitors create a compromise where you get high frame rates but with a sacrifice in resolution. FHD was once a gold standard, but when compared to 27-inch QHD or UHD, it looks soft. There’s no getting around acquired perceptions and after reviewing almost 500 monitors in the past 13 years, I have a few. That said, 27 inches is tolerable for FHD. I can see its appeal to competition gamers who prize response and frame rates above all. The VX2730D-4K is a worthy gaming display for players of all skill levels.

For everyday use, the VX2730D-4K’s high pixel density and accurate color is a real boon. It’s ready to rock out of the box, no calibration needed. I tweaked it to a little higher standard but that’s more to satisfy ego than anything else. It’s colorful and bright and extremely sharp. It has about 60% more contrast than the average IPS panel and that’s a clearly visible difference. The image is cleaner and more textural, almost on the level of a good VA screen. The only thing I missed was an sRGB mode. There’s an sRGB label in the color temp menu, but it doesn’t change the gamut size. You get full DCI-P3 color for all content.

Takeaway: The VX2730D-4K is an extremely capable monitor with just a couple of flaws. HDR doesn’t look any better than SDR thanks to lower brightness and incorrect grayscale tracking. But SDR content is rendered with nearly perfect color accuracy and excellent contrast thanks to low black levels. Gaming is addictive thanks to extremely low input lag and a smooth overdrive. Flaws aside, the VX2730D-4K is a 4K display for $280. That’s about all you actually need to know.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

To compare the VX2730D-4K’s performance, I’ve included Alienware’s AW2725QF, Acer’s XB273K, Titan Army’s P2712V, Gigabyte’s GS27U, and Asus’ XG27UCS. The Asus and Alienware monitors are dual-refresh, while the others make do with a single refresh rate.

Pixel Response and Input Lag

Click here to read up on our pixel response and input lag testing procedures.

I’ve only collected data on six LCDs using the Nvidia LDAT test kit, so there are different displays shown in the panel response test chart. Of note is the KTC H27P3, which is a dual-mode, 5K@60 Hz/2K@120 Hz screen. There is almost no difference in response there, but the VX2730D-4K differs by 1.1ms between 288 and 144 Hz. 288 is also where you can actually use the backlight strobe if you want. I tried it at 144 Hz/4K, and it had too much phasing to be viable. The VX2730D-4K at 288 Hz also matched Acer’s excellent G-Sync Pulsar-equipped XB273U F5 at 2.4ms. At 144 Hz, the ViewSonic is good but not great. You can see some blur, even with the overdrive maxed out.

The VX2730D-4K’s dual-mode switch is the path to super quick input lag. It’s the quickest LCD I’ve tested in a while, with just 14.2ms of total delay when averaged over 25 mouse clicks. It’s also quick at 144 Hz, where it beats the 320 Hz Titan Army P2712V. This is excellent performance.

Test Takeaway: The VX2730D-4K shows some motion blur at 144 Hz, but no more than any competitor. It is much smoother at 288 Hz and FHD resolution. The overdrive works well on its quickest setting with no ghosting artifacts visible. The MPRT backlight strobe can only be used effectively in the 288 Hz/FHD mode. It has too much phasing at 144 Hz. Overall, the VX2730D-4K’s video processing is excellent, especially given its low selling price.

Viewing Angles

The VX2730D-4K has off-axis image quality that’s typical of the newest IPS panels. You can see a red tint in both the 45-degree horizontal and vertical planes. In the side view, light output and gamma remain constant. From overhead, brightness goes down by around 30% with a significant gamma reduction resulting in a washed-out image.

Screen Uniformity

To learn how we measure screen uniformity, click here.

The VX2730D-4K’s 6.29% score in my uniformity test is better than I’d expect for a budget display. Remember that we’re talking about 4K at 144 Hz for $280 here. There are no visible hotspots, nor is there bleed or glow.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

To read about our monitor tests in-depth, please check out Display Testing Explained: How We Test PC Monitors. We cover brightness and contrast testing on page two.

Uncalibrated – Maximum Backlight Level

The VX2730D-4K is firmly in the “not super bright but bright enough” category. For indoor use, 300 nits is plenty, and you’ll be more comfortable at 180-200 nits peak. What’s more important is the black level, which is almost 50% below the next screen from Titan Army. The resulting contrast of almost 1,600:1 is a standout among IPS panels. This feat is achieved without any clipping.

After Calibration to 200 nits

Contrast gets a tad higher after calibration, where it tops 1,600:1, well ahead of the P2712V. This is a difference you can see in a side-by-side comparison of actual content. Though a VA or OLED screen will look even better, the VX2730D-4K easily outpaces its direct competitors.

In the ANSI test, the VX2730D-4K stays consistent, even adding 3.7 points to the total. It’s rare for any monitor to be this consistent in my contrast tests. On planet IPS, it doesn’t get much better.

Test Takeaway: The VX2730D-4K has lower-than-average black levels for an IPS monitor, resulting in a contrast ratio of over 1,600:1. That delivers a greater perception of color saturation, depth, and sharpness. Coupled with its 163ppi pixel density, it delivers a supremely clear, detailed image. It isn’t super bright, but for any indoor environment, it has more than enough light output.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

The VX2730D-4K comes set to its Standard picture mode, which is accurate enough to be enjoyed without calibration. It’s close enough to spec that there are no visible errors for color, grayscale, or gamma.

Grayscale and Gamma Tracking

Our grayscale and gamma tests use Calman calibration software from Portrait Displays. We describe our grayscale and gamma tests in detail here.

If you look very closely at a grayscale step pattern, you might see a hint of red tint at the 90% brightness step. But more than likely, you won’t see anything wrong. And you certainly won’t see any problems in actual content. Gamma is also nearly perfect with close tracking to the 2.2 spec over the entire luminance range.

By calibrating the RGB sliders, which only required one click changes to red and green, I got all grayscale errors below 1dE. That’s pro-level accuracy. Gamma was mostly unchanged but the 70-90% region got a tad darker. This was not a factor in actual content. I noted a slight increase in overall contrast, too so it’s worth making the adjustments.

Though the VX2730D-4K includes an sRGB option in the color temp menu, it does not change the gamut volume, so there is no actual sRGB mode.

Comparisons

The VX2730D-4K’s out-of-box grayscale error of 2.14dE is impressive but not unprecedented. The Asus and Titan Army screens measure a bit better. But the top three monitors are without visible errors, so it’s a wash. After calibration, all the screens look the same, with no visible grayscale issues.

In the gamma test, the VX2730D-4K’s slight rise at 70-90% makes its value range a tad less tight than the rest. The 1.82% deviation represents an actual value of 2.24. There are no visible gamma issues with any of the monitors. They all exhibit excellent performance.

Color Gamut Accuracy

Our color gamut and volume testing use Portrait Displays’ Calman software. For details on our color gamut testing and volume calculations, click here.

Though you won’t see any color errors in content, the VX2730D-4K has slight hue anomalies in the secondary colors, magenta, cyan, and yellow when put under the colorimeter. Red is also a tad oversaturated throughout. Calibration fixes those bits neatly and takes the average error to an impressive 0.99dE. It doesn’t get much better than that.

Comparisons

All the monitors have impressively accurate color gamuts, but the VX2730D-4K is the best at 0.99dE. It’s rare for any monitor, even a professional screen, to get under 1dE for a 31-point average test. This is exceptional performance. In the volume test, the ViewSonic nearly covers the full DCI-P3 gamut, coming up a tad short only in the green primary. My only complaint is that there is no sRGB mode. You’ll have to perform color-critical tasks in P3, which is more than accurate enough. sRGB is useful for photography and web graphics so it would make a good addition for a future firmware update.

Test Takeaway: The VX2730D-4K has better than average out-of-box color with no need for calibration in its Standard picture mode. It covers almost 100% of DCI-P3, 98.38%, coming up a tad short in the green primary. It calibrates to a very high standard that only some pro monitors can match. The only thing missing is an sRGB mode. There is no provision for gamut selection here.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

Our HDR benchmarking uses Portrait Displays’ Calman software. To learn about our HDR testing, see our breakdown of how we test PC monitors.

The VX2730D-4K supports HDR10 signals with an automatic switch if HDR is set to Auto in the OSD. I noted some issues which I’ll detail below.

HDR Brightness and Contrast

The VX2730D-4K has an edge LED backlight with no dimming available for HDR so I expected that it would match output levels between SDR and HDR. I could only get to 207 nits with no provision for improvement. I checked full field and window patterns and got the same result. Black levels are about the same as they are in SDR with the backlight maxed, so contrast drops to 1,080.6:1. This gives HDR content the same appearance as SDR with no additional depth or impact.

Grayscale, EOTF and Color

HDR grayscale tracking is decidedly blue, and I didn’t need the meter to tell me there was an error. I could see it in step patterns and full fields. This further flattens the image to where it has less impact than its SDR counterpart. The EOTF runs a bit dark, below the reference line, but stays consistent enough to keep all detail visible.

In the gamut test, you can see the effect of the skewed grayscale tracking. All colors are pulled off their hue and saturation targets. HDR is colorful thanks to the VX2730D-4K’s large gamut, but it is not impactful due to the errors I noted. In the BT.2020, color runs out at 90% red, 75% green and 95% blue.

Test Takeaway: The VX2730D-4K’s HDR presentation is lackluster at best. Brightness is reduced by around 30% from maximum and contrast goes down as well. Grayscale errors result in a flat image with no more impact than its SDR counterpart. While the HDR mode is usable, it doesn’t provide any advantages over SDR.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

4K monitors have long been in the realm of the exotic, but they’ve come very far from their $3,500 roots. That you can buy a 27-inch UHD screen with 144 Hz, Adaptive-Sync and extended color for $280 is a clear indication of progress to the positive.

The ViewSonic VX2730D-4K delivers tremendous performance and quality for the price. Not only is it 4K at 144 Hz, but it is also Full HD at 288 Hz. Gaming is super quick and smooth thanks to excellent video processing and a convenient resolution switch, and I was even able to toggle it without leaving the game, which is a first in my experience. The overdrive removes nearly all motion blur, and gameplay is precise and satisfying at either resolution. There’s MPRT backlight strobing for smoothness at lower frame rates, but I found it was only usable in the FHD/288 Hz mode.

My only complaint here is with the VX2730D-4K’s HDR presentation. It is less bright than SDR by around 30% and its bluish grayscale is visible in content. Though the monitor supports HDR10 signals, it doesn’t give them much love. I found that after some time, I preferred playing everything in SDR mode.

Minor issue aside, the ViewSonic VX2730D-4K is a 4K gaming monitor for $280. That forgives a lot and at this point in time, it has no serious competition. If you’re looking for a budget monitor that’s speedy, responsive and colorful, you should definitely check it out.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

For a very long time, many of us have been begging Noctua to build a liquid cooler. In a time when CPUs are becoming ever more powerful, their power consumption is also rising, and air-coolers are often no longer up to the task. However, this was a problem for Noctua’s customers, as the brand has stubbornly not built a liquid cooler – at least until now. Meet the NL-LC1-36, Noctua’s first liquid cooler.

Today we’re reviewing the 360-mm variant, but it also comes in 240mm and 420mm flavors.

Let’s start with the elephant in the room: At its core, this is not a liquid cooler designed 100% in-house by Noctua – it’s an Asetek cooler with some Noctua touches. This may be a bit of a let-down, especially for Noctua purists – up until now, every product that had a name in Noctua’s nomenclature, such as NL-LC1-36, or NH-D15 G2, or NF-A12x25 G2, was 100% designed and built in-house by Rascom and Kolink, the two owners of the Noctua brand.

Keen readers will note that there is a Noctua case, Noctua PSU, and Noctua GPUs, but that’s not entirely accurate. The Antec Flux Pro Noctua Edition, Seasonic Prime TX-1600 Noctua Edition, and Asus Noctua Edition GPUs are all, as their name implies, Noctua-Edition products, which are not strictly speaking a core part of Noctua’s own lineup, and they do not carry Noctua’s typical nomenclature.

What’s really going on here is something we’ll get into later, but for now, let’s dive into the box and see what we’ve got on the test bench.

Specifications

Product Walkthrough

The cooler comes packaged in a typical Noctua box with a matte finish and product details written on it. Noctua also sells an auxiliary fan, the NL-ACF1, which can clip onto the pump block to cool the area surrounding the CPU.

Open the box, and we’re presented with a nice appetizer – the large Noctua plaque that can later be installed onto the pump block if you’re not using the auxiliary fan.

This is a truly beautiful plaque, and quite large too – a bit unusually large for Noctua. Is Noctua trying to emphasize its own brand to distract us from Asetek?

An included accessory box contains the mounting brackets, cleaning wipe, instructions, TIM guard, and the classic Noctua Torx screwdriver. Both the AMD and intel mounting brackets come with -7mm offset holes so that the cooler is better placed over the CPU’s hotspots.

The three fans come in their own box too. Here we have three NF-A12x25 G2s, two of which are the PPA variant and one PPB, which have 50 RPM offsets to prevent harmonics from forming between them. These fans spin at such consistent speeds, that if you mount two next to each other on the same fan splitter, they can sing in harmony.

The fans all come equipped with the full rear gasket for a perfect seal on the radiator.

Moving on to the pump assembly, here we find the pièce de résistance, the NL-PNA1. PNA stands for Pump Noise Absorber, and this is a custom pump cover designed by Noctua that features multiple layers of foam, sound barriers, and silicone mounts, all to quiet down the pump.

At its heart, the NL-CL1 is based on Asetek’s Emma Gen8 V2 pump platform that, although very powerful, isn’t the quietest pump on the market. However, the combination of its power and the PNA1 cap means that Noctua is happy to stand behind the product, and we’ll see in testing whether this stance is justified.

This pump cap isn’t only engineered to absorb and deflect noise, however – it’s also a tuned mass damper. Sitting on silicone mounts, its own mass and positioning is engineered such that it dampens mechanical noise. This damping essentially shifts the higher, more annoying frequencies pumps typically produce toward deeper, more pleasant frequencies that typically bother human hearing less.

The cold plate is beautifully milled, and the exposed copper should do a great job at transferring heat.

Hidden between the two tubes is a dip-switch with three positions. They’re not marked, but it’s easy to tell whether you’re in the quiet, balanced, or unrestricted mode. These limit the pump’s RPMs to 2100 or 2600 for the quiet and balanced profiles. Within these profiles, you can use the motherboard’s PWM control to guide the pump. However, the pump’s internal coolant temperature sensor will override the PWM signal at certain duty cycles or beyond certain coolant temperatures and push to 100% duty anyway, which may lead to the pump running louder than you want. In practice, however, we haven’t found this to be a problem.

For full control, you’ll want to use the unrestricted mode, where you have the pump’s full 3700 RPM at your disposal.

What I find a bit of a let-down here, is that there is no two-pin thermal probe output. I have a lot of experience with custom water cooling, and my preferred method of managing pump and fan speeds is by using the coolant temperature as the source value. It’s great that this AIO’s pump will manage itself and I’m stoked to see this feature finally coming to numerous AIOs, but I’d like access to the coolant temperature so that the fans can be controlled by this too.

The AIO’s radiator is a relatively standard aluminum type, but it’s 30mm thick which is an improvement over thinner 25mm radiators on more budget-friendly coolers.

Lastly, the loop is filled with Glycol, and is not user-serviceable.

Optionally, for $20, you can add an auxiliary fan to cool the VRM and memory surrounding the CPU, but this honestly feels like an afterthought and doesn’t look very charming compared to the rest of the cooler. I don’t doubt that it works, but it’s not nearly as neatly integrated into the pump block as we’ve seen with much of Noctua’s premium competition.

Testing Part 1: Pump speed vs. dBA

To test the NL-CL1, we decided to deviate from our usual procedure. Usually, we take AIO’s as they come and pit them against each other in a real-world scenario using a closed case and standard onboard fan curves. However, with the improvements Noctua has made to the pump assembly, we wanted to really focus on the impact of this different pump on both acoustic and thermal performance.

To do this, we’re pitting the NL-CL1 against the Phanteks Glacier One M25 G2 and the Lian Li Hydroshift II LCD-C 360N – but instead of using their included fans (the Lian Li doesn’t come with any), we’re installing the three NF-A12x25 G2 fans that come with the Noctua.

The reason we’re testing this way is because I feel that the primary audience that would be interested in this AIO, is also the type that would fan-swap an existing cooler. It’s also to take the fan variable out of the equation – we know Noctua’s fans are better than any other (they sit atop our Best PC Fans list just for that reason), and what we really want to know is how much better this pump unit really is.

These two competing coolers cost about $150 each, and if you were to install three NF-A12x25 G2’s on them, that brings the total cost to about $250, which is in-line with Noctua’s pricing. So, what do you get for the same money with the old method, vs the new, official Noctua cooler?

Meanwhile, for the purists among you, we’re also adding the legendary NH-D15 G2 air cooler to the test at the end. The main reason to contemplate a Noctua AIO is to minimize pump noise – and for a few, up until today, that simply meant not using an AIO at all, rather sticking to air cooling with fewer moving parts.

Test Setup

We’re using an MSI Godlike X870E motherboard with a 9950X3D CPU and some Kingston Fury Renegade DDR5-6000 memory. We also have a PNY RTX 5080 installed, though it’s there only for display purposes and will be kept at its zero-RPM mode.

We’re doing these tests on an open test bench, as to expose as much of the pump as possible.

Testing Part 1: Pump speed vs. dBA

For the first series of tests, I ran the three AIO’s pumps through their RPM ranges independently, by themselves. To get the system quiet enough, we ensured the GPU’s fans were not spinning, there was no load on the CPU (and thus no whining VRMs or PSU fan), and disconnected the CPU fans – the only moving part was the pump itself – that’s all we’re trying to figure out here.

What’s clear from the results is that the Noctua pump is notably quieter at most speeds, right up until about 2300 RPM, where Phanteks’ Glacier One becomes quieter, and Noctua’s pump actually makes a big jump up in noise levels.

What’s also interesting is that up until about 1900 RPM, the pump on Noctua’s cooler is almost indistinguishable from the noise floor of the room. That’s very impressive if you consider that the dB meter was placed at a mere 20cm from the pump, at night, in a room treated for acoustics.

I debated testing with dB-C weighting, as this is more sensitive to low frequencies, but decided against this. C-weighting spans across a wider range of frequencies, which technically speaking, would offer results more representative of the entire sound spectrum.

However, dB-A weighting is geared more toward the sensitivities of human hearing – and Noctua specifically engineered the pump cover to shift the pump’s sound profile towards lower frequencies, where they would be less bothersome.

With that in mind, I do want to mention that the dB reading alone really doesn’t tell the whole story. As mentioned, the Noctua pump ends up louder than the Phanteks, at least as far as decibels are concerned.

As I was testing, I saw these numbers, and I thought: that’s not right. How can that be? The Noctua sounded much quieter to my ears.

So, I went back, re-installed the Phanteks, and lo-and-behold, it put out the exact same figure it did before – quieter than the Noctua on the dB meter, but my ears told a different story.

Using dB(A) weighting should have compensated for that difference, but it didn’t. So, perhaps my dB meter is wrong, or Noctua knows more about human hearing than dB(A) weighting accounts for.

However, what may also play a role is the differences in pump assembly and how they impact flow rate. If, for example, the Phanteks pushes more fluid per rotation than the Noctua, then it’d be only logical that it’s louder too – but I doubt it does.

Either way, we’ll be going back and testing with subjective noise levels in addition to these objective figures for a complete picture.

Testing Phase 2: AIO Objective Noise-Normalized, + Min and Max

In order to account for the pump’s engineering, I’ll be running a noise-normalized test that has the NF-A12x25 G2 fans spinning at the same 1200 RPM speed on each cooler, on every test. I’ll be running the three pumps at speeds that have matching noise levels, which we’ve figured out in the previous testing phase with the dB meter.

This is not a test meant to simulate real-life performance -- we run the fans at a fixed speed of 1200 RPM, and keep the CPU at stock settings to limit power draw to exactly a hair under 200 watts under all scenarios, so that all coolers make it through the test and to remove any variables. The only factors influencing the thermal results are the pump, its mounting offset, the flow resistance, and the radiator – all factors over which we ultimately have no influence.

This is where things started to get very interesting. Let us start with the minimums and maximums – you’ll note that at the minimum pump speed, the Noctua has the CPU running much hotter than the other two coolers. I’d like to clarify that the minimum and maximum tests are not noise-normalized, but have the pump running at their minimum and maximum speeds. Noctua’s LC1 has an extremely wide RPM range to work with, which the other two coolers don’t. And running a 200W workload with a pump running at just 750 RPM isn’t going to do great things for thermals. It does, though, mean that in low-load situations, the pump can run completely quietly.

However, in the noise-normalized test between 29 dB(A) pump noise and 32 dB(A) pump noise, we saw some shocking results. Noctua’s cooler, even when it has its pump running at a barely perceptible noise level of just 29 dB(A), still runs significantly cooler than both the Glacier One and the Hydroshift with their pumps running at full speed.

Of course, it must be noted that there are more factors at play here than just the pump – the LC1 offers a 7mm offset mount for Ryzen systems which we used, a different cold plate design, and a radiator that’s slightly thicker than its competition.

Testing Phase 3: Subjective Noise-Normalized

As mentioned earlier, according to my ears, some of the pump’s acoustics don’t match up with what my dB meter shows. For this reason, I’m repeating the previous batch of tests, but instead of looking at the dB meter, I’m trusting my ears to guide me in picking an RPM setpoint for each pump that I deem fits the description. If you see a result missing, it’s because the criteria weren’t met. The RPMs of the pumps are in the chart below.

To determine the RPM levels, I first ran the pump at the highest speed at which it was still completely inaudible. I then figured out a middle point where I was happy to tolerate the noise level under load. And finally, I pushed the pump speed up further, and found the lowest value at which its noise started to annoy me.

To give some additional context, I’d like to emphasize that I have highly sensitive hearing, and I’m someone that likes to run their gear as close to inaudible as possible. This is also a very subjective test – you may hear something where I didn’t – but on the flipside, what annoys me, you may be just fine with, so do take these findings with a grain of salt.

Then, with those numbers, I ran the thermal tests again, all with the fans at 1200 RPM to remove any other variables.

Here, the Noctua started to really stand out from the crowd. Even pushing the Glacier one to annoying noise levels couldn’t make it keep up with the thermals produced by Noctua pump running inaudibly. That’s honestly quite staggering.

I contemplated for a moment whether to adjust the fan speeds in this test to match the annoyance factor, but this would make comparison difficult, and I really am trying to focus on the difference the pump makes here, and nothing else.

What’s really important to note here, though, is that from my experience, I can truly say that the dBA figure means almost nothing in practice. The subjective noise profile of the pump matters way more than the dBA, and whereas the Noctua might, strictly speaking, be louder than the Phanteks at higher speeds, it’s also far smoother. Whatever they did with the PNA1, it’s working.

Meanwhile, though, from the results it’s also clear that there is no need to run the Noctua at these higher pump speeds, as there is very little additional benefit.

And that’s not to even mention the Lian Li pump, which is always audible. Consequently, I had to omit it from the inaudible category entirely, and I can tell you from experience, there are many AIOs that are unable to run their pumps at inaudible levels. This is not an isolated case.

Testing Phase 4: Air vs Water (Noise-Normalized)

As mentioned, we also added the NH-D15 G2 to the mix, as it is Noctua’s most powerful air cooler and the most surefire way to avoid pump noise entirely, and I’m sure many of you who want nothing except Noctua, actually want to know how the new liquid cooler compares to Noctua’s finest air cooler – not a bunch of AIO’s you wouldn’t buy anyway.

TheD15 air cooler has – as its name implies – much larger 150mm fans. Consequently, to test fairly against the NH-D15, we decided to set two noise targets, based on the AIO’s internal pump limiters selected with the switch on quiet and balanced.

The test we’re performing here is more representative of real-world performance than the previous bunch. For the quiet and balanced pump profiles, I set the AIO’s fans to a speed that I felt matched up nicely, where the fan and pump noise blended smoothly. I measured the total dB(A) produced, then figured out which fan speed the NH-D15 G2 had to run at to produce the same overall noise level.

For the maximum speed test, dB(A) figures were ignored – all gear was run at full speed.

The results speak for themselves. Noctua’s NL-LC1-36 offers vastly superior performance to the NH-D15. Even with the NH-D15 G2 air cooler running at full tilt, producing a highly annoying 43.9 dB(A), it still runs 6 degrees hotter than the LC1 running in the quietest profile with the fans barely even trying at just 1000 RPM, producing only 29.6 dB(A).

Noctua has been defeated. Long live Noctua.

When I first heard that Noctua’s first liquid cooler would be an Asetek unit with some Noctua sprinkles added to it, I had some major doubts about its efficacy and, quite frankly, Noctua’s business decisions.

I grew up knowing Noctua as a company that prides itself on its standards, and designing its products entirely in-house. With that in mind, the Noctua NL-LC1 is a bit… unlike the classic Noctua that we know and love – at its core, it’s not a Noctua cooler in any way we’ve seen before.

However, if you look at the history of companies that have tried to build their own AIO’s, you’ll be met with a very long list of fallen soldiers -- whether that’s due to being unable to build a reliable pump, or because of litigation from Asetek when their patent was still valid. If history has taught us anything, it’s that it’s surprisingly difficult to build a good AIO with no liquid cooling experience.

In that sense, I somewhat doubt whether the LC1 is the liquid cooler Noctua actually wanted to build. The company is also still working hard on its pumpless Thermosiphon cooler. But in the meantime, the market still wants what the market wants.

With CPUs becoming ever-more powerful, and consumer demands increasing, Noctua either had to find a way to give the people what they wanted, or get left behind in the CPU cooling space – a space they once dominated.

With all this in mind, I believe that the LC1 is the result of Noctua making a compromise – they won’t design the cooler entirely in-house, but instead work with Asetek for the pump and radiator, but adding their own pump cover, mounting bracket, and fans to bring the total package to a new industry standard.

However, this compromise is proving to pave the way to greatness. From my brief testing, I can safely say that this is an absolutely fantastic AIO – and I’m glad to finally see somebody making a liquid cooler that focusses on what really matters: noise and performance.

Priced at $220 to $280, it’s not the cheapest AIO, but certainly isn’t the most expensive either. So, despite the fact that it’s not 100% Noctua, this one comes highly recommended to anybody looking for a high-quality AIO with fantastic performance and a stellar 6-year warranty.

So, as much as I may not feel like this AIO appeals to me as a Noctua purist, as a hardware enthusiast, I really quite like it, and it’ll certainly be finding its way into a personal system.

The best gaming desktops don’t just chase raw speed – they deliver the right balance of power, style, and value. CyberPower’s Gaming Desktop GXi3800BSTV2 ($2,309.99 as tested) comes out swinging with a stylish new case, complete with analog lighting knob, a spotless interior, and plenty of RGB for a showroom look. Backed by an Intel Core Ultra 7 270K Plus and an Nvidia GeForce RTX 5070, it’s built to impress, and despite a few concessions, it’s hard not to admire how much CyberPower gets right for the money.

Design of the CyberPowerPC Gaming Desktop GXi3800BSTV2

The biggest draw for this particular CyberPower configuration is its MA-01 case. It offers analog RGB controls and hidden fans and radiators for a strikingly modern look. Though the fans aren’t truly hidden – they’re concealed behind plastic grates to give the impression that they’re not there – it adds a level of polish and cleanliness to the interior lines.

CyberPower did its best to avoid directly visible LEDs for a soft look and mostly succeeded; The LEDs in the CPU water block are an exception. It missed a minor opportunity to add more RGB lighting by using standard DIMMs rather than RGB-lit ones, but this tower arguably already runs over the line of too much RGB. Everything is visible through the artfully curved panel, which highlights the obvious about this case – it’s designed to be seen and shown off.

The three lighting dials are unlabeled. Starting from the top are controls for colors, brightness, and patterns; the latter includes breathing, strobing, and rainbow effects. (This seems to be the more basic version of the MA-01 case; on other versions, the dials are for Red, Green, and Blue for finer control.) Changing the color mix takes just seconds. You can also press in on any knob to achieve an all-white look. It’s easy to create a unique look without using software. (In fact, you can’t use software to control the lighting on this PC since the individual RGB elements are only recognized as a generic RGB strip in the Asus Armoury Crate app.)

Build quality is where this case, or at least this version, shows some budget elements. While it didn't creak or bend when I picked it up, the rolled steel panels feel thin. Additionally, the viewing panel, which CyberPowerPC says is tempered glass, feels plasticky and is not as clear as it could be. Last, the analog knobs feel cheap, and the bottom one tended to fall off in my testing. Size-wise, this case is decidedly in mid-tower territory at 9.5 x 17.7 x 20.9 inches.

CyberPowerPC Gaming Desktop Specifications

Ports and Upgradeability on the CyberPowerPC Gaming Desktop

CyberPower’s desktop offers respectable input and output, starting on the lower front panel with one USB-C and two USB-A ports and a 3.5 mm audio jack. This port placement should work well if this desktop is on top of a desk. The power button is on the adjacent side next to the case lighting control dials.

Around back, the Asus B860M Max Gaming AX motherboard offers USB-C (20 Gbps), four USB-A 3.2, four USB 2.0, DisplayPort, and three 3.5 mm audio jacks. At least one truly high-speed port like USB4 or Thunderbolt 4 would have been ideal. Also, Wi-Fi 6 from the built-in RealTek RTL8851BE wireless card is several years out of date.

Getting inside this CyberPower’s case isn’t straightforward. Both side panels look like they should pull off, but they are actually screwed in at the top. The top grate, which doesn’t have any obvious fingerholds, must be popped off to access the two Philips-head screws securing each panel. Removable dust filters line the right panel.

Motherboard expansion slots include one PCIe 5.0 x16, two PCIe 4.0 x16 supporting x4, and one PCIe 4.0 x16 supporting x1. Memory can be expanded with four DIMM slots. There are also three M.2 slots, one of which is PCIe 5.0. The other slots support PCIe 4.0, with one on the front and one on the back of the board. The latter is accessible by removing the tower’s right panel. You can also see the 3.5-inch drive cage in the lower chamber.

The 850-watt power supply isn’t modular, leading to a mass of tied-off cables, but it easily supplies enough juice for the components in this PC. CyberPower’s cable management is acceptable, with cables routed and bundled neatly enough for a mass-market PC.

Gaming and Graphics on the CyberPowerPC Gaming Desktop

The CyberPower we tested features a Core Ultra 7 270K Plus processor, Nvidia RTX 5070 graphics, and 32GB of DDR5-6400 RAM.

For some informal play, I launched Crysis Remastered. Playing at a 3840 x 2160 resolution with everything at “Can it run Crysis?” settings, I saw between 75 and 85 frames per second (FPS) outdoors, with slight dips in the jungle and wide-open landscapes with lots of shadows. Overall, the gameplay was exceptionally smooth.

We used two gaming desktops for comparison: the Acer Nitro 65 ($2,069.99 as tested) combines a Ryzen 9 9900X with an RTX 5070 while the iBuyPower Y40 Pro ($1,899.99 when we tested it, but very difficult to find now) pairs a Ryzen 9 7900X with an RTX 5070 Ti.

Starting in Shadow of the Tomb Raider at the Highest detail preset, the CyberPower and Acer tied the 1080p numbers at 209 FPS and were within one frame at 4K. Naturally, the RTX 5070 Ti-equipped iBuyPower dominated both, with 228 FPS at 1080p.

The story was similar in CyberPunk 2077 at Ray Tracing Ultra settings. The CyberPower achieved 71 FPS at 1080p but fell to 21 FPS at 4K, being within one FPS of the Acer in both instances. The iBuyPower surged ahead, with 87 FPS at 1080p, but even it couldn’t make a playable time of it at 4K, with just 29 FPS.

In Far Cry 6 at Ultra settings, the CyberPower led the 1080p scores with 141 FPS, well ahead of the Acer (132 FPS) and iBuyPower (125 FPS). Things evened out at 4K, with the CyberPower and Acer at 81 FPS and 82 FPS, respectively, and the iBuyPower about 25% faster, at 102 FPS.

Red Dead Redemption 2 at Medium settings was a disappointment for CyberPower, with its 126 FPS at 1080p noticeably trailing Acer (141 FPS) and especially iBuyPower (161 FPS). The gaps narrowed at 4K, though CyberPower was still last – it scored 51 FPS next to the Acer’s 54 FPS.

CyberPower redeemed itself in Black Myth: Wukong at Cinematic settings, its 55 FPS at 1080p overtaking the Acer (50 FPS) while being behind the iBuyPower (66 FPS). It tied the Acer with 25 FPS at 4K though it couldn’t catch the iBuyPower (32 FPS).

Overall, though CyberPower’s Core Ultra 7 270K Plus didn’t give its RTX 5070 enough of a boost to overcome the iBuyPower’s RTX 5070 Ti, it still produced impressive numbers, often leading the RTX 5070-equipped Acer by a decent amount.

Equipped with the RTX 5070, this CyberPower offers ample horsepower for gaming at 1080p/1440p and 4K, though fluid playability at 4K in the most demanding titles may require reducing the visual quality settings.

To stress test gaming desktops, we simulate half an hour of gaming by running 15 loops of the Metro: Exodus benchmark at RTX settings. The CyberPower desktop averaged 138 FPS with only a few tenths of an FPS variance between runs, indicating stable thermal performance. The Core Ultra 7 270K Plus ran at an average temperature of 48 degrees Celsius while the P-cores ran 5.34 GHz at and the E-cores 3.77 GHz. The RTX 5070 ran at 61 C and had an average core clock of 2.74 GHz.

During testing, the desktop’s fans were noticeable but not intrusive; while gaming, I could hold a conversation with someone sitting next to me without the fan noise becoming a topic.

Productivity Performance on the CyberPowerPC Gaming Desktop

Equipped with a Core Ultra 7 270K Plus, 32GB of RAM, and a 2TB PCIe 4.0 SSD, the CyberPower proved formidable in our productivity testing.

While the rig's single-core score of 3,317 points in the Geekbench 6 benchmark was imperceptibly behind Acer's 3,348 points, it handily outclassed the iBuyPower’s older Ryzen 9 7900X and its 2,953 points. But it truly showed its strength in multi-core, with 22,485 points to Acer’s 18,282 and iBuyPower’s 17,226.

CyberPower slotted into the middle of our 25GB file transfer test, with 1,616.52 MBps to lead iBuyPower’s 1,501.6 MBps and trail Acer’s 1,719.57 MBps.

Last, this CyberPower dominated our 4K-to-1080p Handbrake test, completing the transcode in just 1 minute and 45 seconds. The iBuyPower (2:29) and Acer (2:51) were well off its pace.

Keyboard and Mouse with the CyberPowerPC Gaming Desktop

The GXi3800BSTV2 includes RGB-centric wired peripherals that might tide you over if you’re sticking to a budget. The CyberPower Nohi 02 keyboard is a basic desktop keyboard with a few notable extras. Though it’s described as a mechanical-like feel, its rubbery keypresses say otherwise. I nonetheless found it comfortable for extended typing and gaming. The layout is as expected and there's a handy volume scroller at top left.

The keyboard's RGB backlighting is bright and vibrant. No peripheral software is included, but you can switch between different lighting patterns using Fn and one of the number keys (1-9). Effects include a respectable assortment of wave, color shift, sweep, solid color, and several breathing variations. Lighting can be toggled with the key next to right Alt, effect speed with Fn + plus or minus, and brightness with Fn + Page Up/Down.

The CyberPower Elite M2 wired mouse is also basic. It’s only for right-handers, featuring two buttons on the left side. There are five sensitivity settings – the DPI numbers aren’t published – which you toggle pressing the button behind the scroll wheel. Average in size, the mouse fits my medium hands well and offers tactile, engaging clicks. The RGB effects illuminate the palm reset logo and the semicircle around the palm rest. Various color cycling patterns are available – simply hold down the DPI button to switch between them.

Software and Warranty on the CyberPowerPC Gaming Desktop

CyberPower’s software loadout is minimal. Asus Armoury Crate provides hardware monitoring and software overclocking but little else – the RGB devices in this desktop are only recognized as an addressable strip, and can’t be controlled there. (That’s what the knobs on the tower are for.) Otherwise, besides the default Windows 11 apps, there’s no extra software.

This CyberPower from Best Buy carries a one-year parts and labor warranty while systems purchased from CyberPower include two years of parts and three years of labor.

CyberPowerPC Gaming Desktop Configurations

Our CyberPower review unit features a Core Ultra 7 270K Plus processor with 360 mm liquid cooling, an Nvidia RTX 5070 graphics card, 32GB of RAM, a 2TB SSD, and Windows 11 Home for $2,309.99 from Best Buy.

Best Buy didn’t have any other CyberPower desktops with the MA-01 case, though CyberPower offered many configurations starting at $1,699, which gets you a Ryzen 7 9700X and an RTX 5060 Ti. The top preconfigured model ran $3,815 with a Ryzen 9 9950X3D, an RTX 5080, 32GB of RAM, and a 4TB SSD. CyberPower also offers built-to-order models starting at $1,245.

Competing desktops with an RTX 5070 bracketed our CyberPower review unit’s price. At Best Buy, an Asus ROG G700 was $2,409.99 with a Core Ultra 7 265KF while an HP Omen 35L was $1,999, though it dropped to a Ryzen 7 8700F and a 1TB SSD.

To compare CyberPower’s prices to what it would cost to build a similar PC yourself, I mock-configured several systems using the AI PC builder on Newegg, landing within $100 in most instances. That's a pretty nominal premium to have the system built and ready to go.

Bottom Line

CyberPower’s GXi3800BSTV2 earns its place among the better prebuilt gaming desktops. The company's MA-01 case gives it a distinctive look, and the analog lighting controls mean you don't need software to dial in a unique theme. Equipped with a Core Ultra 7 270K Plus, 360 mm liquid cooling, and an RTX 5070, it produces strong gaming performance without getting too loud. You even get decent peripherals.

Some concessions were made to hit its price: the case relies on plastic instead of tempered glass, wireless is limited to Wi-Fi 6, and there are no Thunderbolt 4/USB4 ports. Even so, it still makes for a compelling value, even next to DIY builds. For gamers who want a flashy tower with the performance to match – and the convenience of a full-system warranty — the GXi3800BSTV2 is an easy system to recommend.

Correction, July 1, 2026: CyberPowerPC tells Tom's Hardware that the viewing panel on the MA-01 is tempered glass, not plastic. We have addressed this, but still feel that the window did not seem substantial.

Dare I say it, prices on the best OLED gaming monitors are trending downward. They’ve been trickling down for a while, but recently I’ve received some new models that have lopped more than $100 off the prices of their predecessors. You’ll still pay more than an equivalent-spec LCD, but not a ton more.

ViewSonic has always delivered good value, and its latest entry is the VX2738-2K OLED. This 27-inch QHD 2560x1440-pixel panel boasts Quantum Dot color, 240 Hz, Adaptive-Sync and HDR for just $450 as I write this. Let’s take a look.

ViewSonic VX2738-2K OLED Specs

The VX2738-2K uses an OLED panel fronted by a Quantum Dot layer to deliver almost 109% DCI-P3 coverage, according to my measurements. That’s about average for the QD-OLEDs I’ve tested, and it is visibly more colorful than OLEDs or LCDs without the tech. Brightness is claimed to be 200 nits, which I verified with a full-field pattern, and I measured over 300 nits from a 25% window with variable brightness engaged. HDR reaches peaks over 400 nits, earning the VX2738-2K VESA DisplayHDR 400 certification.

Gaming performance is given equal attention with a 240 Hz refresh rate and Adaptive-Sync. The VX2738-2K is certified by both AMD and Nvidia for tear-free operation. Being an OLED, it has super quick response as well. Motion resolution is perfect at speeds above 180 fps. And you can achieve that without spending a fortune on a high-end graphics card.

Despite its attractive price, the VX2738-2K isn’t a barebones monitor. You get four video inputs, two HDMI 2.1 and two DisplayPort 1.4 with Display Stream Compression (DSC), along with a headphone jack and internal speakers. The only things left out are USB ports and LED lighting. But you do get a nice, solid stand and a premium DP cable. I also found an aiming point that turns on with a down-click of the OSD joystick. And there are plenty of OLED care options to keep the panel safe, along with variable brightness for SDR and HDR content.

ViewSonic is offering a solid display with a stunning picture and excellent gaming performance for $450. It wasn’t long ago that a 240 Hz QHD LCD cost that much. The era of affordable OLEDs may indeed be upon us.

Assembly and Accessories

ViewSonic doesn’t ship the VX2738-2K in look-at-me packaging, opting instead for a plain brown wrapper that just says, “ViewSonic Gaming.” The contents are encased in crumbly foam with the base, upright, and panel going together without tools. Build quality is high, and you get a three-year warranty too. The accessory bundle includes an external power supply and a DisplayPort cable.

Product 360

The VX2738-2K features angular styling with simple shapes and slightly rounded corners. The front is all screen with a narrow flush bezel that ranges in width from eight to 23 millimeters. A badge graces the lower left corner, indicating Nvidia G-Sync certification. It also meets FreeSync Premium standards. Other bits include a small ViewSonic logo and a tiny power status LED.

In the back are more triangles and a central box containing the internals. A joystick operates the OSD along with a separate power toggle button. The stand snaps into place over a 100mm VESA mount. If you want to use an arm, ViewSonic provides four large fasteners that you can tighten by hand if you like.

The stand has full ergonomics with 5/20 degrees tilt, 45 degrees swivel, and a 90-degree portrait mode. The OSD can be rotated manually as well. The height range is 5.1 inches (130mm). Movements are smooth and firm, better than what you’ll find with most budget screens. The package with stand weighs 12 pounds and sits on a thick metal plate base.

Underneath are four video inputs, two HDMI 2.1 and two DisplayPort 1.4 that support Display Stream Compression (DSC), which keeps frame rates at their maximum potential. There are a pair of two-watt internal speakers, and you get a 3.5mm headphone jack.

OSD Features

The VX2738-2K’s OSD is very business-like with six sub-menu headers across the top. It’s reasonably intuitive and has plenty of picture modes and image options to tailor things to your liking.

The fun starts with seven picture modes. The Game mode includes four additional presets, one of which includes a black stabilization slider. This can be increased to brighten shadow areas for better visibility. While it makes the picture less dramatic, it helps you find hidden enemies more easily. Standard is the default mode, and it includes everything needed for a precise calibration. And that’s a good thing because the VX2738-2K isn’t quite on spec out of the box, more on that later.

There are five fixed color temps and a user color with RGB sliders. Don’t be fooled by the one labeled sRGB. It does not change the color gamut from the native space, which covers almost 109% of DCI-P3. You’ll see the VX2738-2K’s full color for all content, be it SDR or HDR. You also get three gamma presets. HDR has no options, but you can turn off the effect if you like. Leaving it on Auto means the VX2738-2K switches without user intervention.

Manual Image Adjust is where you’ll find OLED care options like pixel orbiting, screen refresh and static detection. The one labeled OLED APL is a variable brightness setting. Turning it on raises the peak white level for small areas of the screen.

The Setup menu includes an Adaptive-Sync toggle, OSD appearance options, and power settings. The second screen includes PIP and PBP for viewing two sources at once.

ViewSonic VX2738-2K OLED Calibration Settings

In the VX2738-2K’s Standard mode, I observed a slight purple tint in grayscale patterns and very dark gamma, which gave the image a murky appearance. Fixing the RGB sliders took care of the color, but gamma could only be corrected by turning the OLED APL option on. That means you’ll see variable brightness all the time. While this isn’t my favorite approach, ViewSonic’s VB isn’t super aggressive so there is no obvious image pumping or rapid changes in brightness. It is the way to the best image, so I recommend using it. It’s locked on by default for HDR content. Below are my recommended SDR settings. HDR has no provision for adjustment; just set the HDR option to Auto for an automatic switch.

Gaming and Hands-on

There is no activity that cannot be enhanced by an OLED monitor and the VX2738-2K proved that to me yet again. The broad contrast and saturated color offered by this technology cannot be completely duplicated by LCD, and that includes Mini LED panels.

The VX2738-2K isn’t the brightest OLED I’ve reviewed, but it is more than bright enough to be satisfying. I found that the variable brightness option was required for the best picture. Usually, I use constant brightness for work tasks and variable for entertainment. With OLED APL turned off, the picture was dull in comparison to other OLEDs. This turned out to be due to dark gamma, which you can learn more about on page four. Turning it on addressed my concerns. And it wasn’t aggressive in operation, there were no artifacts like pumping or changes in brightness that lagged behind the moving image. I left it engaged for all content. And it’s locked to on for HDR.

HDR looked spectacular as all OLEDs do. Color was rich and vibrant thanks to the Quantum Dot layer. It’s awesome to get this tech in a $450 monitor. Accuracy was excellent once I had calibrated. A few tweaks to the RGB slider took color, grayscale and gamma to reference-level. That showed in all areas, gaming, video and productivity.

Video processing was exemplary and with frame rates staying at or near the 240 Hz maximum, moving objects were rendered with perfect clarity and full detail. There was no smearing or blur of any kind. The VX2738-2K was extremely responsive to my inputs as well. You’ll see how good it is on the next page as it’s one of the fastest 240 Hz screens I’ve tested to date. To get visibly better performance, you’ll need to go to at least 360 Hz. And only the most skilled gamers would be able to tell a difference. My abilities are not at that level. I would happily frag all day and twice on Sunday. The VX2738-2K has that addictive quality.

For daily use, the VX2738-2K serves extremely well. 27 inches is ideal for QHD resolution; it’s a reasonably pixel-dense 109ppi but not so much that frame rates suffer. My GeForce RTX 4090 was more than stout enough to keep speeds at 240fps with games set to their highest detail levels. And the dot structure is invisible from more than two feet away. You can sit close when working in Photoshop or Premiere Pro. Or relax and sit back when watching videos on YouTube.

My only complaint is a small one. I was glad to see the inclusion of internal speakers but the VX2738-2K’s are quite tinny. At 50% volume, they were reasonably loud but somewhat distorted in the highest frequencies, which made them harsh. They’re fine for system sounds but for music or game audio, they don’t offer much impact or detail. I suggest plugging in your favorite headphones to the provided 3.5mm jack. Speaking of plugs, I noted the inclusion of four video inputs instead of the usual three with an extra DisplayPort in the mix, nice.

Takeaway: The VX2738-2K is an extremely versatile monitor with excellent gaming performance and enough brightness, color saturation and clarity to suit it for photo editing, web browsing or any computing task. It delivers fast and smooth video processing along with accurate and vivid color once calibrated. The internal speakers aren’t the greatest but plug in a pair of headphones and you’ll be fine. Above all, it’s a superlative value at just $450 and that’s the best part.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

27-inch QHD OLEDs come in a wide variety of refresh rates, up to 540 Hz in fact. I tried to find screens closer in speed to the VX2738-2K and came up with Alienware’s AW2726DM and AW2725D, Gigabyte’s MO27Q28G, Asus’ XG27AQWMG, and AOC’s Q27G4ZD.

Pixel Response and Input Lag

Click here to read up on our pixel response and input lag testing procedures.

I didn’t have enough 27-inch QHD OLEDs to compare the VX2738-2K’s panel response, so I have a different group there. You can see that refresh rate doesn’t always go in lockstep with response time. The quickest screen is a 240 Hz Acer at 0.12ms, and the ViewSonic isn’t far behind at 0.20ms. Even at the slowest 0.24ms, you won’t be able to see any differences at speeds above 180fps. The Acer would have a slight edge between 150 and 180 fps. But these are tiny issues that are nearly impossible to see with the naked eye.

In the lag test, the VX2738-2K proved to be one of the faster screens available, with a total lag time of just 15.7ms averaged over 25 mouse clicks using Nvidia’s LDAT tester and utility. 240 Hz is in no way a limitation when compared to the 280 Hz screens in the group, which are a tad slower. Again, this is a minute difference that anyone outside the Jedi Order will be unable to distinguish.

Test Takeaway: The VX2738-2K is very quick with low input lag and smooth motion resolution. At frame rates above 180fps, moving objects render with the same clarity as stationary ones. It is quicker than at least two monitors with higher refresh rates. It’s safe to say that if you want to beat the VX2738-2K, you’ll need to go at least 360 Hz or faster.

Viewing Angles

Many Quantum Dot monitors show a slight red tint in the 45-degree side photo, but not the VX2738-2K. It looks the same as the head-on image with consistent color, brightness and gamma. The top view has slightly reduced gamma and luminance. This is excellent performance.

Screen Uniformity

To learn how we measure screen uniformity, click here.

When I first measured the VX2738-2K’s screen uniformity, it was an unimpressive 13.55%, which is much higher than average for an OLED. Some monitors need an initial pixel refresh cycle to look their best, so I ran the procedure and saw a significant improvement to 2.55%. I have encountered this issue before with Samsung monitors, so ViewSonic is not alone. And once I had done it, the uniformity value did not change. Once set up, the VX2738-2K delivers excellent performance.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

To read about our monitor tests in-depth, please check out Display Testing Explained: How We Test PC Monitors. We cover brightness and contrast testing on page two.

Uncalibrated – Maximum Backlight Level

ViewSonic rates the VX2738-2K at 200 nits. This tracks when measuring a full field pattern with OLED APL (variable brightness) turned off. To see its full potential, I turned it on and measured a 25% window for 381 nits. This also improves gamma, which in turn improves image quality by quite a bit. I’ll show more detailed tests on the next page. Black levels and contrast cannot be measured.

After Calibration to 200 nits

Calibration makes no difference to the contrast result. I set 200 nits based on a full field pattern so that meant brightness was turned up all the way. This maximizes the VX2738-2K’s image quality and fidelity. Black levels, static contrast and ANSI contrast cannot be measured.

Test Takeaway: The VX2738-2K delivers contrast performance that’s equal to any OLED monitor. With infinite black levels, the picture has a depth and quality that no LCD can match. To get the best result, I recommend leaving OLED APL (variable brightness) on all the time. It’s not aggressive in operation, so it is appropriate for productivity and color-critical tasks.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

For the grayscale, gamma and color tests, I used the VX2738-2K’s Standard picture mode, which is the default. It offers color temp presets, RGB sliders and gamma presets.

Grayscale and Gamma Tracking

Our grayscale and gamma tests use Calman calibration software from Portrait Displays. We describe our grayscale and gamma tests in detail here.

You can see in the first chart what I meant earlier by skewed gamma. Grayscale tracking isn’t too far off, with just a bit of purple tint seen in steps above 50% brightness. But gamma is dark enough to make the picture look flat and undefined. I normally leave variable brightness off for all tests and calibration, but in this case, its use is necessary.

Turning it on and tweaking the RGB sliders makes the VX2738-2K a whole lot better. All grayscale errors are below 1dE, and gamma is visibly and measurably perfect. This is an improvement that is easy to see in an A/B comparison. With the proper settings, the VX2738-2K is capable of reference-level performance.

Comparisons

With a starting value of 5.01dE, the VX2738-2K needs some attention to look its best. In this group, only the AOC and Asus displays can be enjoyed without calibration. Once adjusted, all screens have perfect grayscale tracking. There would be no visible difference between them. In the gamma test, the VX2738-2K excels with a tiny 0.08 range of values and a perfect average of 2.2, 0% deviation. This is excellent performance.

Color Gamut Accuracy

Our color gamut and volume testing use Portrait Displays’ Calman software. For details on our color gamut testing and volume calculations, click here.

You can see how the VX2738-2K’s inaccurate gamma affects color in the first gamut chart. Every color is almost 20% oversaturated, with red looking a bit pink and cyan being too blue in tone. These are clearly visible errors.

With OLED APL turned on and the grayscale calibration in place, color is far better than before, with just some slight oversaturation in red and magenta that is well within visually tolerable limits.

Comparisons

The VX2738-2K’s gamut result goes from 4.49dE to 1.44dE with calibration, which is an obvious improvement. You’ll need to turn OLED APL on and tweak the RGB sliders to achieve this. If you don’t have the means, put in my settings from page one and you’ll be very close to the above result.

With a gamut volume of 108.86% of DCI-P3, the VX2738-2K is average among Quantum Dot OLEDs. The extra three-ish percent shown by the Alienware displays would be hard to spot in a side-by-side comparison. But it’s easy to tell the ViewSonic from a monitor without a QD layer. The sRGB volume is huge at 161.41%. This is because the VX2738-2K has no sRGB mode. There is an sRGB option in the color temp menu, but it doesn’t change the gamut size, and there is no other color space selection in the OSD. You can use it for a DCI-P3 reference only.

Test Takeaway: The VX2738-2K is very color accurate once calibrated. Out of the box, it has a slight purple tint and dark gamma. You must turn on the OLED APL option to fix gamma and adjust the RGB sliders for neutral grayscale tracking. Color covers almost 109% of DCI-P3, which is about average for Quantum Dot displays. And users needing the smaller sRGB gamut for color-critical work won’t find it here.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

Our HDR benchmarking uses Portrait Displays’ Calman software. To learn about our HDR testing, see our breakdown of how we test PC monitors.

The VX2738-2K switches to HDR mode automatically if the HDR option is set to Auto in the OSD. There are no adjustments or additional picture modes available.

HDR Brightness and Contrast

The VX2738-2K delivers its brightest possible picture in HDR mode with 419 nits measured from a 25% window pattern. Variable brightness is in operation and cannot be turned off. It easily qualifies for VESA DisplayHDR 400 like much of its competition. Asus and Gigabyte drive their panels a bit harder to get 580 and 653 nits peak, respectively. The VX2738-2K has decent impact with deep blacks and infinite contrast, but is not exceptionally bright.

Grayscale, EOTF and Color

The VX2738-2K’s HDR color is fixed with no provision for adjustment. Grayscale tracking is excellent in the darker parts of the image where there are no visible errors. At levels above 50%, it gets a bit warm in tone. These errors do not significantly diminish image quality. It would be nice to have the RGB sliders available, though. The EOTF runs a bit dark from zero to 40%, then stays close to the reference and a 65% tone-map transition point, which is correct given the measured black and white thresholds.

In the color tests, the VX2738-2K tracks hue points well with only slight variation in magenta. The chart is generally oversaturated like nearly all HDR monitors but with points that progress in a linear fashion. This approach keeps fine detail sharp and clear in all parts of the image. DCI-P3 is fully covered as it should be. BT.2020 runs the VX2738-2K out of color at 90% red, 80% green and 95% blue.

Test Takeaway: The VX2738-2K delivers an HDR image of average brightness with infinite black levels and contrast like any OLED display. With 419 nits peak, it has a satisfying impact and makes an obvious difference between SDR and HDR. Color is nicely saturated and accurate enough to be pleasing. No one will complain about the VX2738-2K’s HDR quality, it is very high. I noted slight grayscale errors that could not be corrected, but this did not affect the image in any significant way.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

There was a time when I couldn’t be sure desktop OLED monitors would ever exist. The tech only came in exotic TVs that cost $6,000 for a 55-inch panel. Today, they’re almost as common as bison were in the Wild West. And with the quantity of choice comes cost savings. Prices only trickled down for a time, but now we’re seeing some real values hitting the market. The ViewSonic VX2738-2K is a high-performance gaming OLED with Quantum Dot technology for $450. That’s almost all you need to know for a purchase decision.

Despite its low price, the VX2738-2K doesn’t skimp on image quality or performance. Though I noted some gamma anomalies out of the box, it just took the right settings and a quick calibration to bring it to reference-level. I achieved visually perfect color, grayscale and gamma with little effort. And with the extra color afforded by Quantum Dots, it delivered a great SDR and HDR image. The only miss there is the lack of an sRGB mode. Most users won’t care since they’ll prefer more color even if it isn’t strictly correct. But photographers needing that gamut choice won’t find it here.

Gamers definitely won’t complain about the VX2738-2K’s smooth motion response and low input lag. It’s one of the quickest 240 Hz monitors I’ve tested and it keeps up with its higher refresh rate competition. You’ll need to buy 360 Hz or more to beat it and that victory will be a slim one. Competition gamers looking to add a value-priced monitor to their toolkit should have a look.

The ViewSonic VX2783-2K OLED breaks price barriers by offering a high-performance gaming OLED display for $450. In perspective, that’s about what you’d have spent on an equivalent LCD a few years ago. That’s progress for sure. If you have been waiting for OLED prices to drop, now might be the time to pull the trigger. Definitely check it out.

MORE: Best Gaming Monitors

MORE: How We Test PC Monitors

MORE: How to Buy a PC Monitor

What we have today is the WD Black SN8100 by another name, with one very important and exciting change: it’s 8TB. This puts the SanDisk Optimus GX Pro 8100 on an entirely different plane, as it delivers a capacity everyone wants with insanely fast hardware. The 2TB Black SN8100 that we reviewed previously was a pleasant surprise given its high power efficiency, but its absurdly low random read latency remains the real draw. Can all this goodness be repeated at 8TB with a new SKU firmly under the SanDisk name? And will anyone be able to afford it?

The previous 8TB champion, the WD Black SN850X, still remains interesting from a value perspective – if you can call current prices a value in any circumstance – but we’ve had a hard time finding high-end PCIe 5.0 drives at this capacity. The only one we’ve reviewed is the 8TB Samsung 9100 Pro, which, if you missed it when it was on sale, is probably promoting a stinging sensation right now. We’ve been promised an 8TB SKU of the Kingston Fury Renegade G5 – which has the same hardware as the Black SN8100 and Optimus GX Pro 8100 – and 8TB versions of Phison E28 drives, too. So far, these promises haven't materialized. The good news is, if you like the Optimus GX Pro 8100 at 8TB, then you can pick it up in its Black SN8100 form if that happens to be the better bargain.

Whether or not it’s a deal at all is up to you, but it’s hard to beat this level of performance at 8TB. The drive is power-efficient, too, and we have some faith in WD’s and SanDisk’s reliability and willingness to offer support. It’s probably best to buy something you know than roll the dice with unknown hardware, even if the latter comes with some price savings. We think the Optimus GX Pro 8100 qualifies as a better-known option, and given the level of investment at 8TB, that’s the safest way to go, even if there were other options.

SanDisk Optimus GX Pro 8100 Specifications

The SanDisk Optimus GX Pro 8100 is available in the same capacities as the WD Black SN8100: 1TB, 2TB, 4TB, and 8TB. At the time of our Black SN8100 review, the 8TB model was not available, but it is now.

The current price disparity is worth a mention. The Optimus GX Pro 8100 is, right now, priced at $349.99, $694.63, $1,339.99, and $2,799.99. This is significantly higher than the Black SN8100’s $259.99, $439.99, $789.99, and $1,799.99. These prices are without heatsinks – check the table for the Optimus GX Pro 8100’s heatsink price per SKU, which only makes sense at 1TB, as is the case with the Black SN8100. You’re better off providing your own heatsink in most cases, and going for the Black SN8100 instead of the Optimus GX Pro 8100 makes sense until the price of the latter comes down to reality. Alternatives in this price range include the Kingston Fury Renegade G5 and the Samsung 9100 Pro, both of which also have 8TB SKUs, although we’ve only reviewed the latter at that capacity.

The drive’s maximum performance is impressive, with bandwidth up to 14,900 / 14,000 MB/s for sequential reads and writes and up to 2,300K / 2,400K random read and write IOPS. This is about as good as it gets. This level of performance is, on average, attainable with less than 7W of power consumption – a far cry from the 11.5W+ we saw on the earliest high-end PCIe 5.0 SSDs – which makes the drive even more impressive.

SanDisk includes support for TCG Opal 2.02 for encryption. In addition, the drive is backed up to 600TB of writes per TB of capacity within the five-year warranty period. This is standard and comes out to 4,800TB or 4.8PB of writes for the 8TB sample we have today.

SanDisk Optimus GX Pro 8100 Software and Accessories

SanDisk features two pieces of software for the Optimus GX Pro 8100: the SanDisk Dashboard, based on WD’s Dashboard, and Acronis True Image for SanDisk. Both applications support other products, and SanDisk also has other software that applies to other products, so check the Software Downloads page if you own a different drive. While the OEM Acronis package has obvious applications – it’s for backing up, cloning, and recovering files – the Dashboard may be more involved for the average user. This SSD toolbox helps you stay on top of your drive’s health, it covers optional features, allows you to update the firmware, and more.

SanDisk Optimus GX Pro 8100: A Closer Look

The 8TB Optimus GX Pro 8100 is a double-sided drive. At all other capacities, it is single-sided. If you’re going for the heatsinked version of the drive, this doesn’t matter, though. In rare cases, you might have a system that needs or prefers a single-sided drive, in which case it’s worth being aware of the fact that you will be limited to 4TB.

As to why the drive is double-sided only at 8TB, that has to do with how many NAND flash packages can fit on the PCB. With the controller and DRAM on the top side, there is only room for two flash packages. Each package can usually have a maximum of sixteen dies. Each die is 1Tb, or 128GB. Therefore, each package contains a maximum of 2TB of flash. Since four packages are necessary for 8TB, the 8TB SKU is necessarily double-sided. There will be exceptions to this on drives that can fit four flash packages per side – DRAM-less drives, and especially ones with only four flash channels, as the controllers may be smaller, or drives with denser flash. 2Tb QLC flash is certainly a possibility, although for the time being, such flash would be prioritized for enterprise.

The only other interesting information here is the power rating. 3.3V / 2.7A puts the power ceiling around 9W, which is more than ample for the average power numbers that SanDisk lists on the specifications sheet. We should point out that those numbers are averages, and for reads and writes separately. In fact, the drive via SMART is rated for active power draw of 8.9W – exactly where 3.3V / 2.7A hits – with a potential peak of 9.5W. Active power could include mixed workloads, that is, both reads and writes, and as that’s what we test, we would expect our maximum value to be closer to that active power number.

The drive has an SSD controller, power management circuitry, two DRAM packages, and four NAND flash packages. The controller is branded SanDisk with the “A101” being telltale, but the “2508” in the middle tells us the truth: this is a Silicon Motion SM2508 and not a proprietary solution. This is a high-end PCIe 5.0 controller with eight flash channels and DRAM support. It’s newer than the original PCIe 5.0 controller, the Phison E26, and is made in a smaller process node. As a result, it is significantly more power-efficient, which is a very good thing considering that some E26 drives came with active cooling. WD and SanDisk do tweak the controller firmware and use nCache 4.0 to separate the drive from more generic designs.

The DRAM is labeled D8CKD, and each package is thus 32 GB or 4GB of DDR4 in a 16-bit configuration. This is 8GB total, which maintains the 1GB:1TB DRAM:NAND flash ratio we expect for optimal performance. You could use more efficient DRAM, but the power savings would be relatively small. The flash – labeled in part with 2527 for the 27th week of 2025, exactly matching the controller’s 2527 – is 1Tb BiCS8 TLC. This flash has proven to be power-efficient with exceptionally low random read latency. There’s a reason we find the Black SN8100 hard to beat, and that’s with this combination of characteristics.

MORE: Best SSDs

MORE: Best External SSDs

MORE: Best SSD for the Steam Deck

Comparison Products

We’ll be comparing the 8TB Optimus GX Pro 8100 to the 2TB WD Black SN8100 – this is effectively the same drive, but at a lower capacity – as well as the 8TB Samsung 9100 Pro, perhaps the best known high-end PCIe 5.0 SSD available at 8TB. The Kingston Fury Renegade, which also uses the same hardware as the Optimus GX Pro 8100 and Black SN8100, is available at 8TB but reviewed at 2TB. It does perform differently from the other two drives, as WD and SanDisk have custom optimization.

We’ve also thrown in the Crucial T710, with the same controller but Micron’s 276-Layer TLC flash instead of BiCS8, and the Phison E26-based Corsair MP700 Pro SE with Micron’s 232-Layer TLC. This latter flash is used with the Silicon Motion SM2508 controller – the same as the Optimus GX Pro 8100 uses – on the Lexar NM1090 Pro and other “budget” high-end PCIe 5.0 drives. Lower-end PCIe 5.0 drives include the Addlink G55 and Crucial P510, included to illustrate the gap between the mid-range and the high end.

This gives a good set of drives for comparison with two areas lacking: PCIe 4.0 drives, which we really don’t think can compare to a monstrous drive like this, and Phison E28-based drives, since we haven’t really seen those on the market and never at 8TB. We’re comparing drives that exist on the market or have been generally available to account for real-world market conditions.

Trace Testing — 3DMark Storage Benchmark

Built for gamers, 3DMark’s Storage Benchmark focuses on real-world gaming performance. Each round in this benchmark stresses storage based on gaming activities, including loading games, saving progress, installing game files, and recording gameplay video streams. Future gaming benchmarks will be DirectStorage-inclusive, and an evaluation for future-proofing is included where applicable.

We just spoke about the excellent latency we see with the Optimus GX Pro 8100’s BiCS8 flash, and here we can see the flash in action. The drive is just as fast as the Black SN8100 – unsurprising, as it uses the same hardware, which puts it above every other drive we’ve tested. The drive does appear to be a little bit slower at 8TB when factoring in bandwidth, but this is also unsurprising. Drives with an excess of flash will have more overhead and perform more weakly. However, this is still an incredibly fast drive on par with something like the Corsair MP700 Pro XT, a superfast drive we don’t have on this list because frankly, we haven’t seen the drive available and certainly not at 8TB.

The only thing we can add here is some discussion about using a drive like this for gaming. In fact, SanDisk does advertise the Game Mode, and we think that’s worth mentioning. The SanDisk Dashboard lets you enable, disable, or use auto-detection for Game Mode, a feature that basically puts or keeps the drive in a higher power state for superior readiness. We can loosely estimate what this will mean for load-time improvement, as we can see the enter and exit latencies in SMART. For this drive, from idle to load, these are 1,500µs and 8,500µs or 1.5ms and 8.5ms, if you prefer. Very small amounts of time in the grand scheme of things, and you will see real gains about an order of magnitude better…which is still really small.

As it so happens, SanDisk does show the difference in Final Fantasy XIV. For the 8TB drive, the average improvement is 0.05s, which is on the order of ~1%. This is on an Intel system, for those keeping count. AMD and Intel platforms will perform differently. Naturally, this result is not super impressive, but some systems and games will see larger improvements. We think for desktop use, and if this is your primary drive – a drive you use for everything, especially then Game Mode could be a free tweak. Just don’t set your expectations too high.

Trace Testing — PCMark 10 Storage Benchmark

PCMark 10 is an industry standard trace-based benchmark that uses a wide-ranging set of real-world traces from popular applications and everyday tasks to measure the performance of storage devices. The results are particularly useful when analyzing drives for their use as primary/boot storage devices and in work environments.

PCMark 10 performance is phenomenal. There is no performance drop evident from going up to 8TB from 2TB, either. This is a very expensive drive – the 8TB Black SN8100 might be the better bet, but it is still expensive – and it’s worth every penny if you care about having the absolute highest levels of performance combined with capacity. This is a much more expensive proposition these days than it was a year or two ago, and, usually, we would recommend putting that money elsewhere in your build. Unfortunately, other hardware components, such as RAM, have also seen massive price increases. We don’t expect SSD prices to come down soon, either, so prepare yourself for a serious investment if this is the way you want to go.

Console Testing — PlayStation 5 Transfers

The PlayStation 5 is capable of taking one additional PCIe 4.0 or faster SSD for extra game storage. While any 4.0 drive will technically work, Sony recommends drives that can deliver at least 5,500 MB/s of sequential read bandwidth for optimal performance. Based on our extensive testing, PCIe 5.0 SSDs don’t bring much to the table and generally shouldn’t be used in the PS5, especially as they may require additional cooling. Check our Best PS5 SSDs article for more information.

Our testing utilizes the PS5’s internal storage test and manual read/write tests with over 192GB of data, both from and to the internal storage. Throttling is prevented where possible to see how each drive operates under ideal conditions. While game load times should not deviate much from drive to drive, our results can indicate which drives may be more responsive in long-term use.

As expected, the Optimus GX Pro 8100 performs well in the PS5. This drive is overkill for the console in every possible way, but it will deliver a good experience if you decide to buy it for this use case. There is some argument to be made for an 8TB PS5 drive – you can’t add multiple internal drives as you can on a desktop PC, and maybe you want extreme capacity, too. This drive running at PCIe 4.0 speeds will also be pretty efficient, which is a nice bonus. Conceivably, you might use the drive in your PS5 temporarily, future-proofing, or making an investment in the drive. Regardless of your potential reasons, we still recommend reserving this drive for a high-end desktop.

Transfer Rates — DiskBench

We use the DiskBench storage benchmarking tool to test file transfer performance with a custom 50GB dataset. We write 31,227 files of various types, such as pictures, PDFs, and videos to the test drive, then make a copy of that data to a new folder, and follow up with a reading test of a newly-written 6.5GB zip file. This is a real-world type workload that fits into the cache of most drives.

With DiskBench, we’re primarily looking at the copy transfer rate to get a good feel for real-world mixed I/O performance. The Optimus GX Pro 8100 matches the Fury Renegade G5 almost exactly, which, considering they use the same hardware, makes sense. WD and SanDisk get more performance out of this hardware thanks to proprietary optimizations; you can see this with the Black SN8100’s superior result, which again makes for a very fast drive. Drives with Phison’s E28 controller, like the aforementioned Corsair MP700 Pro XT, are faster yet, but are just not readily available at this time. We consider the Black SN8100, and therefore the Optimus GX Pro 8100, as being the fastest drives you can regularly buy right now.

We should emphasize that the latter drive is slower than the former in this specific test because we’re comparing a 2TB to 8TB. An 8TB drive can have more overhead with more flash dies. You can see from the specifications table that the 8TB drive pulls more power, for example, and ultimately, performance takes many factors, including power consumption or heat dissipation, into consideration as the controller is always trying to optimize multiple things simultaneously.

Also, write performance is contingent on pSLC caching, which is not precisely the same at different capacities. The cache will be relatively larger at 8TB than 2TB, but with very large caches, as are used with SanDisk’s nCache 4.0, it is likely that the average write speed at 8TB will be somewhat reduced. The controller has to juggle multiple objectives at once, like incoming writes, already-written writes that have to be moved over from the cache, interruptive reads, wear-leveling, etc., which is a job that becomes more complex with 64 flash dies. As a result, slightly lower performance here and in our write saturation test is expected.

Synthetic Testing — ATTO / CrystalDiskMark

ATTO and CrystalDiskMark (CDM) are free and easy-to-use storage benchmarking tools that SSD vendors commonly use to assign performance specifications to their products. Both of these tools give us insight into how each device handles different file sizes and at different queue depths for both sequential and random workloads.

Let’s start with ATTO and take a look at reads, as writes are pretty stable with this drive. For reads, we see the MP700 Pro SE, the only drive using the Phison E26 controller here, doing the best below 4KiB. We’ve pointed this out before, as many E26 drives, the MP700 Pro SE included, have twice as much volatile memory (DRAM) as normal. That is, this 4TB drive has 8GB of DRAM. We posited that this might be because the E26 had enterprise origins, and the one case you’d need more memory is if you went with a smaller granularity for mapping or had other, heavier metadata requirements.

To cover the technicals quickly, most drives opt to map 4KiB logical pages with 32-bit (4-byte) addressing, which delivers the 1GB:1TB DRAM:NAND flash ratio. This ignores compression and non-mapping metadata, but it is decently representative of the general requirement. If you need to map with smaller granularity, many enterprise workloads demand 2KiB I/O, which is one reason Samsung’s original Z-NAND operates in both 2KiB and 4KiB modes; then you want more memory. This may be one reason the E26 is better here. This is probably only interesting to you if you’re buying a high-end drive and specifically need enhanced performance in this area. Other metadata cases also need memory, but many algorithms, including those for wear-leveling, use block-level granularity, which requires far less memory given large modern block sizes.

The other area of note is 2MiB, where we see many, but not all, of the drives dip. In some cases, this might be because of superpage alignment or nuances of the flash, controller, or flash and controller combination. The NM1090 Pro, for example, doesn’t look too happy there, but the Kingston Fury Renegade G5 has no issue. A look at our logarithmic scale graph shows that the differences between the affected drives are otherwise exaggerated, which matches our real-world findings. Nevertheless, we would recommend a drive with newer flash, like BiCS8, 276-Layer Micron, or Samsung’s 236-Layer, if you want the most consistent performance across larger block sizes. If you’re dealing with media and larger files, that means you.

This translates to CDM, where the QD1 sequential read results for 1MB give some idea of real-world file transfer performance. The Optimus GX Pro 8100 suffers to some extent here, although that might be in part due to its high capacity. Worth noting is that the 8TB 9100 Pro doesn’t suffer in comparison to its 2TB peer, but we’re dealing with a different controller and different flash. Samsung specifically is very good at packaging, which could be one reason. Regardless, the Optimus GX Pro 8100 is able to recover at a higher queue depth, but that only pulls it even with the other top drives. Sequential writes are a different story as the drive handles itself well, and it outperforms the 8TB 9100 Pro at both queue depths. This makes it more interesting for some workloads where you might expect more writes.

Random 4KB latency is also good with random read latency, in particular, being off the chart. Other drives just can’t compare. If you thought the Black SN8100 was a fluke, think again. Drives with the same hardware, like the Fury Renegade G5, also can’t compete with WD’s and SanDisk’s optimization. This level of performance is stellar compared to what we considered the old cut-off for high performance at 45µs. The Optimus GX Pro 8100 is a bit slower than the Black SN8100, but all this really does is underline that these drives are clearly the fastest 8TB drives out there when it comes to what most people care about. For many enthusiasts, this result alone makes the drive the de facto choice.

Sustained Write Performance and Cache Recovery

Official write specifications are only part of the performance picture. Most SSDs implement a write cache, which is a fast area of pseudo-SLC (single-bit) programmed flash that absorbs incoming data. Sustained write speeds can suffer tremendously once the workload spills outside of the cache and into the "native" TLC (three-bit) or QLC (four-bit) flash. Performance can suffer even more if the drive is forced to fold, the process of migrating data out of the cache in order to free up space for further incoming data.

We use Iometer to hammer the SSD with sequential writes for 15 minutes to measure both the size of the write cache and performance after the cache is saturated. We also monitor cache recovery via multiple idle rounds. This process shows the performance of the drive in various states including the steady state write performance.

This is an 8TB drive, and an 8TB drive can have a massive pSLC cache. A 3-bit, TLC drive operating in a single-bit mode can have a cache up to one-third its capacity. For an 8TB drive, this is approximately 2.7TB if you’re going by maximum user space, but it can be larger if you’re going by raw flash. The reason we mention this is that it often appears as if WD’s or SanDisk’s nCache 4.0 is a full-drive, dynamic pSLC scheme, but in reality, there is a static portion, and, further, it seems that some amount of flash is left free to improve write consistency. Static pSLC is always available and operates separately from dynamic pSLC, which also improves write consistency. The sizes of each portion can differ with a hybrid scheme, as is the case with Samsung’s TurboWrite 2.0 on the 9100 Pro and other drives.

With that explanation out of the way, let’s look at the drive’s response. It writes at 13.1 GB/s for over 204 seconds with a 2,673GB cache. This is a little bit slower than the Black SN8100’s 13.2 GB/s. Once the cache is exhausted, the drive can write directly to the TLC flash at around 7.64 GB/s. This precisely matches expectations as it’s less than double the speed of the 2TB Black SN8100’s ~3.92 GB/s in this mode. We would expect up to double the speed with double the flash in parallel — the 2TB drive has 16 dies when you can parallelize up to double that amount with four dies per eight flash channels, reached at 4TB and not improved at 8TB. In fact, going up to 8TB can slightly reduce performance.

Finally, the drive runs out of free cache and is forced to wait for data to be moved over from pSLC to native TLC. This “folding” mode slows the drive down considerably. Our steady state write performance for the drive comes in at over 3.1 GB/s, which is less than one-half the TLC speed and, in fact, lower than what the 2TB Black SN8100 achieves. This is because having a large cache means you are fighting a losing battle with sufficiently long writes. You’re trading capacity for temporary speed, but eventually, that has to be paid back. The 8TB 9100 Pro, which also writes a bit slower than its 2TB version at steady state, is way behind the Optimus GX Pro 8100 at just 1.7 GB/s. This is somewhat misleading as the 9100 Pro has very consistent write performance, but it is undeniably slower. Jumping back and forth with write performance as cache is freed is not ideal on top of the added latency to reads during folding. Writing more slowly can improve endurance in part by deferring writes to avoid unnecessary rewrites, so speed is not everything.

Power Consumption and Temperature

We use the Quarch HD Programmable Power Module to gain a deeper understanding of power characteristics. Idle power consumption is an important aspect to consider, especially if you're looking for a laptop upgrade, as even the best ultrabooks can have mediocre stock storage in terms of capacity and performance. Desktops are often more performance-oriented with less support for power-saving features, so we show the worst-case scenario for idle.

Some SSDs can consume watts of power at idle while better-suited ones sip just milliwatts. Average workload power consumption and max consumption are two other aspects of power consumption, but performance-per-watt, or efficiency, is more important. A drive might consume more power during any given workload, but accomplishing a task faster allows the drive to drop into an idle state more quickly, ultimately saving energy.

For temperature recording, we currently poll the drive’s primary composite sensor during testing with a ~22°C ambient. Our testing is rigorous enough to heat the drive to a realistic ceiling temperature, but real-world temperatures will vary due to the environment and workload factors.

The Black SN8100 is an extremely efficient drive, so we had high hopes for the Optimus GX Pro 8100, too. Luckily, the drive remains very efficient even at 8TB, beating all but the newest, most efficient drives. It also easily clears the 8TH 9100 Pro, its foremost rival. At over 600 MB/s per watt, this drive can even be run without a heatsink.

In our testing, it reached 75°C, which is 15°C below the initial throttling state. We usually recommend at least 10°C of headroom, with 20°C being ideal. In-between means that you may want to put a heatsink on this drive in hotter systems to be on the safe side. However, that aside, being able to hammer an 8TB high-end PCIe 5.0 drive with writes at a reasonable temperature and being able to copy files efficiently is, frankly, astounding at this level of performance.

The Optimus GX Pro 8100 is simply peerless until we see more 8TB options on the market. We freely admit that we think the 8TB Fury Renegade G5 would give it a run for its money, but, on the whole, the Optimus GX Pro 8100 would probably win. There are no 8TB E28-based drives available, so we feel safe in saying that this is what peak performance looks like at this time.

Test Bench and Testing Notes

We use an Alder Lake platform with most background applications, such as indexing, Windows updates, and anti-virus, disabled in the OS to reduce run-to-run variability. Each SSD is prefilled to 50% capacity and tested as a secondary device. Unless noted, we use active cooling for all SSDs.

SanDisk Optimus GX Pro 8100 Bottom Line

The SanDisk Optimus GX Pro 8100 is simply the finest, fastest 8TB SSD on the market. Yes, it’s the same as the WD Black SN8100, but this is the one we have before us. You can’t go wrong with either drive. We think it would beat the 8TB Kingston Fury Renegade, given WD’s and SanDisk’s special sauce, and we feel it defeats the 8TB Samsung 9100 Pro overall. Your other option is the 8TB WD Black SN850X, which is, of course, only PCIe 4.0. Right now, it’s $300 less, or to put it another way, the Black SN8100 is 20% more. Is it worth the jump up? Honestly, yes, if you want maximum performance, but having the option of the Black SN850X is nice for older systems and the PS5. Many laptops can’t take advantage of PCIe 5.0 drives yet, either. On the other hand, if you’re dropping this much money, then maybe you should go all out and just future-proof.

The Optimus GX Pro 8100 shines almost everywhere it matters. It has class-leading random read latency, which, for many users, is the metric to use for real-world feel. It’s lightning fast even without the optional Game Mode enabled, although that makes the drive even more responsive. The drive is very power-efficient, especially compared to earlier PCIe 5.0 solutions. It has plenty of bandwidth, and SanDisk backs it with all the software you need.

The drive is not perfect, but it performs well enough in all of our tests that we would be hard-pressed to put any drive in front of it except maybe something like the Corsair MP700 Pro XT. Which would be fine if such a drive were purchaseable. 8TB drives based on Phison’s E28 controller are coming, but we simply can’t bet on them being available just yet.

If this drive has a downside, it’s the pricing. Right now, the Black SN8100 is the better deal. For that matter, so is the 9100 Pro. However, if we assume the price goes down to where it should be, we can feel it would be the best drive available. We also should point out that the heatsink premium on this drive, right now, is too large to consider, with a custom solution being the way to go. This might also change in the future, and as the heatsink matches the drive with programmable RGB LED, it should be a good option if the premium is cut. We have no other complaints with the drive and could even, in some cases, recommend it for PCIe 4.0 slots, as, frankly, it has the performance to sustain PCIe 4.0 speeds and would be even more power-efficient. It’s just that the experience will cost you, whether you go for this or the Black SN8100.

MORE: Best SSDs

MORE: Best External SSDs

MORE: Best SSD for the Steam Deck

The latest liquid cooler from MSI to land on our test bench is the new MPG Coreliquid P22 360. While it packs a budget price of only $129.99, it doesn’t cut corners, featuring good thermal performance, low noise levels in common scenarios, and a fancy 2.1-inch IPS display.

Let's take a look at the specifications and features of this eye-catching AIO, then we’ll go over thermal and noise benchmarks, and decide if MSI’s MPG Coreliquid P22 360 deserves to make our list of the best CPU coolers.

Cooler specifications

Features of MSI’s MPG Coreliquid P22 360

▶️ Frame compatible with both AMD and Intel systems

Usually when you purchase an AIO, there’s at least one frame included, designed for a specific AMD or Intel socket.

MSI has simplified the install process here, with a single-frame design that’s compatible with both AMD AM4/AM5 and Intel LGA 1700/1851.

▶️ 2.1-inch 540x540 IPS display

A 2.1-inch IPS display with a resolution of 540x540 is included with this AIO, allowing users to observe performance metrics by glancing at the screen. Four preset backgrounds are included with the AIO’s software, and custom images and videos up to 500MB are supported.

To customize the AIO, you’ll need to download MSI’s EZ Display software, which is simple to use but a little rough around the edges. You can set it to use one background by enabling the “repeat” button, or you can have it alternate between the images and videos you have loaded in the software.

My biggest complaint here is that you can’t disable or remove the preset images, which means they’ll be included in the rotation of images displayed unless you’ve set the AIO to show only one image or video by selecting the “repeat” button.

▶️ RAM Clearance

As with most liquid coolers, the design of thisAIO is such that the CPU block doesn’t overhang or interfere with the DIMM slots in any manner, ensuring that all sizes of RAM, no matter how tall, are compatible.

▶️ 120mm fans

There’s more to a cooler than just the radiator and pump. The included fans have a direct impact on aesthetics, noise levels, and overall thermal performance. Conveniently, the trio of fans here arrive pre-installed, saving a few moments during installation. They also feature a daisy-chaining quick-connect system, designed to eliminate cable clutter.

▶️ Packaging

The outer packaging is a bit flashy, at least, in comparison to your normal AIO box. It features a rendering of the cooler against a background of black, blue, purple, and pink.

The inner packaging is typical of most liquid coolers, with molded foam, cardboard, and plastic coverings protecting the contents during shipping.

Included with the package are:

• Mounting hardware for AMD and Intel platforms
• Aesthetic cover for mounting hardware
• A small tube of thermal paste
• 360mm radiator and pre-installed 120mm fans
• 2.1-inch display

▶️ AM5 Installation

This section assumes you’ve already mounted the 360mm radiator. Installation of AIOs is much easier when you have already secured the radiator to your case.

To begin putting things together, you’ll first need to remove the default AM4/5 retention.

The next step is to screw in the mounting studs, followed by an application of thermal paste to the CPU. If you’re not sure how to do that, we have a handy thermal paste application guide that covers the different methods you can use.

After thermal paste is applied, take the CPU block and press it against the studs, using the included screws and a screwdriver to secure it in place.

There are two final steps. First, place the aesthetic cover on the CPU block. Then attach the USB, PWM, and ARGB connections to their corresponding motherboard headers.

Real-world testing configuration – AMD AM5 platform

We’ve tested coolers with both the Ryzen 9950X3D and its non-V-Cache sibling, the 9950X. There are some differences in how the 9950X and 9950X3D CPUs are impacted by thermal events. While the heat output of the CCDs of AMD’s 9950X3D is relatively balanced, the 9950X I used has one CCD that runs much hotter than the other, with a difference of over 10 degrees Celsius in some scenarios, shown below.

We’ve since returned to using a 9950X3D for cooler testing, as it has a more balanced heat profile, and is almost certainly a more widely adopted CPU. The benchmark results shared in these reviews may differ from others because I emphasize results that are comparable to real-world use. This means I generally test CPU coolers inside of a closed desktop case, which increases cooling difficulty compared to other testing methods.

Many reviewers test on open test benches, which have a combination of lesser airflow needs and lowered ambient temperatures. This results in making weak coolers appear stronger than they really are. Some have also used generic thermal plates to test cooling solutions. I reject both of these methods because they don’t accurately reflect real-world cooler conditions.

Our latest testing setup uses the FLOVA F50 computer case from Tryx.

This case features a unique “crossflow” fan that pulls air from the side, which the company claims is more effective than traditional intake fans. For air cooling tests, we’ve added a single Noctua NF-A12 G2 intake fan.

PBO Performance and maximum noise levels

We’re going to start this review’s benchmark section by focusing on a traditional maximum performance test, with the CPU cooler’s fans allowed to reach their fastest speeds, for the best cooling possible.Turning on PBO allows AMD’s Ryzen 9 9950X3D to stretch its legs to an extent and consume over 260W. Enabling PBO incrteases max power consumption and heat output. We use MSI’s X870E Carbon Wifi motherboard, and often the CPU will reach its TJ Max (peak temperature) of 95 degrees C (203 F) and thermally throttle to some extent with most coolers. When this throttling occurs, I’ve measured the average power consumption to determine performance.

High-end AIO liquid coolers, however, are expected to be able to keep AMD’s Ryzen 9 9950X3D under TJ Max during the duration of the benchmark – in these cases, we compare the CPU’s actual temperature.

Today’s review is a bit unique, in that we have the results from a prototype and final version of the AIO being reviewed today – giving us some insight into how these products evolve during the design and manufacturing process.

Both the prototype and retail versions of MSI’s MPG Coreliquid P22 360 handle this test with ease, with an average temperature of 84.7C (184.5F) – one of the best results we’ve seen on this test bench.

Some coolers perform well in maximum-strength tests, but run loudly at full speeds. Noisy fans in this scenario aren’t necessarily a bad thing; some folks prefer to hear them as a way to be aware when the CPU is getting hot.

One issue I raised when I tested the prototype of this AIO was that it ran louder than it needed to. 48.9 dBA is pointlessly loud for most users. MSI took that feedback to heart, and the fans included with MSI MPG Coreliquid P22 360 reach 44.9 dBA at 100% RPM, a moderate volume level that is much quieter than the maximum noise levels of competitor liquid coolers. Those who prefer to implement speed restrictions for silent operation need not worry; our noise-normalized benchmarks show good performance even when the fans are set to run quietly.

200W thermal benchmarks

For the next thermal test, I leave the motherboard settings at their defaults, which results in a power limit of 200W when running Cinebench R23.

In this test, with a standard power limit enabled, MSI’s Coreliquid P22 performs in the middle of the results we have from other liquid coolers – 1.2 degrees C behind our best result, and 3.6 C cooler than the worst result from another AIO.

150W + GPU thermal results, noise levels

Our next test runs Cinebench on the CPU with a 150W power limit, while also running Furmark on MSI’s RTX 4070 Ti Super Ventus 3x OC. This causes the GPU to consume ~295W of power. This test is designed to emulate the thermals of games, which stress the CPU and GPU.

In this test, designed to emulate the thermal conditions of a CPU-intensive AAA game, MSI’s P22 continues to deliver outstanding performance – the result of 37.1 C over ambient (59.1 C actual temperature) is our third-best result, and an improvement of one degree compared to the prototype of this AIO.

Noise-normalized testing

Most testing is performed with the cooler tied to the default fan curve of our MSI X870E Carbon motherboard, but some of y’all prefer to see tests when the noise levels of coolers are equalized. This is especially important to those of you who prefer silent computers. This next test has the CPU cooler noise-normalized to 38.9 dBA, with PBO enabled for the Ryzen 9 9950X3D CPU. Most coolers will not be able to keep the CPU from reaching its peak temperature (TJ Max) in this stress test. So we’ll look at this in terms of average CPU power consumption, which averaged 255.8 watts.

This test is especially difficult, because in addition to the reduced noise from the CPU cooler, our current test bench’s system fans are configured to run extremely silently, below the floor of the noise meter I use to measure dBA.

While I prefer to look at the power consumption in a scenario where the CPU reaches TJ Max (the processor’s peak temperature), I’ve also included the average CPU temperature in the chart below.

Karhu DDR5 RAM thermals testing

Your CPU cooler does not operate in isolation. It has an impact on not just your CPU’s temperatures, but also the other components in your build, like your RAM and GPU. To that end, I’ve run the Karhu RAM stress test. This places a load of ~153W on the CPU and ensures system RAM (DDR5 in my case) is fully stable. In this type of scenario, most AIOs tend to produce worse results than air coolers.

DDR5 temperatures were excellent in this test, on par with ASRock’s Steel Legend, which features an integrated VRM fan. The only result showing better performance is from Silverstone’s IceMyst Pro, with fans that directly cool system DIMMs.

Conclusion

MSI is back in the AIO game with the MPG Coreliquid P22 360. This AIO has a relatively low price tag while featuring a 2.1-inch IPS display, excellent thermal performance, and low noise levels in common scenarios. It’s a solid choice if you’re after a cooler with a round display that cools well. Just note that if you can live without a display, there are excellent options that cost quite a bit less, like Arctic’s Liquid Freezer III Pro, which sometimes sells for less than $85.

AGI is probably not a brand you’ve heard of, but they make many memory products that are often sold at a discount. We reviewed the AI818 SSD two years ago with something less than enthusiasm, but we’re happy to say that AGI seems to have taken our critique to heart with the AI858. A budget, high-end drive is a bit of a contradiction, but in this atmosphere of skyrocketing memory prices, every dollar matters. If you can save a few bucks and still get a fast drive, you should, and that’s the segment the AI858 is targeting, but this time with hardware that’s actually good.

AGI AI858 Specifications

If you’re ever upset that a drive only comes in one or two capacities, then the TeamGroup G70 Pro might be for you. Not only does it come in both heatsinked and non-heatsinked versions, but it’s also available at 512GB, 1TB, 2TB, 4TB, and even 8TB. At the time of review, we could only find 1TB, 2TB, and 4TB, with pricing pretty close between the two types – go for the heatsink, if you can. We’re giving the lower prices at $197.99, $326.99, and $505.99. If you’re shooting for DRAM, these prices aren’t too bad, but we’d lean towards the Seagate FireCuda 530R instead at 1TB. The G70 Pro is more competitive at 2TB and 4TB against comparable drives.

The drive is capable of reaching up to 7,400 / 6,600 MB/s for sequential reads. AGI does not list the random read and write IOPS specifications, but we know this controller and flash combination tends to reach at least 700K and can reach 1,000K or more. This is comparable to other drives in this class. We wouldn’t recommend the drive at 512GB as it can’t reach peak performance. Ideally, you would go for 2TB or 4TB for the best results. The drive is backed by a five-year, 740TB of data writes per TB of capacity warranty, which is above average but not exceptional.

AGI AI858 Software and Accessories

AGI offers no downloads aside from the spec sheet. For checking drive health information, we recommend CrystalDiskInfo. CrystalDiskMark, its sibling, is good for benchmarking your drive to make sure everything is working well. For backups and disk imaging/cloning, we recommend MultiDrive for Windows and either Clonezilla or Rescuezilla for other operating systems or for booting.

AGI AI858: A Closer Look

This is a double-sided drive with a heatsink. Great for desktops and, in a pinch, the PS5, but less great for anything else. Considering it’s a high-end PCIe 5.0 drive, that’s fine – you’ll be using this in an enthusiast desktop, most likely.

There’s not too much else interesting here unless you read the PCB silkscreen lettering beneath the rear label. This gives you information after the drive and its components – it’s using the Silicon Motion SM2508 (SM2508) controller in the M.2 (M2) form factor with LPDDR4 (LPD4) volatile memory, four NAND flash packages (4) in the BGA178 (178) form factor with an SMI reference design (Y260A), manufactured during week 23 of 2025 (25.23). This information could be useful to you in multiple scenarios: when you’re looking at review photos of drives, when you are checking a drive you purchased, or when you’re comparing drives.

The drive has the SMI SM2508 controller, as mentioned, with a single Samsung LPDDR4 DRAM memory module and four mysteriously-labeled NAND flash modules. The Samsung memory is low-power, which is good, and is in a 16Gb or 2GB configuration, which means the drive meets the usual 1GB:1TB DRAM:NAND ratio.

Deciphering the flash is a little trickier, but we’ve discovered that this uses Micron’s 232-Layer TLC. This matches other “budget” SM2508-based drives that are using this flash rather than BiCS8 or 276-Layer Micron. It’s not a bad trade-off if you can get the drive for a little less money.

MORE: Best SSDs

MORE: Best External SSDs

MORE: Best SSD for the Steam Deck

Comparison Products

The AGI AI858 gets no respite as it’s up against some of the fastest drives on the market. These include the proprietary WD Black SN8100 and Samsung 9100 Pro. We then have drives with the SM2508 controller, including the Kingston Fury Renegade G5, using BiCS8, and the Lexar NM1090 Pro and Biwin Black Opal X570 using Micron flash. The Crucial T705 and Corsair MP700 Pro SE are using that same Micron flash but with the Phison E26 controller, instead. Lastly, we have the lower-end Addlink G55, which uses BiCS8.

Trace Testing — 3DMark Storage Benchmark

Built for gamers, 3DMark’s Storage Benchmark focuses on real-world gaming performance. Each round in this benchmark stresses storage based on gaming activities including loading games, saving progress, installing game files, and recording gameplay video streams. Future gaming benchmarks will be DirectStorage-inclusive and an evaluation for future-proofing is included where applicable.

The AI858 falls into the middle of the pack in 3DMark, which, all things considered, isn’t that bad. The drives with BiCS8 flash – the Black SN8100 and Fury Renegade G5 – are the fastest drives due to the flash’s famously low latency. The DRAM-less G55 and X570 are at the opposite end and are the slowest. Everything else is within the same zone, and the AI858 actually beats its closest rival, the NM1090 Pro, by a small amount. We could chalk this up to the capacity difference, after all, 4TB could be slower, but our experience with similar drives like the Acer Predator GM9000 suggests otherwise. The AI858 is legitimately slightly faster here.

Games are starting to lean on SSD performance more, but a small loading advantage was more compelling back when drives were affordable. That said, if the AI858 or equivalent is within a reasonable distance from other drives, it could be worth jumping up to the faster platform. The gaming advantage would just be icing, and this gap could widen in the future. Normally, that’s not a big deal for a current-day purchase, but if we’re years out before a consumer SSD industry recovery, then buying speed early is a fine investment.

Trace Testing — PCMark 10 Storage Benchmark

PCMark 10 is an industry-standard trace-based benchmark that uses a wide-ranging set of real-world traces from popular applications and everyday tasks to measure the performance of storage devices. The results are particularly useful when analyzing drives for their use as primary/boot storage devices and in work environments.

We see a similar difference in PCMark 10. The AI858 again beats the NM1090 Pro and, as before, this isn’t due to the capacity difference, given our overall review experience. The drive ends up in the middle again, but this isn’t a bad place to be when facing off against some of the fastest drives on the planet. You’ll get better responsiveness out of the Black SN8100, but the AI858’s level of performance is still exceptional.

As with gaming, it’s hard to justify price leaps when app and app loading performance differences might be relatively small in the real world. However, the extra bandwidth can be more impactful for enthusiasts. We do think it’s worth going with an AI858-class drive over a DRAM-less PCIe 4.0 if you’re building a powerful system with a purpose; it’s just a matter of pricing that advantage against the other hardware components. It’s safe to say that personal computing as a whole has gotten much more expensive, and SSD price scales almost directly with capacity. This means nailing down the right performance range – which could be as simple as, which 1TB drive gives me the best bang for the buck if I’m a developer? – becomes more crucial.

Console Testing — PlayStation 5 Transfers

The PlayStation 5 is capable of taking one additional PCIe 4.0 or faster SSD for extra game storage. While any 4.0 drive will technically work, Sony recommends drives that can deliver at least 5,500 MB/s of sequential read bandwidth for optimal performance. Based on our extensive testing, PCIe 5.0 SSDs don’t bring much to the table and generally shouldn’t be used in the PS5, especially as they may require additional cooling. Check our Best PS5 SSDs article for more information.

Our testing utilizes the PS5’s internal storage test and manual read/write tests with over 192GB of data, both from and to the internal storage. Throttling is prevented where possible to see how each drive operates under ideal conditions. While game load times should not deviate much from drive to drive, our results can indicate which drives may be more responsive in long-term use.

AGI takes care to mention PS5 performance – 6 GB/s sequential reads – for this drive as it is a viable use case. We see no performance issues whatsoever, but putting a high-end PCIe 5.0 SSD into your PS5 is a somewhat questionable decision. You can almost always get away with a less expensive drive that will perform equivalently.

We actually think it’s more valuable to find a reliable drive for this. Having a heatsink, as the AI858 does, is actually an investment towards this goal, as you want to stop your PS5 drive from throttling. Two things to address here. First, yes, NAND flash likes heat, but if your drive is redlining in the PS5 from the ambient under load, then that is very much a problem. Two, the AI858 is a high-end drive, but it’s built with a newer, more efficient controller and will be running at half speed in a PS5. Therefore, it’s likely to do very well even without its heatsink, which makes it an interesting pick if you can find it priced reasonably.

Transfer Rates — DiskBench

We use the DiskBench storage benchmarking tool to test file transfer performance with a custom 50GB dataset. We write 31,227 files of various types, such as pictures, PDFs, and videos, to the test drive, then make a copy of that data to a new folder, and follow up with a reading test of a newly-written 6.5GB zip file. This is a real-world type workload that fits into the cache of most drives.

We’re admittedly a little disappointed in the AI858’s DiskBench scores – 2,231 MB/s for the copy transfer rate is particularly low – but we feel this is a good opportunity for analysis. The drive’s read transfer rate is perfectly normal and even good by some measures, beating drives like the Fury Renegade G5 and 9100 Pro. Then we have the write transfer speed, which is below 2 GB/s and 10% or more beneath its peers. Clearly, the copy transfer rate is hindered by this more than anything else.

This does mean that the drive may take writes more slowly than the competition and will write more slowly to itself. This can lead to a slowdown in some cases, and especially if you have multiple fast drives with lots of transfers, this performance deficit can add up. However, in daily use, it doesn’t mean much, and on the flip side, if you are doing extreme writes, then it could even out over the long term. This is one reason we test write saturation – see our results below.

However, looking specifically at this result, it’s within the margin of error versus the identically-equipped NM1090 Pro and even the Black Opal X570. This is logical as the other drives on the list, except for the T705 and MP700 Pro SE, are using newer flash. In fact, the T705 is using the same flash as the AI858. Our test fits within the pSLC cache, and we would not expect the Phison E26 controller to outdo the SM2508, so we believe this is due to a difference in flash optimization. We know there were some issues getting Micron’s 232-Layer flash up to speed, especially at capacity, and Crucial having a better handle on it is not surprising.

We’re hesitant to call it a flash quality difference; maybe more of a trade-off, as being conservative with write performance, with read workloads being prevalent in consumer usage, can be a smart move. The T705 was trying to stand apart at its launch so we can understand why it pushed the limits at the time. This also partly explains why the AI858’s results are less impressive today. Alternatively, or in combination, it’s also possible AGI is doing this to reduce user flash wear in order to make the TBW in the case they have to swap hardware, or if they have to use lower PEC flash. This is not an issue for the general user, but could be worth consideration if you are buying a drive to do a lot of writes.

Synthetic Testing — ATTO / CrystalDiskMark

ATTO and CrystalDiskMark (CDM) are free and easy-to-use storage benchmarking tools that SSD vendors commonly use to assign performance specifications to their products. Both of these tools give us insight into how each device handles different file sizes and at different queue depths for both sequential and random workloads.

ATTO gives us a good idea of how a drive will operate with transfers of files of various sizes. A queue depth of 1, or QD1, is not unreasonable for this. While usually you look at random read performance to derive a real-world feel, for many games and apps, sequential reads are actually more common. Likewise, 4KiB tends to be what people focus on, while many games and apps operate with bigger I/O. We get to see the full picture with ATTO, including writes, which are more impactful if you intend to use the drive for caching.

The most noticeable thing we see with the AI858 is the drop at 2MiB for sequential reads. This is mirrored with many other drives, including the T705, the Lexar NM1090 Pro, and the MP700 Pro SE, drives that have the same flash. Why the drop here? As two different controllers are used here, we can assume the flash is the culprit. Micron’s 232-Layer TLC uses six planes and the typical 16KiB page size. Multiplying these by the controller’s eight flash channels and, with parallelization, by two or four dies per channel, you get 1.5MiB and 3MiB-sized superpages, respectively. Without outstanding I/O at QD1, this could be the cause, but this remains speculative. The two controllers – The E26 and SM2508 – handle it differently, with the E26 less impacted, so this might be more of a firmware-flash combination.

Chances are you’re not buying a drive for such specific performance, but it’s worth noting if you’re technically-minded. We can also see ATTO map to CDM with the latter’s sequential results, specifically QD1 reads, which tell us a lot. The AI858 and NM1090 Pro, for that matter, do poorly here against the other drives. This is down to the controller. Lots of reads happen at QD1, and 1MB is a realistic I/O size. On the other hand, if you’re doing writes, you’re fine and dandy. If you’re doing reads and writes or reads at a higher queue depth, the AI858 is also right up there. Random write latency is also excellent, so this could make an excellent “budget” caching drive if you need something PCIe 5.0.

Let’s be honest, though, you’re here to look at random 4KB QD1 latency. We’re pleased to say that the AI858 does exceptionally well, beating all drives that don’t have BiCS8 flash. It’s a stellar choice on a budget for this reason, or at least as much “budget” as you can get on a high-end PCIe 5.0 drive. If that’s your primary metric, then you can leave the review knowing AGI hit the mark on this one.

Sustained Write Performance and Cache Recovery

Official write specifications are only part of the performance picture. Most SSDs implement a write cache, which is a fast area of pseudo-SLC (single-bit) programmed flash that absorbs incoming data. Sustained write speeds can suffer tremendously once the workload spills outside of the cache and into the "native" TLC (three-bit) or QLC (four-bit) flash. Performance can suffer even more if the drive is forced to fold, the process of migrating data out of the cache in order to free up space for further incoming data.

We use Iometer to hammer the SSD with sequential writes for 15 minutes to measure both the size of the write cache and performance after the cache is saturated. We also monitor cache recovery via multiple idle rounds. This process shows the performance of the drive in various states, including the steady state write performance.

The drive first writes in the fastest pSLC cache mode at almost 12.6 GB/s for over 32 seconds. The cache is over 400GB in size, roughly 406GB by our numbers, which is large in comparison to some drives like the T705 but is not as large as the cache on the Fury Renegade G5 or NM1090 Pro. A 2TB TLC flash drive can have a cache up to almost 700GB – the Fury Renegade G5’s is almost 684GB in our testing – but this leaves little spare space to handle longer writes. The AI858 gracefully declines to ~3.7 GB/s in a presumably direct-to-TLC mode, which is close to what we see with the T705 and Black Opal X570, drives with the same flash. This is a solid result.

The drive eventually runs out of free space and must wait for already-written data in the background to “fold” from the cache into native flash. This reduces the drive's write speed and increases latency. In this case, the drive manages around 1.25 GB/s, which isn’t super fast but isn’t the end of the world for that weakest performance state. In fact, the drive averages 3.677 GB/s for steady state, which is quite good. We typically expect this level of performance for this flash, although it can be faster in drives like the T705, which again does have a smaller pSLC cache.

Given the NM1090 Pro’s results, we suspect AGI targeted write performance at a certain level – matching what we saw earlier in DiskBench – and the cache was adjusted to reflect this. We think this was a good move, even if it doesn’t always benchmark as well. You’re bound to have a more consistent experience as a result, and the drive is capable of heavier workloads, which makes the drive usable in a caching scenario. However, we would caution against greatly exceeding the TBW within the warranty period on a drive like this. This means knowing your workload’s average write amplification factor, as host write counting won’t cut it – some of the Micron 232-Layer TLC floating out there may be rated significantly below the normal 3,000 PEC.

Power Consumption and Temperature

We use the Quarch HD Programmable Power Module to gain a deeper understanding of power characteristics. Idle power consumption is an important aspect to consider, especially if you're looking for a laptop upgrade as even the best ultrabooks can have mediocre stock storage in terms of capacity and performance. Desktops are often more performance-oriented with less support for power-saving features so we show the worst-case for idle.

Some SSDs can consume watts of power at idle while better-suited ones sip just milliwatts. Average workload power consumption and max consumption are two other aspects of power consumption but performance-per-watt, or efficiency, is more important. A drive might consume more power during any given workload but accomplishing a task faster allows the drive to drop into an idle state more quickly, ultimately saving energy.

For temperature recording we currently poll the drive’s primary composite sensor during testing with a ~22°C ambient. Our testing is rigorous enough to heat the drive to a realistic ceiling temperature but real-world temperatures will vary due to the environment and workload factors.

Is the AI858 efficient? For a high-end PCIe 5.0 drive, yes! It puts the original E26 generation of drives – including the T705 and MP700 Pro SE – to shame. On the other hand, it’s no match for the high-end BiCS8-equipped drives or the more modest G55. The real question is: does this matter? We think that, with a heatsink, this drive will stay cool and should end up in a system where its power draw is fine. It will not require active cooling.

Our temperature testing reinforces this assumption with a maximum rating of 66°C on the primary sensor. This is 17°C below the first throttling point. We could see this drive working without a heatsink in some machines if they are cooled well enough, and certainly in laptops that are using a PCIe 4.0 slot, where it will pull less power. We see no issues here worth mentioning.

Test Bench and Testing Notes

We use an Alder Lake platform with most background applications, such as indexing, Windows updates, and anti-virus, disabled in the OS to reduce run-to-run variability. Each SSD is prefilled to 50% capacity and tested as a secondary device. Unless noted, we use active cooling for all SSDs.

AGI AI858 Bottom Line

AGI’s AI858 does everything it needs to do and nothing it doesn’t. It has a high bandwidth ceiling with good to excellent random read latency. Sustained write performance is also good, and the drive doesn’t overheat or pull too much power. It even comes with a heatsink to make life easier. While it’s no record-setter, it performs well enough on the whole to firmly be a high-end PCIe 5.0 drive. It’s positioned to be less expensive than the more well-known alternatives like the Black SN8100 while providing most of the performance. We have no trouble recommending it if you’re looking for a drive in this class, which is certainly an improvement over the last AGI drive we reviewed.

The downsides are exactly what you would expect here. As we stated, its performance is good but not the best. Its power efficiency is good but not the best. Its sustained write performance and even its excellent random read latency…you guessed it, not the best. It doesn’t have a flashy 8TB SKU to throw at you, either. That’s fine – the AI858 is not trying to be that drive. It’s a budget alternative to high-end options with relatively small trade-offs. If you have a PCIe 5.0 system and don’t want to let that bandwidth go to waste, it’s more than sufficient. Enthusiasts at that level usually know how to manage a drive so the lack of software support isn’t a big issue. There are several drives like the AI858 so, at the end of the day and as always, it comes down to price. Scout accordingly.

MORE: Best SSDs

MORE: Best External SSDs

MORE: Best SSD for the Steam Deck

The entire point of highly customizable creator-oriented macropad-like streaming decks is that they're, well... highly customizable. While most of them still center around a series of translucent plastic LCD keys, we've seen both Elgato and Loupedeck add a variety of other input options — dials, knobs, touch buttons, touchscreens, and more — but what if you could make your own custom streaming controller?

That's the premise of Cooler Master's MasterHUB, a fully-customizable modular macropad/controller for any and all creators — or at least, that's what it was supposed to be.

The MasterHUB was actually announced almost two years ago: Cooler Master initially put it on Kickstarter, where it was fully-funded within an hour, raising a total of $267,491 by the end of its run. It had a somewhat rocky start and began shipping to backers a year ago, and has since lagged on the promised software updates and functionality. You can now find the MasterHUB Creator Kit on Amazon for $170, which is the same price as the Elgato Stream Deck+, so let's take a look at what you're getting.

Design of the MasterHUB

The MasterHUB comes in a few different packages — there's the MasterHUB Creator Kit, which includes the base module, a 15-key IPS module, a 5-fader module, and a 2-roller module. There are also two other modules, which are currently sold separately: a module with three knobs, and a module with an encoder dial, which is a clickable dial with a customizable screen.

The MasterHUB is bigger than most Stream Decks — it measures 7.4 x 4.9 inches (190 x 126.5mm), which makes it slightly larger than the 32-key Stream Deck XL (7.2 x 4.4 inches / 182 x 112mm). The base station is 0.6 inches (16.2mm) thick, but with the modules attached it's about 1 inch (25.4mm) thick, not including the additional height of buttons or knobs. So it's not the most space-friendly customizable macropad, since the base station is always a requirement no matter how you set it up.