On this day in 1996, a freshly inked U.S. patent quietly laid the cornerstone of the digital music revolution. In addition to facilitating this now vast internet-based entertainment business, the humble MP3 file format would propel broadband proliferation, usher in the iPod era, and arguably precipitate the iPhone and all the other touchscreen-slabs that remain indispensable gadgets to this day.

MPEG Audio Layer III (MP3) files were devised by scientists to greatly reduce the amount of data required to represent an audio file. Key personnel behind the invention of MP3 included: Bernhard Grill, Karlheinz Brandenburg, Thomas Sporer, Bernd Kurten, Ernst Eberlein, and Dieter Seitzer. Brandenburg is often credited as being the father of MP3, for leading this and similar research since 1977, but Seitzer (for example) brought expertise in transferring music over standard phone lines.

Central to this digital audio format’s success was its ability to shrink audio files by 75%-95%. It did this by using now familiar lossy media compression techniques. In other words, the MP3 audio codec discarded audio data to achieve significant file size reductions, with little perceptible difference in listening quality due to the perceptual limitations of human hearing. Depending on the target listening device, users could change compression bit rates, though audiophiles often insisted they could hear the difference between raw CD audio source data, or indeed, vinyl.

Work on the MP3 file format began in 1987, and a patent was awarded for the compression method in Germany in 1989. But it wouldn’t be until 1996 when U.S. Patent 5,579,430 was granted to Fraunhofer, securing the format in the home of leading global entertainment businesses, that MP3 started to gain traction. The inventors were primed to make lots of money from the internet boom that had started in earnest.

From Napster shockwaves to an industry accepted ‘1,000 songs in your pocket’

Unfortunately for Fraunhofer, its software was quickly ripped off and shared widely online. Once it was re-packaged into CD ripper and encoder apps, it was a cinch for music collectors to rip CDs into MP3s and share them with friends. Such activity began online via FTP servers, but in 1999, Napster was launched, changing the music industry forever.

Several tech moments converged to make this a perfect storm for the traditional music industry. WinAmp was released and ready to be your default MP3 player in 1998, the year before Napster arrived. At the same time, broadband was proliferating, which meant it could take seconds, instead of several minutes, to grab the latest number one hit single. Broadband became an accelerant for the physical music media inferno.

We also saw the first MP3 hardware player arrive in 1998. Diamond Multimedia’s Rio 100 launched in the U.S., and Saehan Information Systems' MPMAN became available in Korea. Both used solid-state Flash to store music.

Apple would be instrumental in the legitimization of digital music. Its iTunes software launched in January 2001, with the first iPod hardware released within the same year. However, it would take until 2003 to put together the first groundbreaking partnerships with the music business in crisis. That is when the iTunes Music Store opened its doors with 200,000 digital music tracks at $0.99 each.

The MP3 file format patent expired in 2017, and is an old legacy codec, now largely a symbol of an iconic era in media and internet history. For the same purpose – digitizing physical music media – it has been usurped by modern alternatives like AAC and FLAC.

In 2025, music ripping, downloading, and sharing have taken a back seat to streaming services powered by superfast broadband and 4G/5G mobile connectivity. Ardent music enthusiasts and collectors still prefer their physical media, though, as was the case during the MP3 era.

Raspberry Pi has long been the most famous in the world of single-board computers. But there are plenty of alternatives, such as LattePanda and Khadas, and of course, BeagleBoard, which has announced the availability of its RISC-V powered BeagleV-Ahead.

Coming in at $149, the BeagleV-Ahead is powered by a 2 GHz, quad-core RISC-V 64GCV Xuantile C910 CPU and has 4GB of RAM and 16GB of eMMC flash storage on which you can install either Ubuntu or Yocto Linux. The OS is flashed to the board, and we then remotely interface with the board via a terminal or web interface.

The big draw for this board is the RISC-V CPU. In the maker community, RISC-V is gaining popularity, mainly due to its open-source nature and availability in several different price points and packages. This includes Milk-V, which offers three RISC-V machines starting from $9. You can even get RISC-V-powered soldering irons now!

Beagle boards have used Arm-based CPUs in the past, but the switch to an open-source RISC-V CPU is the only significant change between the boards. The distinct BeagleBoard form factor remains, meaning "capes", BeagleBoard's name for add-on boards (think Raspberry Pi HATs or Arduino Shields). Capes connect to the board using the P8 and P9 header, essentially the same as the Raspberry Pi's GPIO. But with the BeagleV-Ahead, we get six analog inputs and a plethora of PWM, I2C, UART, SPI, I2S and good old digital IO.

The BeagleBoard team has a documentation page up; from there, we can delve deeper into the BeagleV-Ahead. There are a few gaps, but this board is just a few days old, so documentation will eventually catch up.

We first reported on this board in 2021, and in the two years since, there has been a pandemic and a global chip shortage which has impacted the delivery time of the BeagleV. That said, it is good to see another RISC-V board being offered to an enthusiastic maker community.

The BeagleV-Ahead is available from several distributors, with an average price of $149. 

In a recent video, YouTuber EposVox reports that YouTube is finally rolling out AV1 live-streaming support to the platform, with the tech currently in a beta. AV1 will provide YouTube live streams with a substantial increase in video quality, and allow users to stream at up to 4K 60FPS with Twitch-limited bitrates. EposVox was able to get early access to a development build of OBS 29.1 to check out YouTube's live streaming AV1 capabilities.

The newest addition to the AV1 rollout is YouTube live streaming support with AV1. YouTube just rolled out beta support for a new video live-streaming standard known as Enhanced RTMP, which will allow streamers to utilize several of the latest video codecs, including AV1, VP9, and HEVC (H.265) to live stream videos to YouTube.

EposVox was able to test drive Enhanced RTMP, with a development build of OBS 21.9 to stream AV1 gaming content to YouTube directly. According to EposVox, the quality difference is night and day compared to H.264. The quality jump with AV1, allowed him to drive higher quality video to his live stream, and remove pixelation altogether.

Just for perspective on how powerful AV1 is, EposVox was able to run an AV1 1440P 60FPS live stream of Halo Infinite at 500kbps - a bitrate 15x lower than the Twitch limit, and the stream was still perfectly watchable.

For normal use cases, EposVox found that 8mbps was the sweet spot for 1440P 60FPS, and around 15mbps for 1440P 60FPS. For a perfectly good-looking live stream with none or close to no pixelization. For users that still want to stream 1080P video, all you'll need is a 4MBps bitrate to achieve the same result. This is a night and day difference to H.264 where 8Mbps was about the minimum you want for a high-quality 1080P 60FPS video stream, and even in this situation, pixelation is still very likely to occur with a lot of streams.

AV1 represents the biggest jump in video quality and performance in a decade. AV1 is the latest-generation open-source codec, that has garnered massive support from many industry tech giants to supersede older standards such as H.264. The media codec is substantially more capable than H.264, and has the capability of running up to 4K 60FPS streams at similar bitrates to H.264 1080P video.

Development has been slow, with AV1 originally being founded in 2015. But after 5 years of development, AV1 is now in a functional state, with all major GPU manufacturers and several major video platforms already supporting AV1 video playback. YouTube is one of these platforms and already supports AV1 video playback and AV1 uploads. (In fact, Discord already supports limited AV1 live streaming.)

The only side-effect of YouTube's implementation is that videos will still get transcoded to VP9. This means AV1 live streams will be re-coded to YouTube's VP9 codec. VP9 is known to be of slightly lesser quality but it supports far more GPU hardware. Thankfully this isn't a big deal, since transcoding doesn't take away much from AV1's encoding capabilities, and video quality remains exceptionally good.

We don't know exactly when AV1 live streaming will roll out officially across the entire YouTube platform, but it will probably be soon since the roll-out is already in beta. OBS isn't far behind either, with a new version that is being developed right now that will support YouTube's new AV1 live stream standard (Enhanced RTMP).

If you want to get ready for AV1 live streaming support, all you need is either a capable multi-core CPU (for software-based AV1 encoding) or purchase a new RTX 40 series, RX 7000 series, or Intel A-series Arc Alchemist graphics card to live stream AV1 to YouTube in the future.

Nvidia has quietly removed some of the concurrent video encoding limitations from its consumer graphics processing units, so they can now encode up to five simultaneous streams. The move may simplify the life of video enthusiasts, but Nvidia's data center grade and professional GPUs will continue to have an edge over consumer products as now Nvidia does not restrict the number of concurrent sessions on them. Obviously, the speed of encoding can suffer with more simultaneous encodes.

Nvidia has increased the number of concurrent NVENC encodes on consumer GPUs from three to five, according to the company's own Video Encode and Decode GPU Support Matrix. This is effective for dozens of products based on the Maxwell 2nd Gen, Pascal, Turing, Ampere, and Ada Lovelace microarchitectures (except for some MX-series products) and released in the last eight years or so.

The number of concurrent NVDEC decodes isn't restricted, though if you're after real-time decoding, there are limits to how many streams can be handled at once. Nvidia provides a table showing the relative decode speeds of various GPUs by codec, though it doesn't indicate the resolution of those streams. Still, even second generation Maxwell GPUs could do over 400 fps of decode, enough for at least six simultaneous 1080p (?) streams — assuming you don't run out of VRAM.

The change does not affect the number of NVENC and NVDEC hardware units activated in Nvidia's consumer GPUs. For example, Nvidia's latest AD102 graphics processor, based on the Ada Lovelace architecture, features three hardware NVENC encoders and three NVDEC decoders. All three are enabled on Nvidia's RTX 6000 Ada and L40 boards for workstations and data centers, but only two are active on consumer-grade GeForce RTX 4090. The AD103 and AD104 each have two NVENC and four NVDEC units, but GeForce cards only enable two of the NVDEC units.

Historically, Nvidia has restricted the number of concurrent encoding sessions on all of its graphics cards. Consumer-oriented GeForce boards supported up to three simultaneous NVENC video encoding sessions while Nvidia's workstation and data center solutions running the same silicon and aimed at ProViz, video streaming services, games streaming services, and virtual desktop infrastructure (VDI) could support 11–17 concurrent NVENC sessions depending on the quality and hardware. It turned out a couple of years ago that Nvidia's restrictions could be removed by a relatively simple hack.

But Nvidia's stance on NVENC and NVDEC limitations has evidently changed a bit. Now, consumer GPU support up to five concurrent NVENC encode sessions. Workstation-grade and data center-grade boards do not have any restrictions and the actual number of concurrent sessions depends on actual hardware capabilities, codec choice, and video quality.

ASRock’s B650E Steel Legend Wi-Fi is a mid-range offering in the B650E board realm. Priced at $269.99, it comes with that steely black-on-grey aesthetic, PCIe 5.0 support, capable power delivery, three M.2 sockets (including one 5.0 128 Gbps socket), eight USB ports on the rear IO, and more. While it’s a capable and performant board, it only sports two SATA ports. Even though M.2 drives have come down in price, many people still use more than two SATA-based storage devices. And its audio hardware, while probably fine for most people, is dated compared to the competition.

ASRock’s B650 lineup (at the time of this writing) consists of 10 different motherboards. There you’ll find all the standard sizes and wide-ranging prices, with familiar names like the Taichi, PG Riptide, PG LIghtning, Pro RS, Livemixer, and our Steel Legend SKU. Prices range from $349.99 (Taichi Carrara) to the ASRock B650M PG Riptide at $169.99. There are plenty of options in the existing lineup, including MicroATX and Mini-ITX SKUs.

Regarding performance, our Steel Legend was average to above average across our testing suite. It performed well overall in gaming and elsewhere in our benchmarks. Nothing was too fast, but nothing was inordinately slow, either. With proper cooling, you’ll be able to get the most out of your AMD processor on this board.

Below, we’ll dig into the details of the board and see whether it deserves a spot on our Best Motherboards list. But before we get into our testing and board details, we’ll start by listing the specifications from ASRock’s website. 

Specifications: ASRock B650E Steel Legend Wi-Fi

Inside the Box of the ASRock B650E Steel Legend Wi-Fi

Inside the retail packaging, below the motherboard, are a few accessories to help get you started. You get two SATA cables, screws/standoffs for M.2, the Wi-Fi antenna and the user manual. ASRock also includes a graphics card holder that attaches to the motherboard to support those heavy video cards. Like the Z790 Livemixer’s accessory stack, there isn’t much here. In fact, we just listed everything in the box.

Design of the ASRock B650E Steel Legend Wi-Fi

ASRock describes the Steel Legend as a “philosophical state of rock-solid durability and irresistible aesthetics.” Aimed at daily users and mainstream enthusiasts, you won’t find any LED screens or a fully covered board, but there are ‘steel’ colored heatsinks on top of an 8-layer, black PCB. The motherboard also has camo-like black white and grey patterns around the socket, audio area, and the chipset. 

The VRM heatsinks and shroud to the left of the socket house RGB LEDs that light up the “S” symbol for the Steel Legend. You’ll find additional RGBs under the chipset heatsink, while a large heatsink sits on top of the PCI 5.0 M.2 socket and another covers the other two M.2 sockets across the bottom of the board. Overall, this is a good-looking mid-range B650E board.

In the upper-left corner, the VRM heatsink sports the ASRock branding in white, along with the Steel Legend Series and B650E stenciled on the heatsink / shroud. The RGBs here (and on the chipset heatsink) are bright and saturated, so they will do a good job illuminating the inside of your chassis. The VRM heatsink isn’t as big as others, but it does the job well. Above the VRM heatsinks are two 8-pin EPS connectors (one required) that power the processor.

To the right of the camo-like pattern around the socket area, we run into four reinforced DRAM slots with the locking mechanism on the top. Supporting up to 128GB of DDR5, the Steel Legend supports speeds up to DDR5-6600+(OC), but your mileage may vary. Our DDR5-5600 and DDR5-6000 kits worked without issue during testing.

Above the DRAM slots are the first two (of six) four-pin fan headers. The CPU_FAN1 supports fan power up to 1A/12W, while the rest of the headers (CPU_FAN2/WP, CHA_FAN1~4/WP) support up to 2A/24W fan power. There are enough of these headers to run most, if not all, of your fans from the motherboard. The BIOS or ASRock’s A-Tune application controls these headers.

Moving down the right edge, we find the first two (of four) RGB headers. In this area are two 3-pin ARGB headers, while on the bottom of the board are additional 3-pin ARGB and 4-pin RGB headers. Just below these is the Post Status Checker. Four LEDs light up during the POST process if there is a problem with DRAM, CPU, Boot device, or VGA. Next is the 24-pin EPS connector to power the board, a front panel USB 3.2 Gen 1 (5 Gbps) connector and the front panel USB 3.2 Gen 2x2 (20 Gbps) header. There are only eight USB ports on the rear IO, so some users may really need the additional headers/ports it offers.

The VRMs on the B650E Legend consist of 19 total phases, with 16 dedicated to Vcore. Power feeds from the 8-pin EPS connectors down to the Renesas RAA229628 20-phase PWM controller, then moves to the 16x 60A Intersil ISL99360 SPS MOSFETs. The 960A available certainly isn’t the most powerful we’ve seen, but it handled our flagship Ryzen 9 7950X flagship without concern.

The audio section is hidden a bit among the stenciling on the left side of the PCB. The ever-present audio separation line is visible, running from the bottom of the board to the top M.2 socket, along with the dated Realtek ALC897 codec. You won’t find any third-party DACs or Amps at this price, so if you’re a critical listener or have a decent set of cans / speakers, you’ll want an external solution for the best results. Most others likely won’t notice the dated codec used here.

In the middle of the board are two full-length PCIe slots and three M.2 sockets. The top PCIe socket under the large heatsink connects through the CPU and runs at PCIe 5.0 x16 speeds. The bottom socket sources its lanes from the chipset, running at PCIe 3.0 x4 speeds. The board lists AMD Crossfire support, if you’re still into that sort of thing.

The top M.2 socket sits above the primary PCIe slot and gets its lanes from the CPU, running at PCIe 5.0 x4 (128 Gbps) speeds while supporting up to 80mm modules. The bottom two sockets are under a smaller (by mass) heatsink below the bottom PCIe slot. Each socket supports up to 80mm PCIe 4.0 x4 drives. With this configuration, there isn’t any sharing of ports or bandwidth, so what you see is what you get!

Moving past the illuminated chipset heatsink to the right edge, we find two SATA ports. In fact, these are the only two SATA ports on the board, which could be a problem for quite a few users looking to carry over older drives. Worst case: You can add more via a roughly $40 add-in-card, but I’d still like to see at least four ports here, especially on more budget-oriented models where M.2 drives may not be as common SATA-based SSD or HDDs.

Across the bottom of the board are several exposed headers. You’ll find the usual, including additional USB ports, RGB headers, and power/reset buttons. Below is a complete list, from left to right.

• Front panel audio
• 4-pin ARGB header
• 3-pin ARGB header
• 5-pin Thunderbolt AIC connector
• (2) System Fan headers
• Clear CMOS jumper
• (2) USB 2.0 headers
• System fan header
• USB 3.2 Gen 1 connector
• Speaker header
• System panel header

The rear IO plate on the B650E Steel Legend, like many others, comes preinstalled to the motherboard. It sports the same black, white and grey background against black labels, along with the Steel Series ‘S’ branding. There are eight total USB slots on the rear IO. You get two USB 3.2 Gen 2 (10 Gbps) Type-C and Type-A ports, four USB 3.2 Gen 1 (5 Gbps) Type-A ports, and two USB 2.0 ports. Video outputs consist of one HDMI and DisplayPort. You’ll also find the Wi-FI antenna connections for the integrated Wi-Fi 6E, a BIOS flashback button, the Realtek Dragon-based 2.5 GbE port, and last but not least, the dual analog plus SPDIF audio stack.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Firmware

ASRock’s firmware for B650 keeps the same general format as we saw in the X670E model. The major changes are the updated options/functions for the Zen4-based processors. The Steel Legend’s background matches the board theme, with a black/grey with white fonts and blue highlighting. You start in Easy Mode, which is mostly informative but lets you change a few options (XMP, profiles, boot order, Fan-Tastic Tuning, etc.). 

Advanced mode displays headings across the top with details below. Here you can tweak everything to your heart’s desire, as ASRock includes every option you can think of. Overclocking is easy, with most options on the same page, although some power options are in a different section. It’s a logical layout. The movement is smooth and it’s easy to read. No complaints from us about the ASRock firmware.

Software

For software, ASRock provides several different options. It has the App Shop to install drivers and software, the Nahimic 3 audio control panel, and the A-Tune application that overclocks your system, controls fans, and more.  There’s even a pop-up to install drivers when you first boot the system. ASRock’s software provides everything a user needs to manage and tweak their system.

Test System / Comparison Products

We’ve updated our test system to Windows 11 64-bit OS with all updates applied. We kept the same Asus TUF RTX 3070 video card from our previous testing platforms but updated the driver. We also updated to F1 22 for our games and kept Far Cry 6. We use the latest non-beta motherboard BIOS available to the public unless otherwise noted. The hardware we used is as follows: 

EVGA supplied our Supernova 850W P6 power supply (appropriately sized and more efficient than the 1.2KW monster we used previously) for our test systems, and G.Skill sent us a DDR5-5600 (F5-5600U3636C16GX2-TZ5RK) memory kit for testing.

Benchmark Settings

MORE: Best Motherboards

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 provide a great way to determine how a board runs, as identical settings should produce similar performance results. Turbo boost wattage and advanced memory timings are places where motherboard makers can still optimize for either stability or performance, though, and those settings can impact some testing.

For our synthetic testing, the B650E Steel Legend performed well overall, achieving average to above-average results in most tests. Surprisingly, it struggled a bit in the Procyon Office tests, which were average to just below average. That said, there’s nothing to be concerned about in this set of test results.

Timed Applications

Overall, the ASRock board had a good showing on the timed tests. LAME and Corona times we average or faster than average, as were both Handbrake results. There’s nothing to worry about here, either!

3D Games and 3DMark

Starting with the launch of Zen 4, we’ve updated one of our games, F1 21 to F1 22, while keeping Far Cry 6. We run the games at 1920x1080 resolution using the Ultra preset (details listed above). As the resolution goes up, the CPU tends to have less impact. 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 differences. We’ve also added a minimum FPS value, which can affect your gameplay and immersion experience.

The B650E Steel Legend is a competent gamer. It scored well in the 3DMark benchmarks and also in the game tests, though the F1 result was on the slower side of average.

Overclocking

Over the last few CPU generations, overclocking headroom has been shrinking (at least on the AMD side) while the out-of-box potential has increased. For overclockers, this means there’s less fun to have. For the average consumer, it means you’re getting the most out of the processor without manual tweaking. Our goal in this section is to increase the load on the VRMs and see if they can handle the additional stress. Overclocking AMD CPUs can be done in several ways (all-core or adjust PBO values). But for simplicity’s sake, we just went with an all-core overclock of 5.4 GHz with 1.30V to increase the power output.

Since our approach is to add power through all cores, we simply raised the CPU multiplier to 54x, manually set the voltage to 1.30V, and adjusted LLC to minimize vdroop. On the memory side, AMD states the sweet spot is around DDR5-6000, so we used the Kingston Fury kit, set the AMD EXPO profile and checked for stability.

Overclocking on the B650E Steel Legend was straightforward and easy, as it is with most boards. We were able to set our voltage, adjust LLC to mitigate vdroop, and off we went on a 30-minute stable stress test. On the memory side, we dropped our Kingston Fury Beast DDR5-6000 kit into the system, enabled XMP, and tested it without issue, as expected (pictured above).

Power Consumption / VRM Temperatures

We used AIDA64’s System Stability Test with Stress CPU, FPU, Cache and Memory enabled for power testing, using the peak power consumption value. The wattage reading is from the wall via a Kill-A-Watt meter to capture the entire PC (minus the monitor). The only variable that changes is the motherboard; all other parts remain the same. Please note we moved to use only the stock power use/VRM temperature charts, as the goal of this section is to ensure the power delivery can handle the chip even when overclocked. Since we’re using less power to get more clocks, those datasets are more novel than useful. We’re also temperature limited on the processor, so adding more power isn’t possible without increased throttling. 

Idle power consumption on our B650E Steel Legend and the Ryzen 9 7950X hit a low of 75W at the desktop/on idle. Load power peaked at only 275W at the wall, which leans more to the efficient side of boards we’ve tested on this platform.

VRM temperatures on the B650E ran a bit warmer than others during stress testing, but are still well within the MOSEFTs limits. The system peaked at just over 50 degrees Celsius while using stock settings and 52 degrees Celcius when overclocked. The VRM cooling solution works well on the 60A SPS MOSFETs hidden below.

Bottom Line

ASRock’s B650E Steel Legend is a solid all-around motherboard. Priced at $269.99 (but often on sale for $249.99), it’s a mid-range offering with a distinct ‘steel’ appearance akin to the Prime boards from Asus or Gigabyte’s Aero line. It’s relatively well-appointed, offering a 20 Gbps Type-C port (via the front panel), three M.2 sockets (one PCIe 5.0 x4), 2.5 GbE, and integrated Wi-Fi 6E. Performance was average to above average in all tests except for the Procyon Office benchmarks, which were average to slightly below average. Performance in our game tests showed there are no worries there.

But the board does have some drawbacks. First, it only has two native SATA ports. If you have more than two SATA-based drives, as many of us do, you’ll have to get an add-in card for more ports. Also, the audio codec is dated, and discerning listeners may yearn for something better. I’d also like quick latches on the M.2 sockets to avoid dealing with tiny screws, though that’s a minor quibble.

As far as the competition goes, there are better options around this price. The MSI MAG B650 Tomahawk Wi-Fi costs $249.99. Gigabyte’s B650 Aorus Pro Ax and the Asus TUF Gaming B650 Plus Wi-Fi are $239.99. These boards offer superior audio solutions, and at least four SATA ports, which is better (and for the most part, cheaper) than the Steel Legend. Overall, we do like the Steel Legend, but with the competition priced $20-$30 lower at the time we wrote this article, it’s a tough sell at the current price.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Discord, a social platform that supports text messaging, voice calls, and video calls, is enabling support for hardware-accelerated AV1 video streaming for Nvidia's GeForce 40-series graphics boards, which are among the best graphics cards money can buy today. This is the first communication platform to support both hardware acceleration of AV1 encoding and decoding using the Ada Lovelace microarchitecture. 

"Discord is currently rolling out an update that enables AV1 streaming with GeForce RTX 40 Series GPUs," tweeted Gerardo Delgado, Nvidia's product line manager for content creators. "The update is rolling out starting this week and will slowly populate to all users. With AV1 you'll be able to stream up to 4K60 with nitro, at 8 Mbps!" 

The AV1 codec significantly reduces bandwidth requirements for high-resolution, high-framerate videos. By adding support for AV1 encoding even for just one family of GPUs, Discord will reduce the upload bandwidth needed for high-quality video streams for content creators and download bandwidth requirements for viewers with hardware that supports AV1 decoding. For those with a graphics processor that does not support AV1 decoding, the service will transcode video into something ubiquitous, such as H.264. 

Nowadays, several video streaming services like Netflix and YouTube stream videos in AV1 to hardware that supports AV1 decoding. Meanwhile, Discord is currently the only communication platform that supports AV1 decoding and encoding. This may not be surprising as Discord primarily aims at gamers, and this audience tends to own the latest and greatest hardware. 

Speaking of hardware, it is necessary to note that all the latest GPUs from AMD, Intel, and Nvidia support AV1 encoding and decoding. Therefore, it is likely that Discord and, eventually, other communication software will support hardware-accelerated AV1 encoding and decoding using other GPUs.

HandBrake, the popular free and open source video transcoder, has been updated to version 1.6.0. This major point upgrade is notable for facilitating AV1 video encoding for the first time in a general release. Moreover, those with Intel Quick Sync Video (QSV) enabled processors, and those with Intel Arc GPUs will be able to encode AV1 video with hardware acceleration.

HandBrake 1.6.0 can encode AV1 videos on any of its supported systems. In the current release its SVT-AV1 encoder offers the widest support, encoding on your processor through software. However, those with Intel QSV supporting CPUs or discrete Arc graphics can use the QSV-AV1 encoder for hardware accelerated processing. QSV isn't supported if your CPU is an ‘F’ suffixed model (i.e. it doesn't have an iGPU), or it is older than the Skylake generation. If you are lucky enough to have multiple QSV accelerators in your system, support for Intel Deep Link Hyper Encode should accelerate processing further. While AMD and Nvidia have AV1 encoders available for their latest GPUs, they currently aren’t integrated with HandBrake.

AV1 video is set to become the dominant codec across app-based streaming services and the wider internet, offering attractions such as; an open and royalty-free architecture, improved compression enabling efficient 8K video streaming, and support for the newest HDR standards. Developed by the Alliance for Open Media, the AV1 standard is expected to usurp the likes of H.264/AVC and HEVC, and it looks like a sure-fire winner with the support of tech giants like Amazon, Apple, ARM, Facebook, Google, Intel, Microsoft, Netflix, Nvidia, and Samsung.

Alongside the new AV1 transcoding support, the HandBrake developers have put together several 10-bit encoder profiles, and a handful of presets for dealing with typical AV1 encoding tasks. For those still interested in H.264 and H.265 encoding there are new profiles too. Meanwhile, a multitude of the app’s built in filters are updated in v1.6.0, with many of the updates implemented to support >8-bit color depths. The latest release notes on GitHub details all the above changes, as well as the updated libraries, and tweaks such as bug fixes on various platforms.

If you are interested in grabbing the latest HandBrake release to dabble in AV1 encoding, earlier versions of the transcoding utility are currently not seeing the update as available. Instead of waiting, you can head on over to the official downloads page and download and install or upgrade your existing version. HandBrake is available for Windows 10 or later, MacOS, and Linux.

MSI’s MAG Z790 Tomahawk WIFI DDR4, our first MSI Z790 board, hails from the budget side of the product stack. Using DDR4 instead of DDR5 saves around $10-$20 on the motherboard cost and potentially a hundred dollars or more on the RAM itself. Going this route could save a lot of money (especially if you already have DDR4 RAM) that you can put into other parts, such as a video card.

At just over $300, the Tomahawk WIFI DDR4 is a full-featured option that gives you four M.2 sockets, a premium current-gen audio codec, capable power delivery, and the signature all-black appearance. Overall, it’s an attractive option in the budget Z790 space, although there are similarly equipped boards that currently cost less.

MSI’s Z790 product stack (at the time of this writing) consists of 14 motherboards in all shapes, sizes, and price points. At the top of the food chain are the MEG boards (Godlike, MEG Ace), followed by the MPG-class (Carbon WIFI, Edge WIFI, Edge WIFI DDr4, and Edge WIFI ITX). The less expensive MAG series has the Tomahawk WIFI in DDR4 and DDR5 flavors. Additionally, six “Pro” boards fill out the budget end of the company’s Z790 offerings.

Performance with our DDR4-based motherboard was solid nearly across the board. Where it lacked was mainly in the AIDA bandwidth tests, 7Zip compression, and Procyon Video editing – all were notably slower than our DDR5 systems. It did well in the Procyon Office tests and kept up with our other DDR5 offerings in the remainder of tests, including gaming. DDR4-based motherboards are just about as fast as DDR5 in utmost common tasks.

Below, we’ll dig into the details of the board and see whether it deserves a spot on our Best Motherboards list. Before we get into the nitty gritty, we’ll start by listing the specifications from the MSI website.

Specifications: MSI MAG Z790 Tomahawk WIFI DDR4

Inside the Box of the MSI MAG Z790 Tomahawk WIFI DDR4

MSI includes a couple of basic accessories along with the motherboard. This certainly isn’t a comprehensive accessory stack, but includes some of what you need to get started. Below is a complete list of the included accessories.

• Quick Install guide
• (2) SATA Data Cables
• EZ M.2 clips
• Wi-Fi antenna
• USB driver stick
• Cable stickers

Design of the MSI Z790 Tomahawk WIFI

MSI’s Tomahawk boards use a military-style theme, with an all-black appearance including the 6-layer PCB. In fact, the only contrast on the board is the reinforced PCIe slot and some numbering for the M.2 sockets. The board uses large, heavy heatsinks around the VRMs, sporting a brushed aluminum finish and other aesthetic cutouts. If you’re looking for onboard RGB lighting, the Tomahawk doesn’t have any, but there are headers to add your own. Overall, the board fits in well with most build themes, and you can add your own RGB glow if needed.

Starting with the top half of the board, we get a better look at the VRM heatsinks and their design elements. There isn’t a heatpipe shared between them, but the parts under the hood are good enough not to need a larger (and more costly) design. Sticking up between the two VRM heatsinks are two 8-pin EPS power connectors to feed the CPU (one required).

To the right of the socket, we run into four unreinforced DRAM slots with locking mechanisms on both sides. MSI lists support for up to DDR4 5333 and a capacity of 128GB. As expected, we had no issues with our DDR4 3600 or DDR4 4000 sticks. If I were to nit-pick about the RAM slots, I’d love to see single-side locking on the slots to make them easier to remove, but few will find this to be a concern.

Located above the DRAM slots, we spy our first two (of eight) fan headers and two (of four) RGB headers. Starting with the 4-pin fan headers, they all auto-detect if the attached device is PWM or DC-controlled. All of the System fan headers support 1A/12W, the CPU_Fan outputs up to 2A/24W, while the pump fan has the most at 3A/36W. Control over the attached 4-pin devices comes from the BIOS or the MSI Center software.

Along the right edge is another fan header, the EZ Debug LEDs, and the 24-pin ATX power connector supplying the board. Just below that is the front panel USB 3.2 Gen 1 (5 Gbps) header and a USB 3.2 Gen 2 (10 Gbps) header.

Moving on to the VRMs, power flows from the 8-pin EPS connector(s) onto a Renesas RAA229132 PWM controller. For Vcore, power then heads on to 16 90A Intersil ISL99390 SPS MOSFETs. The 1,440 Amps available can handle the flagship Intel Core i9-13900K at stock and overclocked. We have no concerns whatsoever about the Tomahawk’s power delivery.

On the bottom half of the board, we see a lot more of the PCB in this class than we do on the more expensive offerings. To that end, the audio section is fully exposed on the left, showing off the Realtek ALC 4080 codec and several dedicated gold audio capacitors. It may not have the fancy DAC some high-end boards come with, but it does have a latest-generation codec. Most users will be satisfied with this audio implementation.

In the middle of the board are a couple of PCIe slots, and all of the M.2 sockets. Starting with the slots, There are two full-length PCIe slots and one x1-size slot. The top full-length slot (primary graphics slot) is reinforced to prevent shearing from heavy video cards potentially help with EMI. This slot connects through the CPU and is the only one that is PCIe 5.0-capable. The bottom full-length slot connects through the chipset and runs up to PCIe 4.0 x4, while the x1 slot is PCIe 3.0 x1 and also sources its bandwidth from the chipset. Neither the specs nor the manual mention Crossfire support, even though it is capable bandwidth-wise.

For M.2 storage, there are four sockets mixed in among the PCIe slots. Starting from the top, M.2_1 sources its lanes from the CPU and supports up to PCIe 4.0 x4 (64 Gbps) modules up to 110mm. The next three sockets get their bandwidth from the chipset and run at PCIe 4.0 x4 (64Gbps) and up to 80mm modules. M.2_3 supports SATA-based storage devices as well as PCIe.

Across the bottom of the board are several exposed headers. You’ll find the usual, including additional USB ports, RGB headers, and power/reset buttons. Below is a complete list from left to right.

• Front panel audio
• 4-pin ARGB header
• 3-pin RGB header
• Thunderbolt header
• (3) System Fan headers
• (2) USB 2.0 headers
• (1) System Fan header
• SATA port
• System panel header

Out back on the preinstalled rear IO, the Z790 Tomahawk sports a black background with light grey labels on all of the ports. There are 10 USB ports: Six 10 Gbps ports (red, plus two Type-C), and four 5 Gbps ports (blue). The DisplayPort and HDMI ports sit on the far left, next to tiny clear CMOS and BIOS Flashback buttons. The Intel 2.5 GbE port is above one of the Type-C ports, with the Wi-Fi 6E antenna connections to the right, next to the 5-plug analog plus SPDIF audio stack.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Firmware

MSI’s BIOS for the Tomahawk hasn’t changed. The menus reside on the sides and information up top, a unique look compared to others. For Z790, the only updates were functionality for the new processors and chipset. It still has a black background with red highlights, making it easy to read. It also has an informative Easy Mode that displays various information about the system and allows for some changes (boot order and XMP enable). Though different than other UEFIs, it is easy to get around; everything has a place, and many frequently used options are readily available, not buried within sub-menus.

Software

MSI has a single utility, MSI Center, that covers quite a bit of functionality. From hardware monitoring to RGB control with Mystic Light, there are many applets to choose from within the software and a one-stop shop to download all of their utilities. I wish there were some overclocking options, however. But for those looking to use the Gamebar feature, Super Charger, or any other utilities MSI offers will all be at your fingertips in MSI Center.

Test System / Comparison Products

We’ve updated our test system to Windows 11 64-bit OS with all updates applied. We kept the same Asus TUF RTX 3070 video card from our previous testing platforms but updated the driver. We also updated to F1 22 for our games and kept Far Cry 6. We use the latest non-beta motherboard BIOS available to the public unless otherwise noted. The hardware we used is as follows:

Test System Components

EVGA supplied our Supernova 850W P6 power supply (appropriately sized and more efficient than the 1.2KW monster we used previously) for our test systems, and G.Skill sent us a DDR5-5600 (F5-5600U3636C16GX2-TZ5RK) memory kit for testing.

Benchmark Settings

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

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 provide a great way to determine how a board runs, as identical settings should produce similar performance results. Turbo boost wattage and advanced memory timings are places where motherboard makers can still optimize for either stability or performance, though, and those settings can impact some testing.

Across our synthetic benchmarks, the Tomahawk is the fastest in some and the slowest in others. But like most results (that aren’t affected by DDR4 heavily like the 7Zip compression test) the peaks and lulls are difficult to notice unless you’re specifically looking for them. Performance is right up there with the others across most tests.

Timed Applications

In the LAME benchmark, the Tomahawk was middle of the pack, but it was the fastest in Corona (by one second, minimal difference). During the Handbrake x264 testing, MSI’s board was the slowest by a couple of seconds, while the x265 test showed one of the faster results. We found nothing anomalous in the timed applications.

3D Games and 3DMark

Starting with the launch of Zen 4, we’ve updated one of our games, F1 21 to F1 22, while keeping Far Cry 6. We run the games at 1920x1080 resolution using the Ultra preset (details listed above). As the resolution goes up, the CPU tends to have less impact. 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 differences. We’ve also added a minimum FPS value, which can affect your gameplay and immersion experience.

The Tomahawk is a more than competent gaming board. The 3D mark results both landed in the middle of the other three data sets while it lead the pack in our Far Cry 6 benchmark in both minimum and maximums (though only by a frame on minimums) and tied for the fastest result in F1 2022. Here again, all of the results are close, and difficult notice a difference beyond benchmarks and frame counters.

Overclocking

When overclocking, we aim to increase the power and add stress to the VRMs. We do so by increasing the clock speed and voltage until we’re at the thermal threshold for the processor during stress testing. However, where these CPUs are allowed to run with ‘boost’ speeds, those speeds are closer to the limit than ever before.

With our i9-13900K, the approach was a bit different. We had to lower the voltage from what we recorded during stress tests to overclock our chip. We increased the clock speeds of the “P” and “E” cores by 100 MHz and 200 MHz, respectively, over the turbo boost and limited by our cooling. We ended up with  5.6 GHz P core and 4.5 GHz E core clocks using about 1.34V (from DMM). Temperatures peak just under the throttling point with this configuration. And as we’ve noted elsewhere, this is the intended performance for Intel’s flagship CPU.

Overclocking on the Tomahawk was the easiest of our tested boards so far. We set 1.35V in the BIOS with LLC (Load Line Calibration) set to 2, and the load voltage in Windows matched what was set in the BIOS. The board then ran stable and didn’t throttle using the same or less power than the out-of-the-box settings.

We had no issues with both memory kits. We used the latest BIOS, running them stable throughout our testing. The Tomahawk’s DDR4-5333 MHz is one of the higher ratings listed, so there’s plenty of headroom for those who want to push past DDR4’s performance sweet spot.

Power Consumption / VRM Temperatures

We used AIDA64’s System Stability Test with Stress CPU, FPU, Cache and Memory enabled for power testing, using the peak power consumption value. The wattage reading is from the wall via a Kill-A-Watt meter to capture the entire PC (minus the monitor). The only variable that changes is the motherboard; all other parts remain the same. Please note we moved to use only the stock power use/VRM temperature charts, as the goal of this section is to ensure the power delivery can handle the chip even when overclocked. Since we’re using less power to get more clocks, those datasets are more novel than useful. We’re also temperature limited on the processor, so adding more power isn’t possible without increased throttling.

Idle power consumption for the Z790 Tomahawk was 62 watts (the lowest so far), while the load power use was 381 watts, several more than the Z790 Aorus Master (but nowhere close to the Taichi’s 498W number). It’s still early, but we expect the Tomahawk’s numbers to be around average after we get more data sets.

VRM temperatures on our board peaked at almost 58 degrees Celsius, just below the 500W actively cooled Taichi and around 10-plus degrees hotter than the Aorus Master and Hero. These temperatures are well within the specifications of the power delivery bits, so there’s nothing to worry about on this front. The combination of  90A SPS MOSFETs and the ample heatsinks is a great match.

Editor's Note 12/15/2022: After contacting MSI about an error on the webpage, they confirmed the power delivery on the Tomahawk (DDR4 and DDR5) is actually a "mirrored" configuration using Renesas ISL6617A phase doublers.

Bottom Line

MSI’s MAG Z790 Tomahawk WIFI DDR4 presents builders with a less expensive way to get into the latest and greatest Intel platform. It does so by using DDR4 instead of DDR5. Since DDR4 memory kits are still much less expensive and the boards are $10-$20 cheaper than their DDR5 counterparts, you can save a fair amount of money going this route.

But it’s not all about the money! The Tomahawk WIFI DDR4 comes well-equipped, integrating many things the platform offers, including a PCIe 5.0 slot, integrated Wi-Fi 6E, and the latest generation audio codec. However, it is missing a PCIe 5.0-capable M.2 socket. Performance was good, except in the few tests that show bandwidth matters. It’s a more than capable gaming machine for those who aren’t concerned with productivity. Overall, it’s a well-rounded option, specs-wise.

There’s not a lot of direct competition close to the $300 price point. If you have the budget for DDR5 sticks, there are several DDR5 boards around this price, including a DDR5 Tomahawk. Other DDR4 options include the Gigabyte Z790 Aorus Elite AX DDR4 ($259.99), Asus TUF Gaming Z790-Plus WiFi D4 ($289.99), and the ASRock Z790 Pro RS/D4 – almost 33% less at $209.99. All of these boards have capable power delivery, 2.5 GbE, four PCie 4.0 x4 M.2 sockets, and Wi-Fi 6E.

Mostly, the difference between these boards lies in the audio codec (Tomahawk is the only one with the latest generation Realtek). However, it would be difficult to discern among most of them outside of the significantly cheaper ASRock board that uses a dated, budget codec. It’s going to come down to looks (which we all know is subjective) and price. I like the all-black look and aesthetic of the Tomahawk and Gigabyte boards over the Asus TUF and ASRock. But I don’t think this board is $50 better looking than the Gigabyte.

In short, we like MSI’s MAG Z790 Tomahawk WIFI DDR4. The hardware is comparable to the competition and it looks good. But the price is high for what we’re currently seeing on the market. If this were somewhere between $250 and $275, it would be a much more compelling option. Still, there’s no doubt it’s worth a look among other DDR4 boards. If you like the look and / or you need the seven SATA ports, it’s a great board to build your new Raptor Lake system around.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Moore Threads Intelligent Technology, a major graphics processors developer from China, on Thursday announced its next generation GPU that can be used for gaming, artificial intelligence, and datacenter workloads. The MTT S80 gaming graphics cards as well as MTT S3000 server board promise to deliver compute performance akin to that of Nvidia's GeForce RTX 3060 Ti, though their actual performance in games and professional applications is something that remains to be tested. 

The Chunxiao GPU

Moore Threads's new graphics processor is based around the company's Chunxiao architecture that supports FP32, FP16, and INT8 precision and is compatible with the company's MUSA computing platform and application programming interface (it can work with standard APIs too though). The MTT Chunxiao GPU is clocked at 1.80 GHz – 1.90 GHz and packs 4,096 stream processors, 128 tensor cores, 256 texture units, and 256 render output. The GPU features a 256-bit memory interface and can work with GDDR6 memory with a 14 GT/s data transfer rate. As for host bus, it features 16 PCIe Gen5 lanes and fully supports the SR-IOV specification for PCIe virtualization in server environments. The GPU up to 32-way partitioning, which will be handy for Android games rendering.  

The chip also comes with a rather powerful video engine that supports AV1, H.264, and H.265 codecs for up to 8K videos as well as can decode up to 32 streams at a 1080p30 resolution. As for output, the GPU has four display engines supporting an up to 8Kp30 resolution and has three DisplayPort 1.4 as well as one HDMI 2.1 interface. 

Moore Threads says that depending on clocks, its Chunxiao GPU can deliver 14.4 FP32 TFLOPS or 15.2 FP32 TFLOPS, which is comparable to single precision compute performance of Nvidia's GeForce RTX 3060 Ti GPU. On paper, this is enough to make it to our list of the best gaming graphics cards available today, but how the ChunxiaoGPU works in real life is something that we will have to test. 

Meanwhile, since the Moore Threads Chunxiao graphics processor is aimed at AI, gaming and datacenter workloads, it comprises of 22 billion of transistors, which is fairly complex. To put the number into context, Nvidia's GA104 features 17.4 billion transistors, whereas AMD's Navi 21 packs 26.8 billion of transistors. Moore Threads does not disclose which process technology it uses to make its Chunxiao graphics processor.

Two Products Coming 

For now, Moore Threads plans to offer two products based on the Chunxiao GPU — the MTT S80 graphics card with a 14.4 FP32 TFLOPS throughput and 16GB of memory as well as the MTT S3000 server card with a 15.2 FP32 TFLOPS compute performance and 32GB of memory. 

Moore Threads has not disclosed power consumption of its MTT S80 and MTT S3000 products, though the former comes with a rather sophisticated cooling system with three fans. 

One of the main advantages that the Moore Threads Chunxiao offers is its broad compatibility; It can work with a variety of client and datacenter hardware platforms (Arm, Ampere, Intel, etc.) and operating systems. It is compatible with Microsoft's DirectX (and therefore will ship with appropriate drivers for Windows), Khronos Group's OpenGL/OpenGL ES, proprietary MUSA as well as multiple specialized APIs. Furthermore, the Chunxiao can work with PyTorch, TensorFlow, PaddlePaddle, Jittor and other mainstream deep learning frameworks and popular AI models. 

To ensure compatibility with modern games, Moore Threads says it had worked closely with developers of Unreal Engine and Unity as well as creators of popular titles (e.g., Call of Duty, Crossfire, Counter Strike, Diablo 3, League of Legends, etc.). Meanwhile, Moore Threads admits that for now its Chunxiao only properly supports about 20 DirectX titles and does not make any promises about performance. 

The GPU developer plans to start selling its MTT S80 graphics cards at JD.com on November 11, 2022.  No pricing has been announced.

Scientists from the Technical University of Denmark in Copenhagen have achieved 1.84 petabits per second data transfers using a single photonic chip connected via a single optical fiber cable. The feat was accomplished over a distance of 7.9 km (4.9 miles). For some perspective regarding this achievement, at any time of day, the average internet bandwidth being used by the whole world’s population is estimated to be about 1 petabit/s.

With the ever increasing amounts of data shifted across the internet for business, for pleasure, and software downloads or updates - infrastructure firms are always on the lookout for new ways to increase the available bandwidth. The 1.84 petabits/s over a standard optical cable using a compact single chip solution will therefore hold much appeal.

Photonic chip technology holds great promise for optical data transfer purposes – as the processor and the transfer medium both work with light waves. The New Scientist explains in simple terms how the Danish scientists, led by Asbjørn Arvad Jørgensen, managed to deliver such bandwidth with the resources at hand.

Firstly, the data stream used in the trial was split into 37 lines, with each one sent down a different optical thread in the cable. Each of the 37 data lines were split into 223 data chunks corresponding to zones of the optical spectrum. What this allowed is for creating a "frequency comb" where data was transmitted in different colors at the same time, without interfering with other streams. In other words a “massively parallel space-and-wavelength multiplexed data transmission” system was created. Of course, this splitting, and re-splitting massively increased the potential data throughput supported by a fiber optic cable.

It wasn’t easy to test and verify 1.84 petabits/s bandwidth – as no computer can send, or receive, never mind store, such a humungous amount of data. The research team used dummy data over individual channels to verify what would be the full-on bandwidth capacity. Each channel was tested individually to ensure data received matched what was transmitted.

In action, the photonic chip splits a single laser into many frequencies and some processing is required to encode light data for each of the 37 data optical fiber streams. A refined fully capable optical processing device should be possible to build at approximately the size of a match box, according to Jørgensen. This is a similar size to current single color laser transmission devices used by the telecoms industry.

Ut is reassuring that we will be able to keep the same fiber optical cable infrastructure, but replace matchbox-sized optical data encoders / decoders with the similar sized photonic chip powered devices, potentially delivering an effective 8,251x increase in data bandwidth. The researchers say there is enough potential shown in their work to inspire “a shift in the design of future communications systems.”

For further information about the record 1.84 petabits/s data transfers you can check out the Petabit-per-second data transmission using a chip-scale microcomb ring resonator source paper.

VESA this week released version 2.1 of the DisplayPort specification. The update brings several important changes concerning efficiency, but as you will read, VESA strove to keep things smooth for both implementers and consumers.

DisplayPort 2.1 is more than just backward compatible with DisplayPort 2.0. Due to the efforts of VESA and its partners, "all previously certified DisplayPort 2.0 products… have already been certified to the stricter DisplayPort 2.1 spec." So yes, any GPU, Docking Station, monitor, cable (or anything) certified for DisplayPort 2.0 will fall under the stricter newer v2.1 specification.

VESA says one of the most important things it has tried to achieve with this update is to offer a robust experience. This philosophy applies to users connecting peripherals via a native DisplayPort cable, via DisplayPort Alt Mode (DisplayPort over the USB Type-C connector), or tunneling through the USB4 link. Whichever solution you might use, the alignment of the spec and PHY (physical layer) has been tightened for dependability.

So, if all these things remain the same, and DisplayPort 2.0 and 2.1 are interchangeable, you might be wondering what's the point of the new specification. Perhaps the most important introduction is the new DisplayPort bandwidth management feature "to enable DisplayPort tunneling to coexist with other I/O data traffic more efficiently over the USB4 link." According to the VESA press release, this efficiency comes on top of new mandated support for VESA's visually lossless Display Stream Compression (DSC) codec and VESA's Panel Replay capability. VESA's DSC implementation will reduce bandwidth usage by as much as 67% without visual artifacts. Moreover, the new Panel Replay capability can reduce tunneling packet transport bandwidth by over 99% in certain situations.

VESA Board Chair and VESA DisplayPort Task Group Chair Alan Kobayashi boldly claimed that the higher bitrates supported by DisplayPort 2.1 "offers ample bandwidth for the needs of virtually every practical application."

The announcement also revealed that cable specs were updated. With DisplayPort 2.1, VESA aims to allow for improved capabilities and longer cable lengths, sometimes extending over two meters. "VESA certified DP40 cables support up to the UHBR10 link rate (10 Gbps), with four lanes, providing a maximum throughput of 40 Gbps," VESA explains. "VESA certified DP80 cables support up to the UHBR20 link rate (20 Gbps), with four lanes, providing a maximum throughput of 80 Gbps."

Interestingly, ex-HardOCP Editor-in-Chief Kyle Bennett tweeted last week that multiple sources had indicated to him that AMD Radeon Navi 31 GPUs would offer support for the DisplayPort 2.1 specification. This summer, there were rumors of RDNA 3 supporting DisplayPort 2.0 and UHBR20. Intel Arc LE model GPUs also support DisplayPort 2.0. Considering what VESA has said, both the AMD and Intel GPUs will support DisplayPort 2.1. Putting these into competitive perspective, the recently launched Nvidia Ada Lovelace flagship, the GeForce RTX 4090, supports DisplayPort 1.4.

According to Phoronix, Google has released a new AOM-AV1 update - version 3.5, that drastically improves encode times when streaming, rendering, or recording from the CPU. At its best, the update can improve encoding times by up to 34%. 

It is a fantastic addition to AV1's capabilities, with the encoder becoming very popular among powerful video platforms such as YouTube. In addition, we are also seeing significant support for AV1 hardware acceleration on modern discrete GPUs now, such as Intel's Arc Alchemist GPUs and, most importantly - Nvidia's RTX 40-series GPUs.

Depending on the resolution, encoding times with the new update have improved by 20% to 30%. For example, at 1080P, encode times featuring 16 threads of processing are reduced by 18% to 34%. At 4K, render times improved by 18% to 20% with 32 threads. Google could do this by adding Frame Parallel Encoding to heavily multi-threaded configurations. Google has also added several other improvements contributing to AV1's performance uplifts in other areas - specifically in real-time encoding.

In other words, CPU utilization in programs such as OBS has been reduced, primarily for systems packing 16 CPU threads. As a result, they are allowing users to use those CPU resources for other tasks or increase video quality even higher without any additional performance cost. If you are video editing and are rendering out a video in AV1, processing times will be vastly reduced if you have a CPU with 16 threads or more.

AV1's Performance Improvements Come At The Perfect Time

AV1's speed improvements couldn't have come at a better time. The video industry is aggressively moving to AV1 more than ever this year, thanks to the introduction of AV1 hardware acceleration engines and allowing content creators to use discrete GPUs to record and stream content with the AV1 codec.

AV1 came on the scene in 2018 as a newer potential alternative to the H.264 video codec. It has become a mainstream item due to its very attractive feature set; for one, it is entirely open-source, allowing anyone to use it for free. It provides up to 50% higher compression ratios than H.264. In addition, it is drastically reducing AV1 video file sizes.

This year we saw our first AV1 hardware acceleration engine in the Intel Arc A-series GPUs, providing drastic video quality improvements over its competitors, including Nvidia's legendary NVENC H.264 encoder. But now, Nvidia has finally entered the game with its RTX 40-series GPUs supporting AV1 encoding for the first time - thanks to the brand-new 8th generation NVENC engine.

It sets up the stage for AV1 to become a complete replacement for H.264, now that we have AV1 encoding present in both discrete GPUs and via software encoding on the CPU, which will make AV1 encoding very accessible to the public. The new AV1 performance update is just the cherry on top and will provide greater flexibility for users running the AV1 codec on live streams and within video editors.

Biostar’s B660M Silver is a Micro ATX motherboard that hailing from the budget end of the Alder Lake spectrum. Priced around $160 in the US, it offers just about everything you need to get the most out of your system–so long as you don’t want to overclock your CPU. The B660M Silver brings a black and silver aesthetic (no RGBs), two PCIe 4.0 x4 M.2 sockets, four SATA ports, capable power delivery, and a 2.5 GbE port. Overall, it’s a well-rounded option in this space, but lacks included Wi-Fi.

Performance on the B660M Silver was comparable to other like boards across our testing suite. It struggled a bit in a couple of the heavily multi-threaded tests, but that’s a common theme among most B660 boards. Biostar’s board excelled in the Procyon Office suite, with results near the top across all those tests. If you’re a gamer, there’s nothing to fear, as the performance there was average, too. Power consumption was also average overall, though load power consumption was higher than many others.

Before we get into all the details and see if this board earns a spot on our best motherboards list, below is a list of specifications from the Biostar website.

Specifications: Biostar B660M Silver

Inside the Box of the Biostar B660M Silver

Biostar, like all other board partners, includes some accessories to get you started. The B660M Silver’s accessory stack is sparse, but includes enough to get you started. Below is a complete list of the included accessories.

• (4) SATA cables
• I/O Shield
• Support DVD
• Quick guide

Design of the B660M Silver

The B660M Silver is a general purpose board that, according to Biostar, focuses on reliability. The black PCB gives way to light gray design elements stenciled on the board around the PCIe and socket areas. Cooling the power delivery, chipset, and one M.2 socket are silver heatsinks with a brushed aluminum finish. These aren't the biggest heatsinks we’ve seen, but they do the job. If you’re looking for integrated RGB LEDs, look elsewhere as the Silver doesn’t have any. But don’t fret if you want lights, because the board has a full complement of ARGB and RGB headers to add on your own.

On the top half of the board, we get a better look at the VRM heatsink that reaches out over the rear IO bits. It’s not the largest or heaviest VRM heatsink, but it keeps these MOSFETs running with spec. Across the top of this heatsink is the Silver branding in black, Biostar is stenciled on the chipset heatsink. Just above the VRM heatsink is the single 8-pin EPS connector (required) to power the processor.

Moving past the socket and small top VRM heatsink, we run into the first (of five) 4-pin fan/pump headers.  Biostar doesn’t list the output of these headers, so I would expect 1A/12W output for each, if only for safety reasons. To adjust the fans, use the Aorura software and AI Fan program.

Continuing right, we run into four unreinforced DRAM slots that only lock on the top side. Biostar lists support for up to 128GB of RAM and speeds up to DDR4-5000+(OC). We didn’t run into any issues with our two test kits with speeds up to DDR4-4000. As usual, your mileage may vary in reaching these (and faster) speeds as it depends on the memory kit and the processor’s integrated memory controller (IMC) to get you there.

In the upper right corner, you’ll find all three RGB headers: two 3-pin ARGB headers and one 4-pin RGB header. Control over the attached lighting comes from the Aorura software and Vivid DJ applet. Down the right edge, we run into the 24-pin EPS for board power, a 19-pin USB 3.1 Gen 1 (5 Gbps) header, and a USB 3.2 Gen 2 (10 Gbps) Type-C header.

Power delivery on the B660M Silver consists of 12 phases for Vcore (13 total). Power comes from the 8-pin EPS connector and onto the Renesas RAA229130 controller. Power then moves to 12 70A OnSemi 5062 DrMOS MOSFETs. The 840A available isn’t the most we’ve seen, but it is plenty to handle our Intel Core i9-12900K at default speeds without getting in the way.

On the bottom half of the board, we spy a fully exposed audio section on the left side. To the left of the audio separation line, we run into a Realtek ALC1220 codec and four dedicated audio caps. Compared to most of its peers (except the B660M Mortar), this is a superior audio solution on paper, but tough to differentiate audibly.

In the middle of the board are three M.2 sockets and three PCIe slots. Starting with the slots, the B660M  Silver sports two full-length slots (one reinforced) and a single x1 slot. The top slot (primary) connects through the CPU and runs at PCIe 5.0 x16. The bottom slot connects through the chipset, running at PCIe 3.0 x4. Multi-GPU support isn’t mentioned, though the board has the bandwidth to support it. The middle x1 slot also connects through the chipset and supports up to PCIe 3.0 x1 speeds.

Regarding M.2 sockets, there are two Key-M slots for storage and one Key-E socket for Wi-Fi (CNVi module not included). The top socket connects through the CPU for PCIe 4.0x4 (64 Gbps) bandwidth and handles up to 80mm modules. The bottom socket supports up to 110mm modules and connects through the chipset for PCIe 4.0 x4 (64 Gbps) bandwidth.

Across the bottom are several headers, including USB ports and RGB, and more. Below is a complete list, from left to right:

• Front panel audio
• (3) 4-pin fan headers
• (2) USB 2.0
• 9-pin Thunderbolt header
• Debug LEDs
• COM port
• Front panel
• 4-pin RGB header

The B660M silver includes a simple rear IO plate you need to install. It has a black background with silver labels and “Silver” branding in the upper-left corner. As far as ports go, from left to right, we run into video outputs – HDMI, DisplayPort, and DVI-D. Above that are Wi-Fi antenna connections which, I find misleading since this model doesn’t come with Wi-Fi. There are eight USB ports: two USB 2.0 (480 MBps) and six USB 3.2 Gen2 (10 Gbps) ports, one of which is Type-C. Next up is the 2.5 GbE port and a 3-plug analog input-only audio stack.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Firmware

Biostar’s B660 BIOS has the same look as the last generation and the Z690 Valkyrie reviewed a few months back. There’s an EZ Mode that's more informational than tweakable, covering high-level status information, including RAM and CPU speed, temperatures and more. Once in the Advanced section, there are headings across the top with options under each. There are a fair amount of switching sections to reach all of the overclocking functionality, and listing some values in milliwatts is confusing. I’d also like to see a description for highlighted items, as most other boards have. But outside of that, the BIOS is easy to navigate and read, with a high-contrast dark grey background and white letters.

Software

Biostar uses a single application called Aurora to cover multiple functions, including Fan control, RGB lighting, monitoring, overclocking and more, but little has changed since the Z690 version. Once opened, there is a menu system in the left column with a larger right-hand side to display the information. With Aurora, you can control system volume (Smart Ear), canned power plans (GT Touch), RGBs (Vivid LED DJ), Fans (A.I Fan), hardware monitoring (H/W Monitor) and overclocking/overvolting (OC/OV).

Overall, I like this software. The user interface is easy both to read and to find the items you are looking for. Fan, RGB, and overclocking control worked without issue. I would like to see more temperatures in the hardware monitor section (VRMs, for example), but what’s there is informative, if only at a high level.

Test System / Comparison Products

We’ve updated our test system to Windows 11 64-bit OS with all updates applied. We kept the same Asus TUF RTX 3070 video card from our previous testing platforms but updated the driver to version 496.13. Additionally, our game selection has been updated, as noted in the table below. We use the latest non-beta motherboard BIOS available to the public unless otherwise noted. The hardware we used is as follows:

Test System Components

EVGA supplied our Supernova 850W P6 power supply (appropriately sized and more efficient than the outgoing 1.2KW monster we used) for our test systems, and G.Skill sent us a DDR5-5600 (F5-5600U3636C16GX2-TZ5RK) memory kit for launch day testing.

Benchmark Settings

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Benchmark Results and Final Analysis

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.

To get the most out of the Intel Alder Lake chips, you need to be on Windows 11 with its updated scheduler. In most cases, Windows 10 performs well. However, some tests (Cinebench R20, Corona and POVRay) take a significant hit. In short, if you’re going with Alder Lake, you must upgrade to Windows 11 for the best results across the board. That may change with patching and updates in the future, though.

Synthetic Benchmarks

Synthetics provide a great way to determine how a board runs, as identical settings should produce similar performance results. Turbo boost wattage and advanced memory timings are places where motherboard makers can still optimize for either stability or performance, though, and those settings can impact some testing.

The B660M Silver performed well during most synthetic benchmarks, landing around the average of other DDR4-equipped motherboards. The only below-average tests were Cinebench R23 and POV-Ray’s multi-threaded tests. But if you’ve read our previous reviews, you know that’s a recurring theme due to the thermal throttling of the processor. The Procyon tests were all averag, as was the AIDA64 memory bandwidth.

Timed Applications

In the timed applications, the B660M Silver mixed in with average for LAME and Corona but was slower than most in Handbrake. LAME testing was completed in 9.60 seconds (average is 9.69) and 52 seconds in Corona (spot on the average). The x264 test completed in 120 seconds, against a 115-second average, and 305 seconds in x265 (barely above the average of 303 seconds). Nothing off here either.

3D Games and 3DMark

Starting with the launch of the Z690 chipset, we’ve updated our game tests, moving to Far Cry 6 and shifting from F1 2020 to F1 2021. We run the games at 1920x1080 resolution using the Ultra preset (details listed above). As the resolution goes up, the CPU tends to have less impact. The goal with these settings is to determine if there are differences in performance at the most commonly used resolution with settings most people use or at least strive for. We expect the difference between boards in these tests to be minor, with most falling within the margin of error differences. We’ve also added a minimum FPS value, which can affect your gameplay and immersion experience.

In F1 2021, the B660M Silver proved itself a performant entrant, lurking around the average of all the boards with 141 frames per second minimum and 164 frames per second average. In Far Cry 6, the story was the same, as our testing showed a 121 frames per second minimum and 137 frames per second average.

In our synthetic 3DMark tests, the Biostar scored 14,201 on Time Spy and 17,899 on Fire Strike Extreme. All the results for games and gaming benchmarks blend in with our other results and shows the B660M Silver is a competent gaming motherboard.

Power Consumption / VRM Temperatures

We used AIDA64’s System Stability Test with Stress CPU, FPU and Cache enabled for power testing, using the peak power consumption value. The wattage reading is from the wall via a Kill-A-Watt meter to capture the entire PC minus the monitor. The only variable that changes is the motherboard; all other parts are the same.

At idle, the DDR4-based Silver consumed 47W from the wall, with load power shooting up to 345W. While the idle value is good and better than most, the load wattage was higher than average.

During stress testing, we saw very little throttling, which was surprising. The 70A MOSFETs and the heatsink keep the power bits running within specification. The board peaked around 66 degrees Celsius on our external sensors and 78 degrees Celsius in software, both well within specification.

Overclocking

Overclocking the CPU isn’t possible on B660-based chipsets, but the platform allows memory speed adjustment. With our DDR4-3600 and DDR4-4000 kits, we simply set XMP, and off we went without a hitch. Surely there’s some headroom left, but unless you’re chasing after every ounce of performance, we suggest sticking with the sweet spot, around DDR4 3600 with a low CL rating.

Bottom Line

Priced at $159.99, the Biostar B660M Silver is an inexpensive Micro ATX Alder Lake option for the general user. The board doesn’t do anything poorly, but also doesn’t deliver anything spectacular. That isn’t a bad thing, though! Hardware-wise, the B660M Silver includes what’s expected for this class board. You get capable power delivery, two M.2 sockets and four SATA ports, premium last-gen audio, plenty of USB ports on the rear IO, and more. While there are plenty of comparables in this space, it’s still a worthy option to for your Alder Lake processor.

As far as competition goes, all major partners have an option around this price range. The ASRock B660M Steel Legend is the least expensive at $114.99 (down from $159.99). Asus’ TUF Gaming B660M-Plus WIFI D4 sits at $169.99. Gigabyte’s B660M Aorus Pro AX is also $169.99, and the MSI MAG B660M Mortar WIFI DDR4 is  $159.99, just like the Biostar. Most of the hardware in this range is similar, offering two M.2 sockets and at least four SATA ports (MSI and ASRock have six). The difference comes in audio codecs (Biostar and MSI have the best here), integrated Wi-Fi (the Biostar and ASRock do not come with it), and maybe a few other things like USB port count and speed. If it’s looks you’re after and need integrated RGB, ASRock and Gigabyte have you covered on that front.

The Biostar B660M Silver is a great affordable way into the Micro ATX Alder Lake space. The black-on-silver design fits most build themes and doesn’t draw attention to itself out of the box. Hardware-wise, it has most everything users are looking for. But for the price, I would like to see Wi-Fi included as most of the other comparables include that handy feature. If you’re looking for a B660-based Micro ATX motherboard, the B660M Silver doesn’t stand out compared to others, but is still a worthwhile consideration for your Alder Lake needs. 

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

In a surprising twist, Intel's future 14th Generation Meteor Lake CPU architecture will reportedly split video playback and encode functionality from the integrated graphics into a new area known as an SMU or Standalone Media Unit on the CPU, as reported by Phoronix. It is a significant change by Intel, which will allow media functionality to be used at all times, even if the integrated graphics chip is disabled.

In a real-world scenario, once Meteor Lake launches, users building gaming PCs or content creation machines on the future platform will have access to all of Intel's high-quality video decoding engines and Intel QuickSync technology, even if integrated GPU is disabled in favor of a discrete graphics card, which frees up memory resources and power resources -- moving them towards the CPU entirely. The situation gets even better if older discrete GPUs are installed that lack modern hardware acceleration codecs, such as AV1 or H265 encoding.

Another perk of this transition is related to Intel's F-series processors, which lack integrated graphics altogether. As a result, F-series Meteor Lake processors should theoretically gain access to all of Intel's media features - including QuickSync, even with the iGP disabled at a silicon level. In addition, Intel server CPUs and potential HEDT chips could also benefit from this change since Intel disables the integrated graphics chip from these CPU lineups altogether.

Phoronix saw this change based on new Linux enablement work for Meteor Lake, which includes patch notes detailing the SMU changes. These changes include further architectural modifications to the graphics unit and transitioning to Intel's Arc Alchemist GPU architecture. As a result, we should expect more advanced features such as AV1 encoding to make their way to consumer-based Meteor Lake chips.

Unfortunately, we won't see these new changes for a while since Meteor Lake is still at least a year or two away from release, with Intel busy preparing its upcoming 13th Gen Raptor Lake lineup. But, thanks to early Linux development work on Meteor Lake, we see exciting hardware changes on future CPU architectures before launch.

AV1 encoding has finally made its way to the public via integration with Intel's new Arc Alchemist GPUs — the first GPUs to feature this technology. According to a new video by EposVox, this new encoder has been tested on an Arc A380 and provides incredible efficiency for low-bit-rate video streaming, dominating every H.264 hardware encoder in existence — including Nvidia's tried and true NVENC encoder.

AV1 is an open-source, royalty-free video coding format developed by the Alliance for Open Media, a consortium founded in 2015. AV1 is a free, state-of-the-art codec that anyone can use on the internet. AV1 provides massively better compression performance with up to 50 percent smaller file sizes compared to H.264. 

AV1 has gained major traction in the video streaming industry in recent years — AV1 decoding engines already supported in many of the latest graphics engines, including Nvidia's RTX 30 series Ampere architecture, AMD's RX 6000 series' RDNA2 GPUs, and Intel's latest integrated graphics GPUs. Even older gaming consoles, such as Sony's PlayStation 4 Pro, support the AV1 codec.

According to EposVox, video streaming platforms such as YouTube have also adopted AV1 extensively in recent years — the YouTuber says nearly half the videos he watches on a daily basis support the new codec.

AV1 encoding for content creating and video streaming has been largely absent, despite AV1's adoption. None of the latest graphics engines support AV1 encode engines. You can definitely use AV1 encoding on the CPU with software encoding, but for some reason nobody has bothered to implement a hardware-accelerated method of AV1 encoding. Until now, of course, with Intel leading the charge with its Arc GPUs.

Arc A380 AV1 Quality Comparison

EposVox tested Intel's new AV1 encoder against a variety of H.264 encoders, including Intel's own Quick Sync technology, Nvidia's NVENC, AMD's AMF, and software-based H.264 encoders such as the one found in OBS. 

Testing was performed at 3.5 Mbps, 6 Mbps, and 8 Mbps using Netflix's VMAF benchmarking tool which analyses video quality with a scoring system of 0 to 100 (0 is unwatchable,100 is perfect video quality — compared to uncompressed video). 

The AV1 results are impressive: the new Intel encoder scored 83 points at 3.5 Mbps and 90 points at 6 Mbps — the highest of all the H.264 encoders at these bit rates. (This first example was taken from videos recorded from Battlefield 2042.)

The Nvidia NVENC encoder — surprisingly — produced one of the lowest test results:  71 points at 3.5 Mbps and 85 points at 8 Mbps. The AMD AMF encoder scored close to Nvidia, while Intel's Quick Sync encoder (the Alder Lake variant) scored higher with 76 and 87 points at the same bitrates. This makes Intel's new AV1 encoder 16 percent "better" than both the Nvidia NVENC and AMD AMF encoders.

The only H.264 encoder that came close to the AV1's results was OBS's x264 VerySlow H.264 software encoder preset, which scored 78 points at 3.5 Mbps and 88 points at 6 Mbps. These results are impressive for the AV1 codec since OBS's VerySlow preset is so resource-intensive it can't be run in real-time (and is therefore useless for live streaming).

For more details on the AV1 tests — including software accelerated AV1 testing — check out EposVox's video.

3.5 Mbps is the Sweet Spot for AV1 Live Streaming

The 3.5 Mbps results are where we see the most gains out of Intel's AV1 encoder. This encoder is so good at 3.5 Mbps that it was able to outperform both Nvidia's and AMD's encoders at 6 Mbps with nearly half the bitrate.

As you get into bitrates higher than 6 Mbps, however, there seem to be diminishing returns. This is especially true of Intel's AV1 encoder which only gains 2.2% better image quality at 8 Mbps (the Twitch limit) over 6 Mbps.

As a result, we can expect bitrates to drop substantially when streaming platforms make the switch to AV1. This will be great for viewers, as the new encoder will require significantly less internet bandwidth to watch videos at the same quality as H.264, and will put less strain on the internet as a whole.

These new AV1 codec results are very impressive overall. Thanks to Intel, AV1 has completely decimated H.264 encoding with superior image quality that requires significantly less bandwidth.

So far we've only seen Intel's implementation of AV1 encoding. We should get a better understanding of AV1's capabilities when Nvidia and AMD implement their own AV1 encoding engines in their next-gen GPUs. If H.264 hardware encoders have taught us anything, it's that encoding efficiency can change drastically between brands and versions.

Nvidia and AMD's AV1 encoder implementations could be even more efficient and higher quality than what Intel is providing right now — but we'll have to wait until the RTX 40 series and RX 7000 series launch to see if that's true.

ByteDance, the owner of the TikTok social media app, has begun to hire specialists to develop its datacenter system-on-chips (SoCs). ByteDance will join multiple high-tech Chinese companies that recently rolled out their own data center SoCs designed to work on specific workloads.

So far, ByteDance has posted 31 chip-related openings on its website, reports TMTPost. Thirty-one chip engineers will barely be able to build an advanced processor, but what they could do is create a set of design requirements for contract chip developers, such as Alchip. Then, once a contract chip developer creates an appropriate design, a contract maker of semiconductors (like TSMC) will mass produce it for ByteDance.

For now, ByteDance only says it plans to develop its chips without elaborating. Typically, chip design cycles take years, so we will only learn about the company's SoCs a few years later.

What exactly ByteDance plans to develop is a big question. CNBC cites a spokesperson from the company that the data center grade system-on-chip from the social media giant will support various business areas, including video platforms, information, and entertainment apps. Meanwhile, several China-based sources (such as) indicate that ByteDance is looking forward to designing artificial intelligence chips.

ByteDance's key business is its TikTok media app, so it would be logical for the company to develop an SoC akin to Google's Argos video (trans)coding (VCU) to transcode video streams uploaded by TikTok users. Meanwhile, to further reduce power consumption, traffic, and storage capacities used by TikTok datacenters, ByteDance could introduce all-new codecs or AI algorithms. Furthermore, adding AI enhancements could improve image quality.

ByteDance will be yet another China-based cloud giant that develops its chips. For example, Alibaba has its own AI inference chip and general-purpose processor. Meanwhile, Baidu has its own Kunlun AI processor. Tencent also has its processors for various cloud workloads.

After almost a decade of quality issues, AMD's AMF encoder has finally received a massive upgrade in image quality thanks to the introduction of B-Frames in AMF Release 1.4.24. The update actually dropped months ago, but it wasn't until yesterday that Chris Griffith from Code Calamity finally put the new update to the test.

According to Griffith, he saw massive gains from the update. The AMF encoder was able to close in on Nvidia's newest NVENC encoder found in its RTX 20 series and 30 series GPUs. This update could see AMD Radeon GPUs as much more viable solution for game streaming, with low bitrate images looking almost as good as Nvidia's counterpart.

Major issues surrounding AMD's encoders have gone back as far as the 400 series Polaris GPUs, and in some cases even further. Support and quality have been constant issues and, for the most part, were always inferior to other encoders such as Nvidia's NVENC or Intel's QuickSync encoder. 

But the quality issue was compounded by the introduction of Nvidia's 6th generation NVENC encoder, introduced into the RTX 20 series, which elevated GPU encoding significantly, putting performance on par with good x264 encoding.

Despite 6th generation NVENC coming out over four years ago, AMD never really made a comeback, with its encoders remaining largely the same, even on the latest Radeon RX 6000 series GPUs. Streaming expert EposVox's video illustrates how terrible AMF's streaming performance is in low-bitrate scenarios running on the H.264 codec.

The only exception to this is HVEC encoding, where AMD's AMF encoder performs exceptionally well. But this capability will almost never be used in the real world, with all modern video players today requiring H.264 support.

B-Frames Are Back

The decisive feature improving AMF's image qulaity is the re-introduction of B-frames into the AMD encoder. Ironically, AMD originally had B-Frame support with its original encoder and decoder engine VCE (starting with the 2nd gen implementation). But AMD seems to have dropped support for it when the company released its new VCN engine, which was first released with AMD Raven Ridge APUs and RDNA1 graphics cards.

B-Frames are a very important aspect of video compression in the modern era. These frames are paired with I and P frames to build a highly compressed video file. B frames allow the H.264 compression algorithm to predict image data from both past frames and future frames in a video stream. It is an optional feature, but one that has proven to significantly increase the image quality of streams at lower bit rates.

Benchmarks

Code Calamity used VMAF to measure the image quality differentiation between AMF, NVENC and Intel QuickSync, with Big Buck Bunny as the reference video. In this benchmark, the highest possible score 100 is points. For reference, NVENC scores 96.13 points, and Intel QuickSync scored 96.37 in this test. According to Code Calamity, AMD's AMF encoder is just half a point behind both those encoders, where as before (without B-Frames), AMD's AMF  was a full two points behind.

This benchmark can be difficult to understand. But basically, it means AMF's image quality is right underneath Nvidia's latest NVENC encoder right now, which is much much better than where it used to sit before the latest update.

Support Is Still Largely Absent

As of now, the only problem with this new AMF encoder update is that no streaming platform has offered support, including OBS, despite the update being out for nearly four months. It's unclear why no one has implemented support for the update yet, but AMD does have a history of not broadly supporting developers in implementing its encoder SDKs. This might be the reason why support is taking so long to implement.

So for now, we have to take this benchmark as a future look into AMF's performance. Hopefully, it won't take long for apps like OBS to take advantage of the new update, so that current and future owners of AMD cards can make better use of them for streaming and other tasks.

Several months ago, we looked at the Z690 Aorus Master and found the upper-midrange board to have robust power delivery and many M.2 sockets for the latest storage. Today, we’ll look at the B660 Aorus Master that lives a bit down the product stack, on a chipset that eliminates CPU overclocking. Priced at a between $190 and $210 depending on where you find it, the B660 version of the Aorus Master also comes well-equipped and is a solid choice from a features perspective.

The B660 Master comes with almost everything you’d expect from a B660 board. There are three M.2 sockets, capable 16-phase 60A power delivery, a last-gen flagship audio codec, along with integrated Intel Wi-Fi 6 and 2.5 GbE. Gigabyte uses a black and gray heatsink combination here, which some may not find particularly appealing. That said, this is still a good-looking board and the heatsinks and shrouds covering most of the board, delivering a premium look.

Tested performance with our B660 Aorus Master was average among its DDR4 peers. It excelled in the Procyon suite but was a bit slower than others in the video and photo editing portion. Gaming results were on par with the rest, while power consumption was better than most overall. As usual, you’d be hard-pressed to notice any performance deficits without seeing them in a benchmark result.

Below, we’ll take a closer look at the Aorus Master’s features, software, and performance to see how this sub-$210 board stacks up against the rest and if it finds a spot on our best motherboards list. First, here’s a detailed list of the B660 Aorus Master’s specs, direct from Gigabyte.

Specifications: Gigabyte B660 Aorus Master

Inside the Box of the Gigabyte B660 Aorus Master

Inside the box, along with the motherboard, Gigabyte includes a barebones set of accessories including SATA cables, a Wi-Fi antenna, etc. As seems to be a growing trend, this board didn’t come with a driver disk/USB stick, but you can find the latest online at the website if your retail version is the same. Below is a complete list of the included extras.

• (2) SATA cables
• (3) M.2 screws
• Wi-Fi Antenna
• Microphone
• G-connector
• Stickers/case badge
• Manual

Design of the Aorus Master

Overall, the B660 Aorus Master looks like many of its more expensive (and overclockable) Z690 siblings. The matte black 6-layer PCB gives way to black and gray heatsinks to cover all the parts that get hot. The large black VRM heatsink dominates the left side of the motherboard and houses Aorus branding, backlit by RGB LEDs. The second RGB area is under the board in the top-right corner. The top VRM heatsink, along with the M.2 and chipset heatsinks, are gray with grooves cutout for more surface area and a design element. The chipset heatsink sports the Aorus falcon as well.

Overall we like the appearance. While there are other boards out there that cover more of the PCB, we have to remember this is a budget-focused board. Still, it will look good lit up inside your case, assuming you don’t mind the contrasting color of the heatsinks.

The top left corner of the B660 Master is taken up by the huge VRM heatsink that reaches out over the IO area, yielding a clean appearance. On top is the Aorus branding lit up from below with RGB LEDs. There are grooves cut out for additional surface area. Above the heatsink are two EPS connectors, an 8-pin (required) and 4-pin, to power the processor. To the right is the second VRM heatsink, but this time it’s gray and contrasts with the black PCB and other VRM heatsink.

Continuing right, we run into four DRAM slots that lock the memory down from both ends. Gigabyte supports up to 128GB of RAM with speeds listed to DDR4 5333(OC). Since you can’t overclock a CPU with this chipset, your only overclocking adventures have to come from the memory.

There are two 4-pin fan headers (CPU_FAN and CPU_OPT) and two RGB headers along the top-right edge. In total, there are eight fan headers scattered around the board, which should be plenty to control the cooling in your chassis. Each header supports both PWM- and DC-controlled devices. You’ll have plenty of power for all of your fans and a water cooling pump too, as each supports up to 2A/24W. Next to the fan headers are 3-pin ARGB and 4-pin RGB headers, with the other two light headers along the bottom edge of the board. Gigabyte’s RGB Fusion application controls the board’s RGB ecosystem.

Working our way down the right edge, we run into the 24-pin ATX connector to power the board, two additional fan headers (SYS_FAN5/6_PUMP), a front-panel USB 3.2 Gen 1 (5 Gbps) port, and slightly offset is the front panel USB 3.2 Gen2 (10 Gbps) connector. Just below the screw hole here are the Q-LEDs that light up during the boot process. If there is a problem during startup, one of the four lights (labeled Boot/CPU/VGA/RAM) remains lit pointing out the general area of the problem. Since there isn’t a 2-character debug LED to share more details, you’ll have to rely on this LEDs for any pre-OS hangups during boot.

Gigabyte uses an 18-phase VRM here, with 16 phases dedicated to Vcore. Power comes in from the EPS connector(s) and then to an ON Semiconductor NCP81530 controller, which works in a teamed/parallel fashion (no phase doubler, but two MOSFETs receive one signal) with 16 55A ON Semiconductor NCP302155 MOSFETs. Since overclocking the processor isn’t an option with this chipset, the 880A for Vcore is plenty to run our Intel Core i9-12900K without issue.

The bottom half of the board sports the audio section, storage, and PCIe slots. Starting on the left side, we see Gigagbyte’s AMP UP! Audio branding covering a Realtek ALC1220-VB codec hidden below. While the B660 Aorus Master may not have a fancy DAC or opamp, it does sport several yellow Chemicon brand audio caps, along with four red WIMA caps to help deliver quality sound. This isn’t the newest audio codec, but being the last generation’s flagship, most users will be pleased with the audio.

The Aorus Master sports two full-length slots and one x1 size slot for PCIe connectivity. The top full-length slot uses reinforcement to prevent shearing from heavy graphics cards and runs at PCIe 4.0 x16 from the CPU. The second full-length slot sources its lanes from the chipset and runs at a maximum of PCIe 3.0 x4. The board supports AMD Quad-GPU Crossfire and 2-Way Crossfire with the available lanes. The bottom slot also sources its bandwidth from the chipset and runs at a maximum of PCIe 3.0 x1.

Hidden under the gray heatsinks are three M.2 sockets. All sockets support up to 110mm, but PCIe modules only. SATA-based M.2 modules are not supported. I would like to see that availability for a budget board, as SATA SSDs are cheaper ( although slower) than NVMe drives. And those on a budget might have more older storage they’d like to carry over to a new system. The top socket, M2A_CPU runs at PCIe 4.0 x4, along with the second socket (M2P_SB). The bottom socket, M2M_SB runs at a maximum of PCIe 3.0 x4. It’s worth noting that when a drive populates the M2M_SB socket, the second PCIe slot gets disabled.

Along the right edge, we run into two Thunderbolt AIC headers (for adding an add-in card) and four SATA ports. The SATA ports remain active regardless of what M.2 devices are used. If you want to RAID the SATA ports, they support RAID0/1/5/10. Four SATA ports will be plenty for most people. But again, if you have multiple older drives you want to carry over from a previous PC, this could be limiting.

Several headers live across the bottom, including USB ports, RGB, and fan headers. Below is a complete list of all the headers in this area.

• Front panel audio
• 3-pin ARGB header
• 4-pin RGB header
• TPM header
• (2) USB 2.0 headers
• Q-Flash+ button
• (4) System fan headers
• 2-pin temp header
• Clear CMOS jumper
• Front panel header

As we move to the rear IO area, Gigabyte was kind enough to preinstall the IO plate from the factory, which gives the B660 Master a premium vibe. All the ports are labeled and easy to read against the black background. There’s nothing fancy back here, just some branding to remind you this is an Aorus product.

One thing that stands out are the 10 total USB ports which is generous for a board in this price class. We run into a stack of four USB 2.0 ports on the left, while next to that is the Wi-Fi antenna connection. Video outputs consist of the DisplayPort (v1.2) on top, with HDMI(v2.1) on the bottom. Next is a USB 3.1 Gen 1 (5 Gbps) port that sits on top of the fast USB 3.2 Gen 2x2 (20 Gbps) Type-C port. Continuing right, we hit four more USB ports (3.2 Gen 2, 10 Gbps), with the Intel 2.5 GbE port on top. On the right edge is the audio stack that consists of five analog plugs and a SPDIF out.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Firmware

Gigabyte’s B660 firmware looks no different than on pricier Z690 boards, except it excludes processor overclocking functionality. The BIOS starts in an informational EZ Mode that displays system information with limited functionality. You can enable XMP profiles from here, access Smart Fan 6 for fan control, use Q-Flash to update the BIOS, or enter Advanced Mode. When working in the Advanced portion of the BIOS, major headers are listed across the top, with sub-headings below. Page up/down functionality has finally been added, and the BIOS is easy to read and well organized to help find what you’re looking for.

Software

On the software side of things, Gigabyte’s primary tool is the App Center. This application is a central repository for all board-centric applications, some Windows settings, and other third-party software. Simply click to download the applications you want, install them, and an icon shows up on the screen. We installed @BIOS (BIOS flashing utility), Easy Tune (overclocking/system tweaking), RGB Fusion 2.0 (to control RGB lighting) and last but not least, SIV (for system monitoring). The Gigabyte website has many other helpful applications, including USB charging (to control power to ports), LAN, and more that aren’t covered here. Overall, we like App Center’s small footprint and found its modular tools helpful.

Test System / Comparison Products

As of October 2021, we’ve updated our test system to Windows 11 64-bit OS with all updates applied. We kept the same Asus TUF RTX 3070 video card from our previous testing platforms but updated the driver to version 496.13. Additionally, our game selection was updated, as noted in the table below. We use the latest non-beta motherboard BIOS available to the public unless otherwise noted. The hardware we used is as follows:

Test System Components

EVGA supplied our Supernova 850W P6 power supply (appropriately sized and more efficient than the outgoing 1.2KW monster we used) for our test systems.

Benchmark Settings

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

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.

To get the most out of the Intel Alder Lake chips, you need to be on Windows 11 with its updated scheduler. In most cases, Windows 10 performs well. However, some tests (Cinebench R20, Corona and POVRay) take a significant hit. In short, if you’re going with Alder Lake, you must upgrade to Windows 11 for the best results across the board. That may change with patching and updates in the future, though.

Synthetic Benchmarks

Synthetics provide a great way to determine how a board runs, as identical settings should produce similar performance results. Turbo boost wattage and advanced memory timings are places where motherboard makers can still optimize for either stability or performance, though, and those settings can impact some testing.

The Aorus B660 Master did well here, with scores in 7Zip, Cinebench R23, POV-Ray, and Procyon Office all around the average for DDR4 boards. The Video and Photo editing tests inside Procyon were also average. AIDA64 memory latency results were also as expected. In short, there’s nothing to worry about performance-wise in this set of tests.

Timed Applications

Starting with LAME testing, the B660 Master took 9.51 seconds to complete, the second quickest result so far. The Corona Ray Tracing benchmark's results were just above average, finishing in 53 seconds. Handbrake results were average for DDR4, with the x264 test completed in 118 seconds, while x265 (which isn’t affected much by the slower DDR4) compared in 301 seconds, or slightly above average. There are no performance issues in the timed applications either.

3D Games and 3DMark

Starting with the launch of the Z690 chipset, we’ve updated our game tests. We’ve updated Far Cry: New Dawn to Far Cry 6 and shifted from F1 2020 to F1 2021. We run the games at 1920x1080 resolution using the Ultra preset (details listed above). As the resolution goes up, the CPU tends to have less impact. The goal with these settings is to determine if there are differences in performance at the most commonly used resolution with settings most people use or at least strive for. We expect the difference between boards in these tests to be minor, with most falling within the margin of error differences. We’ve also added a minimum FPS value, as that can affect your gameplay and immersion experience.

In F1 2021, the B660 Aorus Master DDR4 averaged 165 fps, with minimums of 140 fps, one of the faster results. Far Cry 6 was similar, with a minimum of 125 fps and an average of 137 fps. Both results here are slightly above average.

In the 3DMark tests, the B660 Master DDR4 scored 14,173 on 3DMark Time Spy and 16,912 on Fire Strike Extreme, both slightly below average. Although the synthetic game benchmarks run a bit slower than average, our real game tests showed good results. You’ll have nothing to worry about when gaming on this motherboard.

Power Consumption / VRM Temperatures

We used AIDA64’s System Stability Test with Stress CPU, FPU and Cache enabled for power testing, using the peak power consumption value. The wattage reading is from the wall via a Kill-A-Watt meter to capture the entire PC, minus the monitor. The only variable that changes is the motherboard; all other parts are the same.

At idle, the B660 Master used 43W, which is one of the most efficient results we’ve seen. Load power use peaked at 342W, which is a bit higher than the average. Although the peak load wattage here was higher than many others, it will be difficult to see this on your power bill compared to other boards.

VRM temperatures on the B660 Master topped out at 54 degrees Celsius through the Extech (65 degrees Celsius through the internal VRM MOS sensor) during stock testing with DDR4-4000 RAM. Between the 55A VRMs, heavy heatsink, and the fact that you can’t overclock the CPU on this platform, the power bits and their cooling work well.

Overclocking

Overclocking the CPU isn’t possible on B660-based chipsets, but the platform does allow for memory speed adjustment. With our DDR4-3600 and DDR4-4000 kits, we simply set XMP and off we went without a hitch. Surely there’s some headroom left, but I’d imagine people purchasing a budget-class board aren’t looking to buy expensive RAM that’s far outside the sweet spot (DDR4 3600-4000).

Bottom Line

Priced at $209.99 (though it’s often found for sale at about $190), Gigabyte’s B660 Aorus Master is an affordable entry to Intel’s Alder Lake platform that’s well equipped for the price. The board includes capable power delivery, 10 USB ports (including 20 Gbps Type-C) on the rear IO, three M.2 sockets, a solid audio section, and generally good looks. Performance was right around average for DDR4 boards, so there’s nothing to worry about on that front either. In all, it’s a competent motherboard at a reasonable price.

As far as complaints or improvements go, there aren’t many. On the storage side of things, the three M.2 sockets should be enough for most, but none support SATA-based modules. Another potential shortcoming is the four available SATA ports, where many options in this class have six. Outside of that, the contrasting heatsinks could be a turnoff for some users, though the board still looks good overall.

There’s some stiff competition in this space, though. Asus has a ROG Strix B660-A for $229.99, MSI has a MAG B660 Tomahawk WIFI DDR4 ($189.99), and ASRock’s B660 Steel Legend is the least expensive of the group at $159.99. If you need more native SATA ports, the ASRock and MSI have you covered with six. If you’re looking for the best audio, the Asus is the only one that rocks the latest Realtek ALC4080 codec (although it’s priced significantly higher). All of the boards include a 2.5 GbE and Wi-Fi (Wi-Fi 6), except the ASRock which does not come with Wi-Fi. They all support up to three M.2 sockets. In short, it’s going to come down to price and looks for most users.

In the end, the Gigabyte Aorus Master is a competent board in the wallet-friendly B660 space. If you can get the board at the sale price, it costs the same as the MSI. Between them, I like the Gigabyte over the Tomahawk if only for how the heatsinks cover more of the board, giving it a more premium and clean appearance. But if you’re not a fan of the contrasting colors, the MSI could be a more attractive solution at the same price. If B660 is your platform of choice and you want a solid motherboard that checks off all the expected boxes, the Gigabyte B660 Aorus Master DDR4 is a great option for around $200.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Update: As it turns out, AMD's RDNA 3-based GPUs with VCN 4.x engines support hardware AV1 encoding, according to Linux driver patches released by AMD and discovered by @Kepler_L2. Previously, we reported based on Linux patches that AMD's upcoming Radeon RX 7000-series GPUs did not support AV1 encoding, which is not correct. We've modified the text to reflect the new information.

As part of the ongoing enablement of AMD’s next generation RDNA 3 GPUs, the company has released a series of patches for Linux that reveal details about the company’s upcoming video engines, the Video Codec Next version 4.x. Based on the information we have, VCN 4.x supports decoding of virtually all modern codecs.

AMD's VCN 4.0 engine appears to support H.264/MPEG 4 AVC, H.265, VP9, AV1, and JPEG decoding, as well as AV1, H.264, and H.265 encoding. For now, VCN does not appear to support H.266/VVC (versatile video coding) decoding/encoding. VVC is a next-generation codec that will be required maybe in 2023 or 2024 (when appropriate content becomes available), while AV1 is a current generation codec with expanding usage. There are users and companies that would like to have it now for encoding or transcoding video, but later this year should also suffice.

Intel's Arc Alchemist GPUs fully support AV1 decoding and encoding, the only chips to do so at present. We expect Nvidia may also support AV1 encoding with it's upcoming Ada architecture, though that remains as yet unconfirmed. AMD's RDNA 3 also won't arrive until later this year, but if AMD's VCN 4.x indeed does support AV1 encoding, all three major GPU companies will then offer hardware accelerated AV1 encoding and decoding.

AMD's current-generation GPUs based on the RDNA 2 architecture use the company's VCN 3.0, VCN 3.1, and VCN 3.1.2 video decoding blocks. By contrast, next-generation RDNA 3 graphics processors (at least the so-called SoC21, which is believed to be Navi 31) will feature the next-generation VCN 4.0 engine, according to a new Linux patch posted by AMD and discovered by @Kepler_L2.

Speaking of VCN 4.0, even the current VCN 3.x engine fully supports H.264/MPEG4 AVC, H.265, VP9, AV1, and JPEG decoding as well as H.264 and H.265 encoding. Unless there are more innovations, the key improvement with VCN 4.0 over 3.x will be hardware accelerated AV1 encode/decode support. It may also support higher resolutions, color depths/formats, and/or efficiency improvements, and we anticipate some additional feature set changes as well.

Keep in mind that not all GPU features get enabled ahead of launch, especially in Linux due to various legal and technical reasons. We'll find out for sure what VCN 4.x and RDNA 3 add to the feature list likely late this year, when we expect the RX 7000-series GPUs to launch.

Being one of the Alliance for Open Media founding members, Intel has done a lot to promote the AV1 codec and make it more accessible to content makers/providers and end-users. Intel was the first to offer hardware AV1 decoding with its Xe-LP GPUs in 2020. This week it released version 1.0 of its speedy open-source Scalable Video Technology AV1 encoder and decoder for CPUs. SVT-AV1 works with all modern processors. 

The AV1 open-source video codec was designed for ultra-high-def resolutions, wide color gamut, and high-dynamic range enhancements. AOMedia said in 2018 that its AV1 was 30% more efficient than existing codecs (implying primarily on H.265/HEVC that is designed for similar 4K+ content), which is a big deal. But one problem with highly-efficient codecs is that they are extremely hungry for resources and generally require hardware acceleration to work correctly. Meanwhile, modern CPUs have a lot of resources and new instructions that can be applied to decoding and encoding, which is precisely what SVT-AV1 does. 

SVT-AV1 is a scalable standard-agnostic encoder/decoder library that can take advantage of the multi-core nature of modern CPUs and AVX2 instructions. The SVT-AV1 also adds further AVX2 optimizations to boost performance, image quality improvements, fast-decode for more preset levels, and S-frames support, reports Phoronix. 

Intel's SVT-AV1 libraries are supported on modern x86 machines (Intel 5th Generation Core 'Broadwell' and higher) running Apple's macOS, Microsoft's Windows, and Linux. 

Intel and Netflix initially started the SVT-AV1 project to develop a production-quality AV1-encoder with performance levels applicable to various applications, from premium video-on-demand to real-time and live encoding/transcoding. In August 2020, the SVT-AV1 encoded/decode library was adopted by AOMedia's Software Implementation Working Group (SIWG) to make AV1 more popular. The SVT-AV1 version 1.0 release marks a milestone in the development of the encoder/decoder libraries. 

Version 1.0 of SVT-AV1 encoder/decoder libraries is a milestone and good news for content creators and end-users. However, for companies like Netflix, Intel now offers Arctic Sound-M accelerators based on DG2 silicon that can handle eight simultaneous 4K streams and supports hardware-accelerated AV1 encoding and decoding.

The single-tile Intel Arctic Sound 1T features an Xe-HP GPU with 384 EUs and 16GB of HBM2E memory, offering a peak bandwidth of up to 716 GB/s (which probably means that we are dealing with two stacks of HBM2E that use a 2048-bit interface). The accelerator is a short single-slot full-height card rated for a 150W TDP.  

Intel's Arctic Sound 2T card carries an Xe-HP GPU with two tiles, 960 EUs (480×2 to be more accurate), and 32GB of HBM2E DRAM. The accelerator uses a full-length, full-height (FLFH) form factor and is rated for a 300W TDP delivered using one eight-pin power connector. (One thing to keep in mind is that IgorsLab edited the images of the cards to protect the source.)

Intel's Xe-HP architecture is a far cry from the company's Xe-LP architecture we know from the Iris Xe consumer-grade GPUs. The Xe-HP card supports more floating-point formats (e.g., FP16, FP32, FP64 for general purpose, bfloat16 format for AI/ML computing), more compute-specific instructions, DP4A convolution instruction for deep learning, and Intel's XMX extensions.

The datacenter-oriented Xe-HP GPUs use all-new execution units (EUs) with various IPC improvements, feature HBM2E memory support, and are made using Intel's performance-optimized 10nm SuperFin process technology. In short, the Xe-HP is not the Xe-LP or Xe-HPG on steroids, but something completely different. 

Intel now allowed some of its customers to preview its Arctic Sound compute cards carrying single-tile and dual-tile Xe-HP implementations. Intel announced a quad-tile Xe-HP implementation last year and even demonstrated one of such accelerators in action offering over 42 FP32 TFLOPS of performance. However, the company is not ready to sample it right now or is only sampling it with select customers.

Intel's Xe-HP plans are not completely clear as the company has never detailed them. Meanwhile, the EU count of these two cards is somewhat lower than expected (assuming that one Xe-HP tile features 512 EUs). At present, we don't know how old these cards are and which configurations Intel plans to ship.

Researchers have discovered that the XCSSET malware has started targeting M1-equipped Macs via Xcode, The Hacker News reported Monday and has been updated to compromise accounts on various cryptocurrency trading platforms.

Xcode is the integrated development environment (IDE) used to make apps for the iPhone, iPad, and other Apple hardware. Even if a cross-platform framework is used to develop a particular app, it must pass through Xcode to reach those platforms.

That means XCCSET is limiting itself to technically savvy people who, if we had to guess, would be more likely to own cryptocurrency than the average Mac owner. Targeted attacks like this are often more successful than broader ones.

Kaspersky warned that XCSSET had been updated for Apple's custom silicon in March. The malware wasn't focused on cryptocurrency at the time, the security company said. Instead, it featured a variety of modules that were designed to:

• Reading and dumping Safari cookies
• Injecting malicious JavaScript code into various websites
• Stealing user files and information from applications, such as Notes, WeChat, Skype, Telegram, etc.
• Encrypting user files

Trend Micro then warned on April 16 that XCSSET had been updated to bypass security features introduced with macOS Big Sur, change the icons it uses to match system icons, and attempt to gain access to victims' accounts on crypto platforms.

The company's advice was clear: "To protect systems from this type of threat, users should only download apps from official and legitimate marketplaces," it said. But that's hard to do when it comes to finding Xcode projects to work with or learn from.

XCSSET's expansion to cryptocurrency makes sense. The value of Bitcoin, Ethereum, and even Dogecoin has continued to rise in recent months, and stealing coins from someone else is probably requires fewer resources than mining them would.

Adding support for Apple's custom silicon was also prudent. Devices featuring the M1 chip have been well-reviewed, and with Apple's plan to ditch Intel entirely by the end of 2022, it makes sense to start targeting its chips now.

Other malware creators appear to agree. We saw reports of the first malware targeting Apple silicon in February, and in March, the Silver Sparrow malware was discovered on approximately 30,000 macOS devices, some of which had M1 chips.

The RISC-V CPU architecture is making progress across a plethora of devices. From soldering irons and watches to desktop computers, the open source RISC-V CPU looks set to make an impact in 2021. Seeed Studio and Beagleboard have today announced an official collaboration with StarFive, a leading RISC-V solutions provider. The focus of this announcement is the Beagle V, an affordable single board computer (SBC) which runs Linux and is powered by a dual core RISC-V.

Powering Beagle V is a dual core 64-bit U74 RISC-V CPU clocked at 1.5 GHz and up to 8GB of LPDDR4 RAM. From a CPU perspective Beagle V looks to be slightly faster than a Raspberry Pi 4, but looking at the StarFive JH7100 System on Chip (SoC) we can see a Tensilica-VP6 used in computer vision projects, an Nvidia Deep Learning Accelerator (NVDLA) used for artificial intelligence and finally a Neural Network Engine. All of these point towards a board that looks on paper to be quite the AI / machine learning powerhouse. Best of all Beagle V is a purely open source product, with both the hardware and the software offered under an open source licence.

At its heart, Beagle V is still an SBC and as such it has what we come to think as the "standard complement" of ports. With four USB 3.0 ports, Gigabit Ethernet, Micro SD card and a single HDMI port, capable of 1080P 30fps. A 40 pin GPIO suggests compatibility with Raspberry Pi HATs, and a casual glance at the pinout shows that the standard digital I/O, I2C and SPI seem to be mapped correctly. Dual camera (CSI) and a single display connector (DSI) suggest that Beagle V may be compatible with the official Raspberry Pi camera and display. Wireless connectivity is provided via 2.4 GHz WiFi and Bluetooth 4.2. A shame that 5 GHz WiFi and Bluetooth 5.0 are not present, but we can live without it. Power is provided via a USB-C port, and from the datasheet we can see that Beagle V requires 5V at 2A but a 3A supply would be a real benefit.

Starting from $119 for the 4GB model, or $149 for the 8GB, Beagle V is slightly more expensive than a Raspberry Pi 4 but for RISC-V and AI enthusiasts this is a small price to pay for a RISC-V machine of such power. With an anticipated early access release date of March, with larger availability from September 2021, Beagle V looks to be a worthy contender to the Nvidia Jetson for AI and machine learning projects. It may not be a direct competitor to the Raspberry Pi, but it will be interesting to put the two against one another to see both of their pros and cons.

The MSI X570 Godlike motherboard isn’t new -- in fact, we’ve used it  as our test platform in some AMD CPU reviews.  But we’ve never had the chance to review it, and given that the Ryzen 9 5950X is our new test CPU, we’ve decided to put this flagship board through its paces with AMD’s latest flagship silicon. 

The bold Godlike name infers this is one of the best out there, and to be frank it isn’t very far off, if price isn’t much of an issue. The $699.99 Godlike includes robust power delivery, a premium audio solution, five M.2 sockets and a cool-looking monochrome OLED screen, all of which are fitting for a flagship-class part.

At the time of publishing, MSI’s X570 lineup consists of nine boards, starting with the X570-A Pro ($159.99) and including multiple parts from the MAG line (Tomahawk WiFi), the MPG line (Gaming Plus, Gaming Edge WiFi, and Gaming Pro Carbon WiFi), and the premium MEG lineup (Unify, Ace, and Godlike). From E-ATX flagships like the Godlike to entry-level boards, there is something for everyone in MSI’s X570 range.

Focusing on performance, the Godlike did well in our benchmarking suite, running most tests on the faster side of average. Even in the PCMark 10 MS Office/Application testing, where we often see anomalies, we saw consistent results across the full gamut of testing. The board allowed the 5950X to reach its full 5 GHz boost, so we’re getting all we paid for at default settings. As expected, overclocking on this motherboard went without a hiccup, easily handling our 16-core/32-thread CPU. VRM temperatures at load were some of the lowest we’ve seen as well. 

Performance isn’t everything, however. The features list is long and the Godlike includes a 10 GbE add-in-card (AIC) in addition to the existing 1 and 2.5 GbE ports, dual Realtek ALC1220 codecs and two additional PCIe 4.0 x4 M.2 sockets via add-in-card. Features and specifications aren’t everything either, though. The appearance of the X570 Godlike is also solid. The Infinity RGB display around the VRMs, coupled with the customizable OLED screen and the overall clean look set this board apart from other less expensive models. Read on for additional details about the X570 Godlike, because there’s a lot here to cover.

Specifications - MSI X570 Godlike 

Inside the box along with the motherboard is a whole lot of accessories. Outside of the usual selection of SATA cables, driver disks, and guides, the Godlike includes multiple add-in-cards (10 GbE, M.2 Expander), thermistors, and more. This accessory stack is the biggest we’ve seen in a while. Below is a complete list of the included extras. 

• (6) SATA cables
• 1 to 2 RGB LED extension Y cable (80cm)
• Corsair RGB LED extension cable (50cm)
• Rainbow RGB LED extension cable (10cm)
• (2) Thermistor cables
• Wi-Fi Antenna
• M.2 Expander-Z Gen4 AIC
• 10G Super LAN AIC
• 6.3mm audio adapter
• Case badge
• SATA Cable labels
• Driver DVD
• Manual, Quick Guide, Quick Installation Guide

When you first take the Godlike out of the box, you’re greeted by an attractive board that will fit in with most build themes. The PCB is matte-black, as are all heatsinks and shrouds. The large brushed-aluminum finish VRM heatsinks connect via heatpipe to each other and the chipset heatsink, sharing the thermal load between them. The DRAM and PCIe slots are all reinforced using MSI’s PCIe Steel Armor, protecting against bending/shearing and EMI. 

On the RGB front, the Mystic Light Infinity II feature above the VRM area steals the show. Through the magic of light and mirrors, the RGB lighting looks like it goes down to infinity, offering a truly unique look. The chipset heatsink also sports RGB lighting, with the familiar MSI Dragon illuminated just above the chipset fan. Additionally, the customizable OLED panel located to the DRAM slots’ right brings another unique aesthetic to a motherboard. MSI’s Mystic Light software controls all of the LEDs and the OLED screen. In short, the X570 Godlike looks the part of a high-end motherboard and includes some cool features to set it apart from other high-end alternatives.

Starting with the board’s top half, we get a closer look at the Infinity RGB display above the heatsink on the left-hand side. The mirror reflection system does a great job of showing off the RGBs differently from most motherboards. Above this and the large VRM heatsink are two 8-pin EPS connectors (one required) that send power to the CPU. The socket area is clear of most hazards and begs to have a pot on top for some sub-ambient fun.

To the right of the socket are four reinforced DRAM slots. Capacity is listed up to 128GB, while the supported speeds are close to the fastest we’ve seen listed, at DDR4 5000+(OC). As always, your mileage may vary on reaching sky-high speeds such as that. AMD’s sweet spot, even with the 5000 series processors, still sits at DDR4 3600/3733 with the Infinity Fabric tied to it (1:1). Reserved for benchmarking, these speeds benefit those shooting for world records more than it does for real-world performance gains.

Above the DRAM slots are the first three (of nine) 4-pin fan headers. Each fan header supports both PWM and DC controlled devices. The board will automatically detect what is connected and adjust the mode accordingly. MSI doesn’t list how much power these will output in the manual, but we reached out to the company and have answers. All 4-pin headers are 1A/12W, except for the pump header at 3A/36W. If only by the sheer count of fan headers, there should be plenty of power available to connect your pump and fans to the motherboard. Each header even has a LED next to it that tells you what mode the board is using. You’ll see red for PWM fans and white in DC mode. I don’t recall seeing this before, but this nifty little add-on can be helpful for troubleshooting and gives users information on how the fans connect at a glance, without going into the BIOS.

To the right of the socket is MSI’s OLED-based Dynamic Dashboard. Used for a slew of things, the Dynamic Dashboard displays CPU temperature/speed, BIOS flash status and error messages. The monochrome dashboard is customizable, even allowing you to upload a .GIF animation file to display. If there is an error during POST, the dashboard displays the part causing the issue listing the location and error on the screen. Couple that with the two-character debug LED and the four ‘dummy’ LEDs onboard and you should know what’s going on with this motherboard at all times. While it’s arguably overkill for the average user, for those benchmarking and pushing overclocking limits, this can be very helpful to troubleshoot issues efficiently.

In between these headers on the top of the board is the first (of four + Corsair RGB) RGB headers. In this case, it’s a 4-pin RGB. You’ll find the other three headers on the bottom edge of the board. The Mystic Light RGB application controls the RGB lighting. We’ve had no issues with this software providing proper control in our testing.

To the right of the DRAM slots and Dynamic Dashboard are a few more headers and connectors. From the top-down is the first of two temperature sensors, SYS_FAN2 header, the Corsair header, 24-pin ATX power connector, SYS_FAN3 header and a USB 3.2 Gen2 Type-C front panel connector.

The MSI X570 Godlike runs a total of 14 phases for Vcore and a 4-phase configuration on the SoC. Managing power is a premium International Rectifier IR35201 (x+y=8) buck controller that sends power to seven IR3599 doublers, then on to 14 70A Infineon TDA21472 MOSFETs. The 980A available for the CPU/Vcore only is one of the highest we’ve come across so far. We had no issues whatsoever overclocking our flagship Ryzen 9 5950X CPU. The VRMs remained cool throughout all of our testing. If sub-ambient cooling and benchmarking are your thing, the VRMs will serve you well. Of course, ambient overclocking won’t be an issue either.

Moving to the bottom half of the board, we’ll start on the left side with the audio section. Hidden underneath the shroud are two (yes, two) Realtek ALC1220 codecs. The first is for the rear panel audio connectors, with the second dedicated for front-panel audio. An ESS E9018 Sabre DAC drives the 6.3mm stereo headphone output on the back panel. If you don’t have a 6.3mm headset, a 3.5mm-to-6.3 adapter is included if you’d like to use this port. Rounding out the audio section are 12 yellow Nippon audio caps and four WIMA audio caps (red). Overall, the audio solution is well above average, using premium codecs and a quality integrated DAC. The 6.3mm jack on the back is truly something unique on a motherboard.

In the middle of the board are four reinforced PCIe slots. Intended for GPUs, the top three slots support PCIe 4.0 x16 and break down to x16/x0/x0 or x8/x0/x8, or x8/x4/x4 when all three slots are populated. The fourth/bottom slot is sourced from the chipset and provides PCIe 4.0 x4 lanes. If you plan on using the M.2 Xpander and 10 GbE AICs, that leaves you with two full-size PCIe slots for video cards.

Between the PCIe slots are three M.2 sockets, each with their own heatsinks. The top two (M2_1 and M2_2) support both PCIe 4.0 x4 and SATA-based modules. M2_1 supports up to 110mm modules while M2_2 supports up to 80mm devices. The bottom slot, M2_3, supports PCIe only and drives up to 110mm long. If these three aren't enough, the M.2 expander card adds two more M.2 sockets. To get the full bandwidth from the AIC, install it in a slot that offers PCIe 4.0 x8 bandwidth (third slot down with nothing installed in the second slot). If you need a whole lot of wicked fast NVMe based storage, this motherboard supports it!

To the right of the M.2 and PCIe area are the chipset heatsink and fan. The large fan was inaudible during our testing, though we weren’t using any chipset-connected PCIe 4.0 devices so the stress level is lower. On the right edge are two USB 3.2 Gen 1 front-panel ports along with six SATA ports. SATA ports 1-4 are fed from the chipset, while ports 5/6 source their lanes from the ASMedia ASM1061 chip. The AMD chipset connected ports support RAID0, 1 and 10 modes, which is typical.

Across the bottom are several buttons and headers ranging from front panel audio to power and reset buttons. A lot is going on down there, especially for hardcore overclockers, where you’ll find BCLK buttons and other extreme overclocking functions. Also worth noting in this section are the temperature sensors and water flow headers. With these on board, you’re able to monitor and then control attached devices through the MSI software and the BIOS. Here’s the full list, from left to right:

• Front panel audio
• TPM header
• 3-pin RGB header
• LN2 mode header
• Fan header
• Temperature sensor header
• Fan header
• BIOS switch
• Fan header
• BCLK “-” button
• USB 2.0 header
• BCLK “+” button
• Reset and Power buttons
• OC knob
• USB 2.0 header
• Front panel header
• Water flow header
• Clear CMOS header
• Slow mode jumper

MSI’s rear IO includes an integrated plate with a black and red theme. On the left are two buttons, one that clears the CMOS and the other used for the board’s BIOS flashback capabilities. Moving right, we spy the two WiFi antenna mounts, two USB 3.1 Gen 1 ports and above that, a legacy PS/2 mouse/keyboard port. Next are the two Killer-based Ethernet ports that reside above the four USB 3.2 Gen 2 ports (one Type-C). To the right of that is one of the most unique features on this motherboard, a 6.3mm audio jack that uses the integrated DAC. Finally, the rest of the audio stack consists of a 5-plug analog and SPDIF stack.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Software

MSI doesn’t offer an excessive amount of software, but it covers all the bases. In the utility section of MSI’s support section for this board is an MSI-skinned version of CPUz, Nahimic Audio control and Dragon Center. Dragon Center (DC) is a one-stop-shop to monitor the system, control RGB lighting (Mystic Light), LAN, and gaming highlights. DC is relatively comprehensive, though it is missing software-based overclocking and fan control.

Dragon Center

Nahimic Audio

Firmware

To give you a taste of the Firmware, we’ve gathered screenshots showing most BIOS screens. MSI’s layout is informative and straightforward to use in EZ Mode or Advanced, with the latter separated into the more familiar sections on the main page. When overclocking, the most frequently used options are located on the opening BIOS screen in the OC section, so you don’t have to dig down for the most common functions. Overall, the MSI BIOS is easy to navigate and read with plenty of options to tweak your system.

Test System

We’ve updated our test system to Windows 10 64-bit OS (20H2) with all threat mitigations applied. We also upgraded our video card driver to 457.30 and reran all the game tests moving to the new 5000 series CPU. We use the latest non-beta motherboard BIOS available to the public unless otherwise noted. The hardware used is as follows:

Benchmark Settings

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Our standard benchmarks and power tests are performed using the CPU’s stock frequencies (including stock Thermal Velocity Boost), with all power-saving features enabled. We set optimized defaults in the BIOS and set the memory set using the XMP profiles. For this baseline testing, Windows is set to High Performance before we switch over to Balanced during power testing, so the PC idles appropriately.

Synthetic Benchmarks

Synthetics are a great tool to figure out if a board runs out of spec, as identical settings should produce similar performance results. Advanced memory timings are the one place where motherboard makers can still optimize for either stability or performance, though, and those settings can impact some testing.

In our synthetic benchmarks, the X570 Godlike was one of the better performing boards overall. Even in PCMark 10, where the results can vary, it performed consistently well throughout. The results from 7Zip and Cinebench were also some of the highest we’ve seen so far with the Ryzen 9 5950X.

Timed Applications

Results from our LAME testing shows the Godlike in the middle of the pack, while the Corona testing has it in the faster half. The x264 portion of Handbrake testing produced the quickest time, while in x265 it was among the fastest. This is a good showing in the multi-threaded timed applications as well.

3D Games and 3DMark

We’ve updated our game tests to The Division 2 and Forza Horizon 4. The games are run at 1920x1080 resolution using the Ultra preset. As the resolution goes up, the CPU tends to have less of an impact. The goal with these settings is to determine if there are differences in performance at the most commonly used resolution with settings most people use or at least strive for.

During our gaming tests, we saw appreciable gains when moving from the 3900X to the newer 5950X. Between the clock speeds, IPC and large cache differences, performance increased using this CPU at our 1080p test resolution. If you game at a higher resolution, chances are you won’t see those gains unless you’re using one of the new, ultra-fast Nvidia or AMD GPUs. Disclaimer aside, our X570 Godlike had no problems here, easily mixing in with the rest of the newly tested boards in both 3DMark and actual games.

Power Consumption / VRM Temperatures

For power testing, we used AIDA64’s System Stability Test with Stress CPU, FPU and Cache enabled, using the peak power consumption value. The wattage reading is from the wall via a Kill-A-Watt meter to capture the entire ecosystem. The only variable that changes is the motherboard; all other parts are the same.

The X570 Godlike used 78 Watts from the wall at idle, which is on the higher end of boards tested so far. Load wattage peaked 228W, a result that falls around the middle of others we’ve tested. If we average this out, the 153W value is just about in the center as well—nothing out of the ordinary here.

The 19-phase 70A VRMs handled our CPU testing without issue. At stock speeds, the hottest probe on our Exetech peaked a bit over 50 degrees Celsius with the internal sensor peaked at 56 degrees Celsius. After overclocking our Ryzen 9 5950X CPU to 4.4 GHz and ~1.3V, VRM temperatures peaked around 56 and 62 degrees Celsius for the Extech and internal sensors, respectively. Compared to the other boards tested, the temperature lands in the middle of the others tested so far. This VRM is plenty capable of handling the flagship CPU at stock, overclocked, and even sub-ambient/extreme overclocking.

Overclocking

There are several ways to overclock on AMD platforms, depending on your goals. If your focus is single-threaded performance, you may want to focus on using Precision Boost Overdrive (PBO) and adjusting its parameters. If you can use all cores and threads, setting a manual CPU multiplier and voltage is likely the better route. While the latter clips peak single-threaded performance a bit, it increases all core/thread performance over the all-core boost. To that end, we settled on 4.4 GHz at ~1.3V for an all core/thread overclock.

As you’ve probably figured out by now, the robust power delivery on the motherboard isn’t holding anything back. Cooling on the CPU is the limiting factor. The Godlike efficiently handled our 16-core/32-thread Ryzen 9 5950X as expected. Vdroop was insignificant when LLC left on auto for this board. For ambient and extreme overclocking, you have nothing to worry about.

On the memory side, we know AMD is limited to around 3600/3733 MHz when keeping FCLK at a 1:1 ratio with the memory. With this in mind, we add two more sticks and run 4x8GB at DDR4 3600, which is AMD’s current sweet spot. We didn’t run into any issues on this board as expected. All we did was enable the XMP profile and we had a stable system.

Final Thoughts

Although the X570 Godlike has been out for a while now, it was still a treat to test it with the formidable Ryzen 9 5950X. As we saw from the testing, it was one of the better performing boards with the new CPU. Overclocking was a breeze, literally a set-it-and-forget situation for our 4.4 GHz and 4x8GB 3600 settings. It feels like she’s begging for more.

Outside of its good performance, the features list is long as well. The M.2 Expander card is novel, allowing the board to support up to five M.2 storage modules. This is a good configuration for those who want a lot of ultra-fast storage. On the networking side, some competitors use 10 GbE, so it was nice to see the 10 GbE add-in-card to supplement the internal Killer based GbE and 2.5 GbE LAN ports. If I had to nitpick, I would like to see the 10 GbE integrated, perhaps a 2.5 and 10 GbE setup instead. Another unique feature of the Godlike is the audio section. Not only are there two Realtek ALC1220 chips onboard feeding the front and rear audio, but it also uses a premium ESS Sabre audio DAC that outputs to a 6.3mm jack on the integrated IO panel. You don’t see that on motherboards, period.

Competition in the flagship space is fierce as each company puts its best foot forward and says, ‘look what we can do.’ Ignoring the water-cooled boards such as the $1000 X570 Aqua, ASRock has the X570 Creator ($499.99), Asus the ROG Crosshair VIII Formula ($599.99) and finally Gigabyte’s X570 Aorus Xtreme ($699.99). Of all the boards, the Asus and Gigabyte are the group’s beauties, with MSI riding their coattails. ASRock’s Creator board is the least expensive of the bunch, but still packs flagship-level features, though it may not be the best looking or fit into your build theme. So long as you don’t need 10 GbE, the Asus motherboard makes a strong case here as the Goldilocks board between price, features and appearance.

MSI’s X570 Godlike proved itself to be a worthwhile flagship-class motherboard in the X570 space. Whether you plan on gaming, overclocking or using it for productivity, the Godlike handles it all. Do you need loads of ultra-fast M.2 storage? Check. A VRM capable of extreme overclocking? It has that too. Temperature and water flow headers? Premium audio section? 10 GbE? Check and mate. If you’re looking to build an X570 based system with flagship-type money to spend, the MSI X570 Godlike makes for a great option in that space.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Yesterday Microsoft confirmed that the new Radeon RX 6000 Series GPUs from AMD would support the all-new AV1 codec designed by Alliance for Open Media (AOM). This marks the point where all three graphics manufacturers, AMD, Nvidia, and Intel, support the encoder. Naturally, that should accelerate its popularity and support and seemingly indicates that the rival H.266 spec isn't receiving as much uptake. However, whether AMD's integrated graphics in it's desktop and mobile APUs will get AV1 encode and decode capabilities remains to be seen. 

Regarding the encoder, AV1 is 50% more efficient than H.264 and 20% better than VP9, which means AV1 video sizes are cut in half compared to H.264 encoding, but they still have the same level of quality. This is great news as more people continue to shift away from smaller resolutions like 1080P and upgrade to 4k and even 4k 60FPS, which can be expensive to store and stream on slower internet connections.

If you have an AV1-supported GPU, you can check out videos running on AV1 already with YouTube. All you need to do is select a video, right-click on the video and select "Stats for nerds' to check out whether YouTube is running AV1 or not.

Floppy disks, for some it brings back memories of a bygone age and for others it is just a save icon. But for Terence Eden the floppy disk became a means to listen tho his music, and with the help of a Raspberry Pi and a USB floppy drive. He built a “floppy-disk Walkman”.

In Terence’s own words “I have built the most inconvenient way of playing music! It is lo-fi awfulness and cyberpunk grungy.” Using a Raspberry Pi 3B+, an external USB battery and a USB floppy drive, Terence has created the means to play the music, but his method of reducing the file size lead him to use the Opus audio codec.

Using the Opus codec Terence compressed the entire 30 minutes and 45 seconds of The Beatles album “A Hard Day’s Night” into a 1.4MB file matching the capacity of a 3.5” floppy disk. The music is low quality but as Terence admits “Beatles audio was designed to be played over crappy AM radio in mono, so is well suited to being compressed using the latest audio codecs.”

Full build and the steps taken to recreate the project are on Terence’s blog. On their you will find his extensive to-do list for future additions to the projects which includes creating a circuit to control track playback, auto play disks as they are inserted, and to 3D print a case for taking the unit with him when jogging.

Fraunhofer HHI (Heinrich Hertz Institute) today announced the new H.266 codec, also known as the Versatile Video Coding (VVC) codec. The H.266 standard is the result of three years of work from Fraunhofer HHI and its partners.

With an enhanced compression algorithm, the H.266 codec  promises to maintain the visual quality of the current H.265 codec, also known as High Efficiency Video Coding (HEVC), but at an about 50% smaller file size. That would make H.266 perfect for efficiently transporting high-resolution content, such as 4K or 8K media, through mobile networks. 

Furthermore, the H.266 codec also supports HDR content and 360-degree videos.

Let's put this in perspective. A 90-minute 4K video consumes up to 10GB of space with the current H.265 codec. According to Fraunhofer HHI's figures, the same video with an identical level of quality would only require 5GB of space with H.266. The transition from H.265 to H.266 could indeed be a quantum jump if the latter can deliver on its promise. 

In its press release, Fraunhofer HHI said that the CPUs required for H.266, especially the mobile chips, are still in development. This suggests that content used with H.266 may require a hefty amount of firepower to encode or decode. However, encoding might not be a big deal if you're not a content creator, since most are more likely to consume H.266 content than create it.

Dr. Thomas Schierl, head of the Video Coding and Analytics department at Fraunhofer HHI, said that "this autumn Fraunhofer HHI will publish the first software (for both encoder and decoder) to support H.266/VVC." 

Tech companies rarely draw the line between competitor and collaborator. Intel demonstrated that duality on September 4 when it updated the open source SVT-VP9 encoder with optimizations for AMD's EPYC processors, along with other AVX2 CPUs, which debuted in 2011.

Those optimizations were made in an update pushed by an Intel engineer, Jing Li, who titled it "Fix the perf gap for Epyc CPU." But that didn't tell the whole story: Phoronix reported on Sunday that the update was "easily the largest speed-up we've seen to SVT-VP9 in its history, at least as far as AVX2 CPUs are concerned," which means some Intel processors will benefit from the change as well.

VP9 debuted in 2013 as a successor to the HEVC and h.265 video coding formats. It's set to be replaced by the AV1 codec, for which Intel also released an encoder in February. But for now it's popular because of its use by YouTube. Phoronix said in February that Intel released SVT-VP9 with the goal of "supporting real-time encoding of up to two 4Kp60 streams on an Intel Xeon Gold 6140 processor."

SVT-VP9 historically performed best on AVX-512 processors. The update pushed last week improves performance on the older AVX2 processors, which could be a welcome update for anyone who publishes on YouTube but doesn't have a top-of-the-line system. Intel's joke about fixing the performance gap on EPYC undersold the update--it should make SVT-VP9 more viable for people on both platforms.

Like we said: tech companies often work with their competitors. That doesn't mean they can't have a little fun at each other's expense, though, and it's even better if that fun is related to performance improvements for older processors that both companies benefit from.

Linux users received some good news today: Phoronix reported that Linux 5.3 will finally address issues with audio input on systems with AMD processors. Those fixes will be added to currently available versions of the Linux kernel, too, so users won't have to install the point upgrade.

Phoronix said that Linux users who rely on AMD processors have been reporting "crackling" audio input since at least 2017. The problems don't usually appear to affect the audio output, so many people were probably unaware of the issue, but there were sporadic reports of "occasional playback hiccups."

These issues were said to affect systems featuring motherboards built around AMD's X470 and X370 chipsets that used Realtek audio codecs. Linux users couldn't find a workaround to address the issue--which isn't common for that particular community--so they simply had to accept the crackle.

Or at least they did until SUSE engineer Takashi Iwai shared a series of not-quite-fixes discovered after "lengthy debugging sessions."  Those workarounds are:

Set up the proper driver caps for this controller, similar as the other AMD controller.Correct the DMA position reporting with the fixed FIFO size, which is similar like as workaround used for VIA chip set.Even after the position correction, PulseAudio still shows mysterious stalls of playback streams when a capture is triggered in timer-scheduled mode. Since we have no clear way to eliminate the stall, pass the BATCH PCM flag for PA to suppress the tsched mode as a temporary workaround.

Iwai said the patch set to debut with Linux 5.3 introduces these workarounds by setting the driver caps preset to AXZ_DCAPS_PRESET_AMD_SB, enabling FIFO-corrected position reporting and enforcing the SNDRV_PCM_INFO_BATCH flag. This should temporarily resolve audio input problems.

But the patch notes were clear about these being workarounds rather than legitimate fixes. "Note that the current implementation is merely a workaround," Iwai said. "Hopefully we'll find a better alternative in future, especially about removing the BATCH flag hack again."

Samsung has announced Exynos 9825, the first SoC manufactured on the its 7LPP extreme ultraviolet (EUV) process. The chip itself has largely remained the same, but the new process should give it a substantial boost in power efficiency. The sustained performance, (directly linked to power consumption and heat management), address areas of criticism for its predecessor, the 8nm Exynos 9820.

Samsung was the first foundry to adopt the EUV lithography, which the company said helps it print finer circuits and, thus, develop a more power-efficient processor.

However, adopting EUV also meant staying a little behind TSMC, which stuck to the more conventional deep ultraviolet (DUV) lithography for its first-generation 7nm process. Qualcomm used TSMC’s 7nm process to manufacture its Snapdragon 855 flagship processor this year, which, combined with the new Kryo 485 Gold cores based on Arm Cortex-A76 IP, gave the Exynos 9820 a run for its money, especially in the power efficiency department. The 8nm Exynos 9820, used in some Galaxy S10 SKUs this year, didn’t stand a chance.

Samsung Exynos 9825: A Mere Refresh

It seems that instead of improving the Exynos 9825 processor significantly, Samsung used a largely unmodified processor. The chip uses the same 2xM4+2xCortex-A75+Cortex-A55 core configuration at the same clock speed. The exception is the Cortex-A75 cores, which have been bumped to a 2.4 GHz clock speed, compared to 2.31 GHz in the Exynos 9820. 

The new SoC even uses the same Shannon 5000 Cat. 20/13 modem with up to 2 Gbps download speeds and only 316 Mbps upload speeds. The SoC does support integration with Samsung’s 5G Exynos Modem 5100, but the 5G modem won’t be built into the SoC this generation.

The new Exynos 9825, like the 9820 before it, also supports up to 8K resolution video encoding and decoding with the h.264, VP9 and HEVC codecs. Neither Samsung nor other chipmakers seem ready to implement the next-generation royalty-free AV1 codec, but we hope to see it in Samsung’s 2020 SoCs.

In terms of security, the Exynos 9825 also supports Samsung’s Physical Unclonable Function (PUF) to store and manage user data in isolation from the main operating system. The PUF provides features such as cryptographic key generation and storage, too, but it’s not clear if the SoC supports the new Android StrongBox Keymaster hardware security module implementation yet.

As it doesn't offer too many improvements over the older Exynos 9820, the new SoC is likely to have a short life, potentially only going inside Samsung's Galaxy Note 10, later this year. Qualcomm stepped it up this year with both the adoption of TSMC's 7nm process, as well as the adoption of Arm's excellent Cortex-A76 core. It remains to be seen if Samsung can one-up Qualcomm next year with a new "M5" custom core, or perhaps by adopting Arm's next-gen Cortex-A77 core. 

Intel published its own open source CPU-based encoder for the next-generation and royalty-free AV1 codec (a codec is a program for encoding / decoding a digital data stream or signal). Intel is one of the main founding members of the Alliance for Open Media (AOM), the non-profit group behind the development of the AV1 codec.

Intel SVT-AV1

Intel’s new encoder, called Scalable Video Technology AOMedia Video 1 (SVT-AV1), aims to fill the role of a good CPU-based encoding software tool until dedicated AV1 encoders are ready for prime time. The encoder supports the Linux, macOS and Windows operating systems.

A CPU-based encoder requires a beefy system, so it's no surprise the real-time encoding specifications for SVT-AV1 are no joke. SVT-AV1 requires Skylake-generation or newer Xeon processors with at least 112 threads and at least 48GB of RAM for 10-bit 4K video encoding. Outside of video streaming companies, these type of systems are out of reach for most. Consumers that want to encode AV1 videos may want to wait for dedicated AV1 encoding hardware to appear, which make take another year or so.

AV1 Moves Forward

The AV1 video codec is widely regarded as the next-generation codec to be adopted after h.264 and (to some degree) HEVC, not only because of significant improvements in video compression and the fact that it’s royalty free, but most of all because it seems that almost the entire technology industry is backing it through the AOMedia group. In the end, it’s adoption by video streaming sites, video-enabled applications and operating systems that will make or break a next-generation codec, and it seems that AV1 will get that support in spades.

Intel’s open source SVT-AV1 encoder joins the recently released open source dav1d AV1 decoder, so now video streaming companies have the full solution if they want to be early with their support for AV1 videos.

They say that you never get a second chance to make a first impression, but Qualcomm's just-announced Snapdragon 850 processor could change the way you look at ARM-powered Windows machines. The new platform promises significant performance, battery life and connectivity gains over the Snapdragon 835 chips that live in the first generation of Qualcomm-powered PCs.

Released this spring to mixed reviews, the Asus NovaGo TP370QL and HP Envy x2 were the first devices to be part of Qualcomm's ACPC (Always Connected PC) initiative. Thanks to their low-power 835 chips and 4G modems, these 2-in-1s deliver long battery life and convenient connectivity no matter where you.

However, when our sister site Laptop Mag tested them, they felt sluggish and laggy, particularly when using standard WIN32 apps such as the Chrome browser (they worked much better with Windows Store apps). The problem is that Windows Store apps will run natively on ARM chips, but the chip needs to emulate an x86 processor in order to run everything else.  With Snapdragon 850 both the emulation and native apps should be much quicker.

Snapdragon 850: More Performance, More Battery Life, More AI, More Systems

Qualcomm’s new Snapdragon 850 follows on from the Snapdragon 835 almost a year after. The Snapdragon 850 is built from the same silicon as the Snapdragon 845, and uses the benefits that the flagship mobile chip introduces for this generation.

This time around however, rather than calling it the Snapdragon 845, similar to the chip inside popular smartphone flagships such as the Samsung Galaxy S9/S9+, LG G7, HTC U12+, and OnePlus 6, Qualcomm is differentiating it by upping the number. The Snapdragon 850 uses the same silicon floor plan as the Snapdragon 845, but the frequencies and firmware are tweaked for extra performance.

Compared to the S845, the S850 has additional CPU frequency in play, with the Kryo Gold 385 cores having a 2.95 GHz turbo frequency rather than 2.80 GHz. The bigger differences occur comparing the S850 to the S835 platform.

Aside from the microarchitectural updates (Kryo 280 to Kryo 385, Adreno 500 to Adreno 630), the Snapdragon X20 modem can now reach speeds of 1.2 Gbps, and Qualcomm is going to leverage the AI capabilities of their Hexagon DSP a lot more with the new platform. The new platform, according to Qualcomm, should afford an additional +30% performance, +20% battery life, and +20% wireless speeds.

While the the Snapdragon 835-powered laptops can already last all-day, Snapdragon 850 promises multi-day battery life. We'll have to see how that claim stands up to our tests.

The Snapdragon X20 LTE modem now supports licenced 10 MHz spectrum with LAA, which Qualcomm states should allow for Gigabit connectivity speeds in 90% of the worldwide LTE markets when combined with carrier aggregation technology.

The S850 also uses the upgraded audio and video codec features we saw in the S845. This includes support for H.264 and HEVC 4K encode and decode for video playback services, and support for extended 4K video capture from integrated cameras. On audio, Qualcomm is stating improved dynamic range and reduced harmonic distortion metrics on its audio codec, but no exact numbers were given. Both analog and digital audio output is supported with Qualcomm’s Aqstic and aptX technologies.

Updates to Windows on Snapdragon

With the latest Windows 10 April 2018 update, the Snapdragon based systems will be compatible with Windows machine libraries compiled for ARM based processors. At present, Cortana is accelerated using the Hexagon DSP, and other assistants are in the works to also be hardware accelerated. Microsoft Edge will also be re-optimized for the platform.

Qualcomm stated that they have been working further with Microsoft to bring RS4 to the next generation of ACPC devices, fixing a number of initial software compatibility issues that arose from the initial launch, allowing more programs to run as intended.

Qualcomm has implemented a 64-bit SDK for developers looking to optimize their code, with the general mantra being that it is better to focus on native 64-bit applications on ARM rather than emulate. 64-bit emulation is still a work in progress, however Qualcomm has noted that a fair amount of software is compiled in 64-bit mode even if the program cannot take advantage of it, so the company is hoping that the 64-bit SDK can alleviate some of those edge cases.

One poignant part of 2018 has been security, and Qualcomm is stating that most of the Spectre and Meltdown issues are patched with the new platform. With new Smelt-style attacks coming through the pipe, they cannot legally say ‘all’ of the issues, but there is a team of engineers in play, as with all vendors, for what will be whack-a-mole for the foreseeable future. Qualcomm states that users will be patched as new problems arise as soon as is possible.

Snapdragon 850 Always Connected PCs: New Partners, More Systems

The focus of S850-based ACPCs, according to Qualcomm, is going to shift from traditionally PC-based companies to more mobile-focused companies. HP, Lenovo, and ASUS are all expected to implement refreshes or new designs. However we were told to expect the traditional mobile companies to get onboard.

These companies are more used to a yearly cadence update cycle, which combines well with the research (as published by Intel and others) that new form factor users tend to update more frequently than traditional users. No particular companies were mentioned, but we could imagine phone vendors like Samsung, LG, Xiaomi, and Huawei might be some of them. Each company will perform their own announcements in due course.

Computex 2018

This announcement is during Computex 2018, and we might be seeing some devices using the new platform behind closed doors or on the show floor. Qualcomm and its partners still have to fight the early first/second generation adoption pains that come with a re-imagined device category, bit hopefully we will see a wider selection and a wider distribution this time around.

For those waiting for 5G, the smart money is likely to wait for the Computex 2019 announcements next year.

The Alliance for Open Media (AOM) announced the first version of AV1, a next-generation open and royalty-free video codec, developed by major players in the technology and video streaming industries. 

A Next-Generation Royalty-Free Codec

AV1 was born out of a necessity to bring an open and royalty-free video codec to the web for which video platforms or device makers wouldn’t have to pay royalties to the Moving Picture Experts Group (MPEG), the developer of video codecs such as h.264 and HEVC. As we saw with the HEVC codec, MPEG wasn’t the only group that could ask for royalties. Other patent groups and companies came out saying that users of HEVC infringe on their patents.

The AOM brought together Google, which developed the VP9 and was working on the VP10 codec; Mozilla, which was working on "Daala"; and Cisco, which was also developing its own “Thor” codec. However, the three joined together to create one codec to beat them all, while also avoiding all the potential patent infringements. AOM’s founders’ list also included companies such as Amazon, Netflix, and Microsoft.

The IETF has also been watching closely these companies’ efforts, as it plans to make the codec an official IETF standard, that could get the name NETVC.

AV1 Specification

AOM’s announcement said that AV1 can deliver a 30% and even up to 40% greater compression over competing codecs such as VP9 and HEVC, based on multiple tests done by the alliance members.

AOM also said that AV1 will enable cross-platform 4k UHD video playback, deeper colors, brighter highlights, darker shadows, and other enhanced UHD imaging features, all while using less data.

Paul Gray, a Research Director at IHS Markit, a global business information provider, said the following:

We expect that the installed base of 4K television sets to reach 300 million by the end of 2019 and therefore there is already latent demand for UHD services over today’s infrastructure. AV1 will be widely supported across the entire content chain, especially including services. We forecast rapid introduction of AV1 content delivery to help the widespread proliferation of UHD streaming.

The release of the AV1 video codec includes a bitstream specification for future chips, an experimental software decoder and encoder to create and consume the bitstream, as well as reference streams for product validation.

AVIF vs HEIC/HEIF

Although Apple recently embraced HEIF/HEIC, and Google and Microsoft also plan to adopt it in their next versions of their operating systems, HEIC remains a proprietary format developed by MPEG, with many of the same potential issues as HEVC.

Some of the AOM members, led by Netflix, have started the development of a next-generation open image format called AV1 Still Image File Format (AVIF). The format will presumably be competitive with HEIF in terms of file size and quality but will do away with royalty and patent issues.

A More Open Web

Google’s VP8 codec, which it purchased along with On2 Technologies, never really caught on, because it took too long to achieve performance parity with MPEG’s h.264 codec. By the time it reached performance parity, MPEG’s HEVC had already arrived.

Later, although Google’s VP9 still arrived a little late on the market, it was more competitive performance wise. This along with Google’s decision to encode YouTube videos with it and convince chip makers to support it, too, helped VP9 achieve moderate success.

AV1 should succeed because virtually all the main tech players support it, it has significantly better performance than HEVC, and MPEG hasn’t had time to come up with a better alternative. Plus, all the patent issues likely didn't help. Depending on how soon Netflix and other AOM members can develop a free and open alternative to HEIF, the next-generation image format will also be an open format.

Features & Layout

What does $200 get you? By way of comparison, the number of added features is akin to the $180 boards of yesteryear and the $160 boards of Tom’s Hardware’s earliest days. We find a reasonably oversized voltage regulator, a few high-end codecs, and a couple extra controllers. In the case of MSI’s Z370 Gaming Pro Carbon AC, those controllers are USB 3.1 Gen2 and Wi-Fi.

Unlike most competing boards, the Z370 Gaming Pro Carbon AC’s Wi-Fi comes on a PCIe/USB to M.2 Key-E adapter card. The module is worth over $20, and adapter cards with the module are generally priced a few dollars higher, yet the Z370 Gaming Pro Carbon AC costs only $10 more than its non-Wi-Fi version: Could this be the starting point of good value?

Specifications

Around back we find a classic PS/2 port for legacy mice and keyboards, two USB 2.0 ports, a DisplayPort 1.2 connector with a meager 24 Hz refresh rate at 4096X2304, USB 3.1 Gen2 ports in Type A and Type-C versions fed by a dual-lane PCIe 3.0 controller, four SATA 3.0 ports (MSI calls these 3.1 Gen1), an HDMI 1.4 output with a still-meager 30 Hz refresh rate at 4096x2160, five analog audio jacks, and a digital optical audio output. Meager 4K refresh rates for on-board video shouldn’t bother gamers, because that target market uses graphics cards.

Since you are using graphics cards, you’ll be pleased to see that the two CPU-fed PCIe slots are reinforced to reduce the chance of damage that often occurs when moving around a PC with extra heavy graphics coolers. Spaced three slots apart to make room for large card coolers, these two slots switch from x16/x0 mode to x8/x8 modes when a card is detected in the second metal-shrouded slot.

A third x16-length slot is fed with four lanes from the PCH. Though technically compatible with 3-way CrossFireX, we prefer to have all of our graphics cards on the same controller. The Z370 Gaming Pro Carbon AC also has three PCIe x1 slots, and this is where we question the inclusion of the Wi-Fi adapter card: If the two graphics cards you’re using are more than 40mm thick, you probably won’t have a place to put it. If your card thicknesses are 40mm or less, you’re stuck sticking the adapter in either the black x16-legth slot or middle x1 slot. Putting it in the middle x1 slot means wrapping its USB cable over the top of the second card to a header on the board’s bottom edge. Meanwhile, the top x1 slot will be covered up by all but the thinnest of graphics cards, which causes us to question its dedicated resource.

The only reason we’d need to question the implementation of a dedicated resource to a slot we can’t use would be if the board had a resource shortage, and here’s the shocker: All feature-packed Z370 boards have resource shortages. In this case, the lower M.2 slot steals two HSIO resources from SATA and transfers them to PCIe, unless the M.2 card is SATA, in which case it steals only one port. On the other hand, the upper M.2 slot is fully provisioned for PCIe, stealing a SATA lane only when the M.2 drive is SATA.

The upper M.2 slot also has MSI’s M.2 shield heat spreader, which helps to hide those ugly M.2 drives. That part is also removable, just in case the M.2 drive you’re using isn’t ugly.

Four SATA ports and a USB 3.0 header point forward along the front edge, in front of the cards. A second USB 3.0 header is found next to the main power jack, and uses a perpendicular connector since it’s not in the way of any cards. Similarly, the two other SATA ports along the board’s bottom edge are perpendicular, because you’re probably not going to use that bottom slot for a graphics card.

The bottom edge also contains, back-to-front, a front-panel audio, RGB LED, PWM-style fan, TPM, LED power indicator, PC Speaker, Front-panel button, two USB 2.0, and Rainbow LED headers. In the midst of those headers is a demo button for on-board RGB LEDs and a jumper to change the function of the demo button.

A second standard RGB LED header is found next to the EPS12V connector, and a special Corsair LED strip connector is located near the top front corner. Many readers have asked us not to talk about RGB connectors in terms of value, but other readers will certainly be able to appreciate the breadth of connectivity.

The Z370 Gaming Pro Carbon AC includes six four-pin PWM fan headers, and all six have voltage control modes available. In addition to automatic detection, the board provides the ability to manually force the alternative fan control method.

The Z370 Gaming Pro Carbon is packed with cables: Though SATA is limited to one straight and one right-angle cable, it also includes an 80cm RGB splitter cable, an 80cm Rainbow LED extension cable, a 50cm Corsair HD RGB LED extension cable, and a USB cable for the M.2-based Wi-Fi module.  Buyers also get an HB-SLI Bridge, the Wi-Fi adapter card with installed module, two Wi-Fi antennas (the Brits will hate me for that), an I/O shield, a sheet of cable labels, a quick install guide, a full user manual, and a driver/application disc.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Software & Firmware

The Z370 Gaming Pro Carbon AC contains the software suite of MSI’s high-end Godlike Gaming, minus the Killer suite that’s unique to those network controllers. The Nahimic 2+ software suite, TriDef VR, OBS Multiplatform, WTFast, an MSI customized version of CPU-Z, and Dragon Eye are available in addition to MSI’s App Manager group.

A customized interface for cFos packet prioritization, MSI’s Gaming LAN Manager replaces the missing Killer suite. Live Update 6 continues to offer users a quick way to poll MSI servers for both software and firmware updates.

MSI Mystic Light finally worked without crashing or excessive lag, and the image shows that we were even able to set up some custom colors. The list of lighting modes now includes Off, No animation, Breathing, Flashing, Double Flashing, Marquee, Meteor, Stack, Rainbow, Lightning, CPU Temperature, Random, POP, RAP, JAZZ, Play, and Movie.

MSI Gaming App offers the above-linked submenus, and its OC Mode refers to the board’s single factory-programmed overclock of 1x over stock CPU multiplier at 1.20V CPU core.

Command Center hasn’t changed either, which is good for now as it still works. We were able to verify CPU multiplier, BCLK, and CPU core voltage using CPU-Z. Most other changes require a reboot. The Game Boost button found here functions identically to the OC Mode of MSI Gaming App.

Firmware

MSI Click BIOS 5 opens into its EZ Mode interface the first time, but is also configured to remember what mode was being used at last exit. Clicking the F7 keyboard function brings users to the advanced menu, where they can do more than just enable Game Boost (1x over-stock multiplier), XMP, or change boot order.

Our Core i7-8700K sample reached 4848 MHz on most boards, so the Z370 Gaming Pro Carbon AC’s matching result wasn’t exactly exciting. Getting the desired 1.30V under load required a combination of two settings—a 1.295V CPU core from the main OC menu and Mode 3 Loadline Calibration from its “DigitALL Power” submenu.

Our TridentZ DDR4-3866 memory was stable at data rates up to DDR4-3946 with two modules installed, and DDR4-3906 with four modules installed. It probably would have gone farther, but we’re using a hard limit of 1.355V measured at the slot to keep our DRAM overclocking comparisons somewhat consistent. Most companies try to cheat, and the Z370 Gaming Pro Carbon AC reached 1.348V at its 1.330V setting. DRAM voltage steps are only 10mV on this motherboard, and our experience indicates that an extra 5mV would have likely pushed our RAM over DDR4-4000 without violating our limits.

Memory tweakers should be able to find everything they need to optimize performance within the Z370 Gaming Pro Carbon AC’s Advanced DRAM Configuration submenu.

If you get lost in your memory tweaking, the Memory-Z submenu can help you figure out what the original timings were supposed to have been.

Although many motherboards now require customized power limits to prevent CPU throttling under full load when overclocked, the Z370 Gaming Pro Carbon AC kept our CPU at 4848 MHz under 12 threads of AVX-optimized Prime95 without manual intervention.

Few things are more frustrating than getting close to an overclocking limit and having the firmware crash, only to lose those settings forever. The Z370 Gaming Pro Carbon AC has enough space to save six custom profiles, and will even let users transfer them to and from a USB flash drive.

The Hardware Monitor tab provides custom fan mapping capability, along with the previously-mentioned PWM/DC control mode overrides for all six fan headers.

MSI Board Explorer shows detected hardware, making it a great tool to figure out whether something not detected in the OS isn’t being detected by the board. Pointing the cursor to different areas allows user to see which SATA, USB, and I/O panel devices have been detected.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

How We Test

The Z370 Gaming Pro Carbon AC goes up against its most closely priced rivals in today’s comparison.

Back during the X299 launch, we upgraded our test bed to handle the tremendous heat produced by the Core i9-7900X. Our award-winning Fractal Design S24 liquid cooler system sample serves the same purpose for the newer, lower heat Core i7-8700K. Cooler Master’s HAF-XB provides an optimal layout to blow the Celsius S24’s fans sufficiently over each motherboard’s voltage regulator.

Comparison Products

All of the boards in today’s test have a broad enough range of overclock setting to reach similar heights, but some have finer increments than others. EVGA’s 1mV setting steps are deceiving in that we couldn’t tell where we’d end up without excessive testing (the 15mV jump in DRAM, for example, was merely discovered after testing many 1mV increments).

Benchmark Settings

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Benchmark Results & Conclusion

As of this writing, the Z370 Gaming Pro Carbon AC comes in at the same price as EVGA’s Z370 FTW. That makes these products direct competitors. Less expensive boards are listed beneath these, in order of cost. Our cost-to-features analysis is reserved for the bottom of this page.

Synthetic Benchmarks

The Z370 Gaming Pro Carbon AC grabs a few top scores in 3DMark and PCMark, but that’s not always a good thing as it sometimes indicates cheating. Fortunately, we didn’t find any hidden overclocks.

Memory tweaks are one of the few places a firm can optimize its motherboard for performance, without getting called out for cheating. MSI edges out two competitors in Sandra Memory Bandwidth, and that advantage puts it in a tie for first in Sandra Cryptography.

Cinebench and Compubench show all four boards producing extremely similar performance.

3D Games

The two boards closest to the top of the chart have price and memory performance in common, and they even have similar performance in Ashes. The Z370 FTW squeaks past the Z370 Gaming Pro Carbon AC in F1 2015, however.

Nahimic Audio Solution is Talos compatible, but that compatibility comes with a performance penalty. The lighter bars indicate performance without the software enabled.

Timed Applications

Less time means more performance in our applications suite, yet the only thing remotely noteworthy is the shorter completion time for the Z370 Aorus Gaming 7 in Adobe Illustrator. The other boards make up for that loss in other tests.

Power, Heat, & Efficiency

We expect our Core i7-8700K to run at 4.30 GHz in Prime95 small-FFTs, and that’s exactly where all three landed. The Z370 FTW required a little coercion to get there (in the form of increased cache current via Intel XTU), but the Z370 Gaming Pro Carbon AC did not. This is legit power savings for performance enthusiasts.

EVGA had a cooler voltage regulator, but both boards did a good job of keeping the components well below throttling thresholds.

That means MSI takes second place in efficiency, though EVGA stretched its first place percentage by also having slightly better overall performance.

Overclocking

The Z370 Gaming Pro Carbon AC matches Gigabyte’s Z370 Aorus Gaming 7 in CPU overclocking for a first place tie, yet the Gigabyte board’s slightly closer-to-max DRAM voltage also allowed it to get a slightly higher data rate.

Performance is the point of overclocking, and the Z370 FTW produced more bandwidth at full O/C than any of its competitors. Then again, it never reached the same CPU clock.

Final Thoughts

Some of you only care about the value chart. That’s a shame, because performance-per-dollar doesn’t benefit from the added features of the two more expensive boards, the Z370 Gaming Pro Carbon AC and the Z370 FTW.

The Z370 Gaming Pro Carbon AC comes out one feature ahead of the Z370 FTW, and that feature is a Wi-Fi card that, due to the board’s layout, will likely go unused in a significant number of systems. Notice we didn’t say most systems, since most people run a single graphics card, and since there are many SLI or CrossFire options using cards that are no wider than two slots. And it’s not like “Gaming Pro Carbon” buyers who don’t use the Wi-Fi adapter have run out of options, since the non-AC version is available without the adapter, and a decent seller could ditch the adapter for more than the $10 price difference.

There’s also the overclocking stuff. MSI’s voltage settings got our CPU closer to its limit without passing it. On the other hand, EVGA’s board did a better job of overclocking our memory at even lower DIMM voltage.

The Z370 FTW already got our stamp of approval, so the Z370 Gaming Pro Carbon AC at least deserves that. Perhaps it deserves a bit more since the Wi-Fi adapter is worth something (even if you’re giving it to a friend). And MSI’s USB 3.1 Gen2 controller is a generation newer than the one used by EVGA, giving it twice the total bandwidth . . . but we no longer carry the “bit more” award.

Oh, and don’t forget MSI’s RGB (wink).

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

Asus announced that its Tinker Board is now available in North America.

Previously available only overseas, the Asus Tinker Board is a single-board computer powered by a 1.8GHz Rockchip RK3288 quad-core processor and 2GB of dual-channel LPDDR3. The 85.60 x 56 x 21mm device features 1Gb Ethernet, a Mali-T764 GPU with HD/UHD video playback support, H.264/H.265 decoding, 192kHz/24-bit audio support, four USB 2.0 ports, Bluetooth 4.0, and built-in 802.11b/g/n WiFi.

Despite its credit card-sized package, the Asus Tinker Board packs as much power as many Chromebooks. Enterprising enthusiasts and hobbyists have made use of single-board computers for gaming emulators, robotics, car computers, HTPC media streaming, mini web browsers, file storage servers, cloud servers, home automation--the list goes on and on. All that power in such a tiny package means the possibilities to create personalized devices are endless. Think of it as a Raspberry Pi on steroids.

The included TinkerOS is a Linux distribution based on the latest Debian 9 core. The Debian-based TinkerOS provides a foundation for basic tasks such as web browsing, watching videos, and listening to music. The LXDE desktop environment includes an optimized Chromium web browser and Python and Scratch coding apps. It should be noted that the TinkerOS media player, co-developed with Rockchip, is required for hardware-accelerated 4K video playback.

The Asus Tinker Board is available now at Amazon and Micro Center with a suggested MSRP of $54.99.

VideoStitch released version 2.2 of its 360-degree video post-production software. The update brings support for the H.265 video codec, compatibility with Nvidia's GTX 10-Series of graphics cards, and other features that are supposed to make it easier for video editors to work with 360-degree content.

Among those features are feathering that improves blending of input videos; better audio and motion synchronization; and support for Apple's ProRes line of video codecs to allow VideoStitch Studio to "retain high quality during the editing process" and play nice with other software. The update also includes rig presets for automatic stitching. VideoStitch said many of the current presets are for 360Heros rigs, but it plans to add more in the future.

All of these added features are supposed to help VideoStitch's latest software make the creation of 360-degree videos easier than ever. "With the new version of VideoStitch Studio, we give professional 360-degree content creators a great new tool that will save them a lot of valuable time during the post-production process without compromising the quality of their output,” VideoStitch CEO Nicolas Burtey said in a statement to Tom's Hardware.

The 360-degree video market is heating up. YouTube started to let its users live-stream 360-degree video in April, Lytro shared some of its work on a light field camera to capture even better full-circle videos in August, and in November, a company called Molanis VR revealed plans for a consumer software suite that would let more people edit 360-degree videos. VideoStitch is but one company working to help video creators embrace this tech.

There's also some exciting news on the consumer end of the rush to support 360-degree content. VideoLAN released a technical preview of VLC 360, which adds support for 360-degree videos and photos to the popular VLC media player, for public testing. The company said it plans to include this feature in the next major release of its software, VLC 3, which will make it the "first major video player app to play 360-degree videos out of the box."

Anyone who wants to make videos for those VLC users to watch can download the latest version of VideoStitch Studio now. A license for the software costs $295--similar to what Molanis VR wants to charge for its software--and a list of the changes in the 2.2 update can be found at the company's site.

The Alliance for Open Media (AOMedia) announced that it gain three major new members in ARM, AMD and Nvidia. The group was formed last year by Amazon, Cisco, Google, Intel, Microsoft, Mozilla and Netflix to create the next-generation open source video codec.

"The open source availability of our AOMedia Video project with active contributions from industry leaders marks the beginning of a new era of openness and interoperability for Internet video," said Gabe Frost, Executive Director, the Alliance for Open Media."We’re delighted to welcome AMD, ARM and NVIDIA to the Alliance for Open Media, reflecting the importance of hardware support to achieve broad industry adoption,” he added.

The new codec is supposed to replace the current generation codecs such as VP9 and HEVC. Mozilla (Daala), Google (VP10), and Cisco (Thor) were already working on their own next-generation codecs before they decided to share their resources and knowledge to create one “best codec” that would be royalty-free and standardized for the Web by the Internet Engineering Task Force (IETF).

Although the VP8 codec was already standardized for the WebRTC protocol by the IETF, it hasn’t gained that much adoption from chip makers and other major platform stakeholders. Even Mozilla stuck to using an open source implementation of the h.264 codec made by Cisco. Google’s VP9 codec hasn’t gotten much more adoption than VP8, either, even though Google uses it to encode all of its YouTube videos now.

The HEVC, created by the same organization that made the h.264 codec, has its own problems, mostly related to expensive royalties and patent challenges from other groups.

The new codec from AOMedia is meant to solve all of the existing issues by being:

Interoperable and open;Optimized for Internet delivery;Scalable to any modern device at any bandwidth;Designed with a low computational footprint and optimized for hardware;Capable of consistent, highest-quality, real-time video delivery; andFlexible for both commercial and non-commercial content, including user-generated content.

The fact that now all the major chip vendors, including ARM, AMD, Intel and NVidia, have embraced it ahead of its release means that all the new chips will support it from day one. With IETF standardizing it as a Web protocol, it should be embraced by all the browser vendors, as well.

Lucian Armasu is a Contributing Writer for Tom's Hardware. You can follow him at @lucian_armasu. 

Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Introduction, Specifications And Features

ASRock is keen on value awards. So keen, in fact that the company was once known for dropping the price of a board at only one seller for only a few weeks until after reviews were published. Motherboards normally drop in price anyway, so when ASRock got called out, it wasn't too difficult for it to stay with that low price until the rest of the industry caught up.

That's not what we're seeing in the Fatal1ty Z170 Gaming K4. Instead, the cheapest of ASRock's Fatal1ty series for Skylake CPU overclocking is just a slightly-improved board with a slightly higher price, compared to bottom-priced products such as Gigabyte's Z170-HD3.

An extra $15 gets buyers a Killer 2400 network controller, four extra CPU voltage regulator phases, one extra SATA-Express connector, and DTS Connect on an improved ALC1150 audio controller. None of those things have anything to do with the elephant on the I/O panel, a Type-C USB connector, and that's because the Type-C connector is connected to nothing more than the chipset's USB 3.0 controller. Still, those other things should be worth at least $15, in theory.

The thing about voltage regulators is that it's really hard to determine worth without O/C testing. Different components have different capabilities, though Nichicon solid capacitors on all of the Fatal1ty Z10 Gaming K4's power circuits should help to assure longevity.

Similarly, the extra SATA-Express connector comes at a time when hardly anyone is asking for one. It doesn't come with any extra SATA ports, but instead uses two of the ports supplied by the chipset. And one of the SATA-Express ports is disabled whenever an M.2 card is installed. Extra value is there, but the amount is indeterminable.

Specifications

MORE: Best MotherboardsMORE: 
How To Choose A MotherboardMORE: All Motherboard Content

Features

The Killer E2400 network controller costs more than the Realtek 8111 that's most commonly used on low-cost boards, and its actual worth depends on the type of user. Rivet networks recently showed a fairly convincing demo of its packet and bandwidth prioritization capabilities, also discussing its heuristic packet analysis as opposed to the white-list method used in third-party prioritization software.

Realteck rates its ALC1150 slightly better than its old 889 and far better than its low cost 892, and ASRock enables the codec's DTS Connect capability so that users can encode live multi-channel audio (such as games) on-the-fly to the board's digital output. That's important if you have a compliant receiver, but doesn't do much for those using 3.1 sound systems, 5.1 analog connection, or headphones. In total, all these different upgrades have different importance to different buyers, and only the buyer who needs them all can count on a value win based on features alone.

The Fatal1ty Z170 K4 uses perpendicular SATA port connectors, placing two of these above the upper PCIe x16 slot and four three-spaces below the second x16-length slot. This allows a triple-slot upper card and a double-slot lower card to be used with all six perpendicular-connector SATA ports. They also leave the slot space beneath both graphics slots empty, since these are typically covered up by GPU coolers. We'd probably put something a little weaker in the second x16-length slot, since it gets only four PCIe 3.0 lanes from the Z170 PCH and shares that bandwidth with all other PCH-connected devices over a PCIe 3.0 x4 DMI. This also eliminates the possibility of SLI, since the driver's applet requires a minimum of eight lanes in addition to Nvidia certification.

Two PCIe x1 slots are open-ended, just in case you want to add a bunch of x4-or-greater cards and don't care about reduced bandwidth. That might be good for certain GPGPU applications, or perhaps for adding a bunch of cheap graphics cards to drive a wall of displays.

Only a single USB 3.0 front-panel header is present, and it's conveniently placed in front of DIMM slots. The front-panel audio connector isn't as easy to reach, since it's in the lower-rear corner. Whether or not your audio cable reaches is between you and your case manufacturer, as most motherboard manufacturers have used this location since the 1990s.

Since it costs so little (by Z170 standards), the Fatal1ty Z170 Gaming K4 includes only four SATA cables, an I/O shield, an installation disc, and documentation.

Software And Firmware

ASRock's OC Tweaker menu includes several factory-configured overclocks, from 4.40 GHz at stock voltage to 4.70 GHz at 1.36 volts (CPU Core). The 4.6 GHz overclock also uses 1.36 volts, but we were able to reach the same frequency (at full stability) using 1.30 volts.

The Fatal1ty Z170 Gaming K4 configured G.Skill's XMP-3600 memory appropriately, but wasn't stable above DDR4-2626 with all four modules installed. The same 1.35V pushed DDR4-3501 with only two modules installed, which leads to questions about the ampacity of its DRAM voltage regulator. Then again, four-DIMM frequency limits can also be firmware related.

Should a builder choose to go with two DIMMs for enhanced data rates, they'll also find a wide enough range of timing adjustments to push the performance envelope. I could have spent weeks in here, but feedback has suggested that you'd rather see the article this month.

The Fatal1ty Z170 Gaming K4 includes "Load-Line Calibration" to boost CPU voltage as it drops under load. The amount of compensation required to keep a steady voltage depends on the current requirements of the CPU, which in turn varies with the number of active cores, core temperature, and clock. At 1.30V set core voltage and 4.60 GHz, and under eight-threads of Prime95, we found that "Level 1" allowed the CPU voltage to sag by less than 10mV.

Level 1 is the highest compensation setting. The voltage would have decreased by a greater amount had the CPU been clocked higher or needed more voltage. Conversely, a lower clock at lower voltage would have also required a lower compensation level to avoid overshooting the voltage target.

Unlike the relatively-accurate core-voltage setting, the Fatal1ty Z170 Gaming K4 produced slightly more than 1.35V with a mere 1.325V setting. Everyone cheats a little on DRAM voltage on their enthusiast boards, and some of the fastest-rated memory appears to require that "little extra stability."

Certain features traditionally reserved for software are addressed in the Z170 Gaming K4's firmware tools, including a System Browser hardware detection schematic, Online Management Guard network access scheduling, a UEFI Tech Service utility for sending support requests, Easy RAID Installer (which transfers RAID drivers to a thumb drive), Easy Driver Installer downloading tool, and Internet Flash to download firmware ROMs from ASRock's web server to a USB flash drive.

ASRock's F-Stream utility provides most of the overclocking controls from DRAM, a better-featured hardware monitor page, fan slope control and a tech service request utility. "Optimized OC" settings mirror those from firmware, and "Advanced Turbo" modes feature similar 4.60 and 4.70 GHz settings but at 1.31 and 1.36 volts CPU core, respectively. The "Auto Tuning" algorithm proved less effective, dropping the frequency to 4.30 GHz after raising the core voltage to 1.50V. That core voltage is too high for my comfort, and the board's lower achievable frequency could be due to the high voltage initiating the motherboard's over-current protection.

Driver and firmware updates are available through ASRock Live Update, which also offers links to several freeware and ASRock-specific applications. Most of these are described in great detail on the Fatal1ty Z170 Gaming K4 webpage, including the three-month premium membership upgrade to XSplit streaming services.

Test Results, Overclocking And Conclusion

Test System Configuration

To assure accurate comparisons, we repeat the benchmark configuration from our initial Z170 roundup in every Z170 motherboard review. The Fatal1ty Z170 Gaming K4's performance and overclocking capabilities are compared to previously-reviewed $115 to $160 Z170 motherboards.

Drivers

Benchmark Suite

Comparison Motherboards

By default, Intel's Turbo Boost ratios decrease as the number of loaded cores are increased. "Enhanced" modes added by various manufacturers tell the CPU to ignore default Turbo Boost ratios and use the highest single-thread ratio regardless of the number of cores deployed. ASRock calls this "Multi Core Enhancement" and, as with the other products in today's comparison, it's disabled throughout these tests. Advanced power modes that are usually disabled on enthusiast boards are also enabled through these tests. Enabling Intel's complete range of performance-boosting and energy-saving technologies allows all motherboards to be compared in a completely fair and consistent manner.

Synthetic Benchmarks

The Fatal1ty Z170 Gaming K4 scores right down the middle of our synthetics suite, showing that the motherboard is neither cheating (via flexible BCLK seen in past products) nor misconfigured.

Manufacturers have a little wiggle room for performance optimization in advanced DRAM timings, but most choose not to risk stability to seek further performance. The Falat1ty Z170 Gaming K4 roughly matches ASRock's previous review sample, the Z170 Extreme4.

3D Games

The Fatal1ty Z170 Gaming K4 leads slightly at the low settings of Grid 2 and Far Cry 3, but not by enough to be visible on the screen. A look back at test notes also showed that the minimum scores were on par with other machines, so we'll just keep this tiny difference in mind for further performance discussions.

Timed Applications

The Fatal1ty Z170 Gaming K4 was quickest in Lame MP3 encoding but tied for slowest in iTunes and Adobe Premiere workloads. In total, these tiny differences appear to approach the "average of averages."

Power, Heat And Efficiency

Equipped with fewer features than its Z170 Extreme4, ASRock's Fatal1ty Z170 Gaming K4 also draws slightly less energy.

The Z170 Claymore pulls down an "average of all" performance, allowing the Fatal1ty Z170 Gaming K4 to produce a 0.6 percent above average performance rating even as its other rivals score slightly higher. Reduced power consumption gives it a 3.5 to 4.4 percent efficiency advantage over the Z170-HD3 and Z170 Extreme4.

Overclocking

The Fatal1ty Z170 Gaming K4's range of firmware settings far exceed the capabilities for any of our hardware or our ability to keep it cool, yet we have some doubts about the board's ability to maintain some of those high settings without bumping up against the limits of over-current protection, which prevents the board from melting down.

Even though the Fatal1ty Z170 Gaming K4 had a slight problem with voltage sag, it pushed our Core-i7 6700K to a stable 4.60 GHz. That far exceeds the limit of Gigabyte's cheaper Z170-HD3, and proves that the higher-current voltage controller is probably worth the difference in motherboard price for any user who relies on the performance advantage of CPU overclocking. DRAM overclocking was another matter, as the board's class-leading 2-DIMM data rate fell to a third-place 4-DIMM stability limit.

One might argue that four-DIMM overclocks aren't important on a dual-channel board, but the insufficiency left the Fatal1ty Z170 Gaming K4 unable to reach our DDR4-2933 evaluation setting with all four modules in place. Dropping it down to two modules caused it to lose by around 10 percent against ASRock's own Z170 Extreme4, which used the same settings with four modules installed.

Conclusion

When rating the suitability of a product for a particular market, we have to keep in mind the limits for that market. The Fatal1ty Z170 Gaming K4 for example is a dual-channel board priced around $130, so we might expect that builders won't use four-DIMM configurations rated beyond DDR4-2600. Some of our tests indicate that a Skylake processor can extract a little more performance from four DIMMs instead of two, so we really should put aside the "but it's only dual-channel" argument and focus primarily on price. A 4x 4GB (16GB) DDR4-2400 CAS 15 kit sounds like a good bet, especially since these are now available for less than $80.

On the CPU core side, the Fatal1ty Z170 Gaming K4 is a stellar overclocker. That's where the most performance can be found, so that's where most of us focus our overclocking efforts. Though we've seen high-end boards push our CPU to 4.60 GHz at a relatively-conservative 1.30V, the Fatalt1ty Z170 Gaming K4 is the only sub-$160 model to accomplish this small feat.

Enabled by a larger voltage regulator, extra CPU overclocking probably gives the Fatal1ty Z170 Gaming K4 an extra $15 worth over its cheapest rival. That's before we begin discussing its DTS-Connect-enabled audio controller and Killer packet-prioritizing network controller upgrades. There's also an extra SATA-Express interface on an identical number of ports...silly me, let's talk about overclocking!

MORE: Best Motherboards
MORE: How To Choose A Motherboard
MORE: All Motherboard Content

Introduction

NAS industry veteran QNAP has brought a new product series to market and given AMD a design win along the way. QNAP plans to tap AMD's extensive GPU knowledge to solve problems with playing back codecs that were never fully resolved in Intel processors. The flagship model using an AMD APU in the QNAP product line is the TVS-863+, the unit we're testing today.

Windows Media Center is dead, but that doesn't mean the principle will not live on. Video on a big screen is the ultimate viewing experience. Sadly, many of us don't have a lot of time to binge-watch  "Star Wars," so we absorb content on our phones and tablets. Keeping content on demand requires storage, the native function of NAS products. By supporting high-definition video playback on local and remote devices, NAS products finally have the killer app that pulls many consumers in the door.

The TVS-863+ is a step above what we see in most home theaters and is designed for small business use. The unit still leverages AMD's powerful APU processor that provides superior video support for a wide range of codecs. This model goes well beyond video playback capabilities. Video is just a big feather in the hat. The unit we're testing today comes equipped with a 10-gigabit Ethernet and 16 gigabytes of DRAM to run more applications than most off-the-shelf models. The TVS-863+ is so powerful that it can operate as a desktop computer without degrading network storage performance.

QNAP leads other NAS builders in software features and has no plans to slow development. In June, the company displayed QTS 4.2, the next-generation proprietary operating system for QNAP products. Version 4.2 adds container support for Docker and LXC while improving on a number of existing functions. What are containers you ask? That is the point. QNAP incorporates the next big thing before most even know it's a thing.

Technical Specifications

NAS products should be treated as complete system solutions rather than components. That said, the hardware allows users to run more software features at the same time and plays a role in the user experience.

MORE: All Storage ArticlesMORE: Latest Storage News
MORE: Storage in the Forums

The TVS-x63 product series consists of four models that range from four to eight drive bays. The TVS-863+ leads the series in features, while the TVS-863 nonplus model is nearly identical but lacks 10GbE connectivity out of the box and ships with either 4GB or 8GB of low-power DDR3 DRAM. All TVS-x63 products include a user-accessible PCIe 2.0 x4 expansion slot for 10GbE add-in-cards and two DDR3 memory slots that accept up to 8GB memory sticks each.

Today, we're testing the TVS-863+ -16GB, the flagship of the series with every hardware option included right out of the box. This series is considered an affordable way to give users access to 10GbE network capabilities and up to eight drives in a storage pool. The next tier up in the QNAP product line is the TVS-871, which costs $2,199 (Newegg) and is configured with a Core i7 processor, 16GB of DRAM and 4x gigabit Ethernet ports (optional 10GbE, but not included). The affordable AMD-powered TVS-863+ allows the system to be used for more virtual machine storage, thanks to the 10GbE network.

QNAP is an innovation leader in hardware features. The TVS-x63 product series can use any two drive bays for SSD cache. When enabled, SSD cache increases read and write performance and can be tuned to accelerate small random files or all file types.

The heart of the system is an AMD 64-bit APU that features Radeon HD graphics technology for hardware-accelerated transcoding. The processor also provides a hardware-accelerated encryption engine with AES-NI that is full AES 256-bit.

I've always felt the increased video capabilities are out of place on many of the SMB NAS products but this market is all about filling in boxes with check marks. The system can be used for digital signage, playing videos like presentations in conference rooms and so on. The transcoding features paired with the always connected software does give the video features some legs. Businesses can keep sensitive presentations in house on the NAS but allow company representatives to play them in remote locations without the digital data actually moving to insecure offsite computer systems.

Software Features

Early in the setup process QNAP SMB systems ask if the system will be used in a business or home. Your answer dictates the initial system configuration and the applications installed. At any point you can add business or home applications depending on your need. I almost always choose business because I don't use all of the applications for home users.

Over the last quarter QNAP released QTS 4.2 Beta, our first look at the upcoming operating system GUI that sets the stage for years to come. QTS 4.2 isn't a radical departure from previous releases but it does look and feel a little different.

QNAP's QTS software ships with several pre-installed features from the factory. The loaded QPKG packages, QNAP's version of an app for your NAS, are enough to get you started. The company has more than 160 additional QPKG apps available for free. Users can download the applications from the NAS operating system via a dedicated page. The software will automatically download the application and  install it without hassle. The apps offer a good mix of client software, the things you may use in your home, and pure business applications. You can see the full list here.

QNAP's software also goes beyond the NAS as well. A number of mobile apps for iOS and Android allow users to take the content stored on the NAS on the road through an Internet connection. The transcoding feature in the TVS-863+ plays an important role when streaming video to remote locations. The mobile applications support much more than just video. You can see the full list of mobile applications here.

Pricing, Warranty & Accessories

We found the QNAP TVS-863+ 8GB and 16GB available from a number of e-tail stores in the U.S. SimplyNAS, a specialty store for NAS storage products, had the lowest prices when we researched this product. The 8GB DRAM model cost $1,399, while the 16GB model was $100 higher, at $1,499. Newegg and Tiger Direct also carry this model.

The TVS-x63 product series includes a standard two-year warranty. QNAP offers optional extended warranty periods that take effect after the original warranty expires. The extended warranties come in one-, two- and three-year increments but must be purchased within the first 60 days of the initial purchase.

The TVS-863+ model under test today already ships with every option, but a few accessories are available. The system comes with a single port 10GbE NIC but QNAP offers a dual 10GbE card that works in the system. The RM-IR002 remote control is also supported (but not included) with this unit for media playback.

A Closer Look

The QNAP TVS-863+ ships in a full retail package that gives shoppers a lot of relevant information in a retail environment.

Most SMB NAS products are sold online in North America, but we give QNAP kudos for printing the information on the package. In some countries you can find NAS in retail stores or computer malls.

Inside the package, we found the TVS-863+ secured in closed cell foam with the accessories in a separate compartment from the main unit. The corners are well protected, making it very unlikely for this model to be damaged during shipping.

The TVS-863+ ships with a quick installation guide, fine and coarse screws for mounting 2.5-inch and 3.5-inch drives and two sets of keys for the locking drive bays. The system also ships with a power cables and three Ethernet cables.

The gold standard? The TVS-x63 series comes in a champagne gold color that is popular in home theater products in Japan. I'm not sure if that was QNAP's inspiration for producing this series in gold rather than the traditional black, but it's a nice change-up in the test lab.

The system allows users to see IO activity in a number of ways. LAN and USB LEDs on the front fascia indicate activity and LEDs that shine through the drive bays also show disk connectivity for the bay and activity. A long, thin LED-lit area on the far left indicates power.

The power button to turn the system on is just above the one-touch-copy USB 3.0 port. To shut the system down you need to hold the power button. It's a thoughtful feature given the nearness to the copy function for external storage devices.

The front LED display gives system status information and warns in the event of a drive failure. The display is useful for the initial setup as it displays the number of loaded drives, IP and other useful information.

A sticker on top of the system gives users the key to connect to QNAP's myQNAPcloud feature for mobile devices.

The system is designed so most of the air flow passes over the hard drives to keep them cool. The compute side does have vents so air can pass through to keep the processor, chipset and RAM cool.

The back of the system has a majority of the IO ports. This is a full feature system with four USB 3.0 ports on the back as well as two HDMI ports for 1080P video playback and NAS-as-Desktop functions. Users can use the NAS as a computer by plugging in a keyboard, mouse and video to the system.

The star of this system is the 10-gigabit Ethernet. This one port has the same throughput as 10 traditional gigabit Ethernet ports. Two gigabit Ethernet ports are also on the back of the system for fail- over or connectivity to a separate network.

This system has an internal power supply so you don't have to manage placement of an external power brick.

Software Interface

In the future we will add comprehensive feature stories showing NAS operating systems and GUIs. There are too many pages, too many features and too many add-on packages to introduce in a product review.

There are hundreds of pages in the QNAP QTS operating system but they all start here on the main page. The icons are similar to your Android phone icons -- you can move and position them. There are three pages, so you can organize them by pages if you like. Users can also add or remove icons as needed.

A health check feature on the lower right corner of the dashboard opens up to what we see here. The monitor is a quick way to catch the latest QNAP software news and details about firmware updates, and  to view the status of your system. You can monitor the status of your array, CPU and memory usage, and network traffic activity. There is also a pane that shows if other users are logged into the system. This is useful in an office environment. You don't want to update the firmware and restart the system if a co-worker is using the system for a presentation at a remote location.

Before we even get into the applications the system has a lot of configuration features that are user- accessible. Everything comes configured from the factory in at usable settings. Users can easily configure options to suit their needs giving a custom experience.

Users can elect to disable and hide all of the home/multimedia functions with a single check box. This is a nice feature to save processing power and system memory if the features are not being used.

In the Storage Manager of the QNAP QTS OS, users can build arrays either in volumes or by pools of disks. Disk pools were introduced just a few years ago and allow the NAS appliance to tackle storage like an enterprise SAN. The same disk in the system can be part of two separate RAID arrays. For example, part of the drive may be part of a RAID 0 array and another section of the disk may be part of a RAID 5 array.

When we test QNAP NAS appliances we use the single-volume option. Systems with seven drive bays and lower are tested in RAID 5, while systems with eight or more disks are tested in RAID 6.

Users have a lot of options when it comes to the network. QNAP enables several modes of network teaming, including 802.11ad link aggregation. There are a few ways to connect to the system from outside of your intranet. The server is accessible from QNAP's cloud feature, a free service through a specialized website, or through a third-party DDNS service. 

Sequential Data Transfer by Size

In the coming weeks we will detail our extensive testing process and share how we configure systems. In today's review we're comparing the QNAP TVS-863+ to two other products on the market. We are also testing the QNAP TVS-863+ with and without SSD cache-enabled.

Without SSD cache, the TVS-863+ uses eight Seagate NAS hard-disk drives and 10GbE connectivity back to the network. For the SSD cache test we used a SanDisk Extreme PRO 240GB SSD as the cache device and seven Seagate NAS hard disk drives. We configured the SSD cache with the default settings that omit sequential reads and writes but accelerate random data. All QNAP tests were run with the EXT4 file system and RAID 6 for dual redundancy.

For comparison products, we chose the ASUSTOR AS5108T, an 8-bay solution without 10GbE. This system uses the EXT4 file system. The 10GbE comparison product is the Netgear ReadyNAS 716X with six drive bays and the BTRFS (B-tree File System). Both the AS5108T and ReadyNAS 716X were tested in RAID 6 to match the QNAP TVS-863+.

To start the testing portion of the review, we swept sequential block sizes from small to large to see the throughput scaling and find peak QD1 performance. This test also shows us that with the default settings, the QNAP cache algorithm doesn't waste time with sequential data. Users can uncheck a box in the setup procedure to cache sequential data, but most users will not run the system that way.

Here we perform the same type of test-writing data instead of reading it. Again, the cache-enabled test doesn't show a difference, as expected. The QNAP TVS-863+ does show a dip in the performance curve at 128KB, the standard measurement block size for sequential data. In our testing we also show sequential performance with 1MB block sizes as well, so the dip will only affect half of our sequential tests.

Random Data Transfer by Size

As a pure storage server, random data makes up only a very small portion of the workload. The closer you push the NAS into an application server role, the more important random data performance becomes.

The SSD cache really takes off here and accelerates random read performance. The performance increase will depend on the cache hit ratio that you can monitor in the operating system software. The monitor is important for tuning the SSD cache with the limited but effective settings.

The SSD cache has a slightly negative effect on random writes outside of 4KB blocks, where it helps performance. With or without the cache, the QNAP TVS-863+ delivers blistering random write performance. QNAP spent a lot of time tuning systems for virtual machine storage. Many VM writes take place at 8KB. The chart shows QNAP's tuning has paid off at that block size.

Sequential Performance

In this section we measure sequential reads at two block sizes while we increase the queue depth. The 128KB block is the traditional size used by many manufacturers to measure sequential performance. The 1MB block size helps us to determine performance while transferring larger data such as pictures, videos and so on.

Users can tune the storage for heavy sequential or random workloads during the array setup process by changing the stripe size.  We use the default 64KB stripe that balances sequential and random performance. This is important to remember when setting up an array. QNAP uses pools so it's possible to direct your data to tuned areas that can use the same physical disks.

Reading back sequential data, the QNAP TVS-863+ is a little slower than the ReadyNAS 716X. That said, the TVS-863+ does deliver consistent performance after the initial ramp-up in queues.

The QNAP unit delivers the highest peak sequential write performance but only overtakes the Netgear model after 4QD. When we add several streams of sequential load, the data takes on a random nature as the array takes in all of the steams and pushes the data to the drives. This will come into play in the next set of charts where we measure random performance at scaling queue depths.

Random Performance

The SSD cache random read results delivered by the TVS-863+ are truly impressive to a degree that is rarely seen in NAS products at this price level. QNAP has improved the algorithms since first releasing the feature just over a year ago on several models. Without the cache enabled, the TVS-863+ falls in line with the Netgear comparison NAS but trails the ASUSTOR model from QD4 on. 

The TVS-863+ delivers exceptional random write performance as well. Here the cache slows random write performance slightly but again we see more consistency over the ASUSTOR unit and a larger increase in IOPS over the Netgear.

Sequential Mixed Data Sweep

A sweep across the storage area starts out on the charts as 100 percent reads and adds writes 10 percent at a time. On the far right side, we get to 100 percent data writes. The charts highlight 70 percent reads with 30 percent writes, traditional workstation workloads.

The 128KB sweep is quite different from the 1MB sequential sweep. The 128KB sweep shows signs of tuning for 80 percent reads, a workload associated with client systems and possibly for virtual machine desktop system storage. The 1MB load gives us the traditional bathtub curve we see on most storage products.

The TVS-863+ fares well in these tests but trails the RN716X slightly.

Random Mixed Data Sweep

All three systems use the same hard disk drives and the random performance at 4KB and 8KB are very close in all tests. The SSD cache test does stand out, as you would expect. The three systems running without SSD cache show sharp spikes with 100 percent random writes. This is from the DRAM cache in the systems.

The TVS-863+ APU is able to keep up with the Intel Xeon in the Netgear ReadyNAS 716X. We really expected the Xeon system to run away from the others in random data tests but that wasn't the case today.

Single-Client SMB Performance

Windows users mostly employ two methods for transferring data to network storage: SMB/CIFS and iSCSI. In this set of tests, we're using SMB 3.0 from a Windows Server 2012 R2 machine to transfer data to a network share. In this case the share is //nas/test; in other words, a shared folder on our test devices. Intel's NASPT tool tests with workload traces and is designed specifically for network storage.

In many of these tests the SSD cache increases performance over the non-cache configuration with the TVS-863+. The system configuration and file system also play a role in these tests. The Netgear with a Xeon processor and BTRFS dominated many of the tests. Both Thecus and Synology support BTRFS now, but QNAP hasn't moved beyond the EXT4 file system. QNAP's pool system is more in line with the type of storage we'd find on large enterprise SAN systems, so we don't think a move to BTRFS makes a lot of sense at this time.

Multi-Client SMB Performance

Our SMB multiclient test represents years of hard work, a lot of money and several companies coming together to make the test a reality. There are 10 dual Xeon servers, each with two Hot Lava Shasta 6-port NICs. Each server has 12 Windows 7 Hyper-V installations and each Virtual Machine has a dedicated gigabit Ethernet connection back to a series of linked enterprise network switches. The Windows 7 VMs run custom software that runs Microsoft Office traces with programs like Excel, Access, Word and Power Point.

In high-client count environments, the sequential data is randomized because so much data is reaching the system at one time. With a low number of clients, the sequential data stays sequential but the block size of the file updates is still very small since we are dealing with Microsoft Office files.

This is the first test where we really start to see the AMD APU have issues with performance compared to the other products in the charts. The first chart shows throughput and it takes the SSD cache to bring the TVS-863+ performance to the same level as the Netgear ReadyNAS 716X. We've run this test on a large number of QNAP devices over the years with various Intel processors, and this is the first system that performed so poorly on this specific test.

The second chart shows the average latency for each client. This test is actually more important than the first as it has a direct relationship to the user experience at each office desktop. If you plan to use the TVS-863+ with several users running office software off of the central storage, you will want to take advantage of the SSD cache option.

Single-Client iSCSI Workloads

iSCSI is an amazing technology that allows users to mount a volume to a host computer and have it control the volume as a local drive. You can even set the computer up to boot from the iSCSI share, just like a SAN.

QNAP's iSCSI configuration delivers some of the best performance in the business. The cache-accelerated volume manages to shave a few seconds off of each test that you may use at home. Even without the cache enabled, QNAP's iSCSI runs circles around the other systems.

Standard Workloads

In this series of tests we ran traditional server and workstation storage workloads. There are two charts for each workload. The first chart simply measures IOPS performance with increasing queue depths. The second chart is one you don't see used very often. In the snake charts we show latency (vertical) with IOPS (horizontal). Ideally, the second chart will show the IOPS moving to the right and staying very close to the bottom of the results area (showing lower latency).

In many of the workloads we found very little performance scaling as the queue depth increased with the TVS-863+ without cache. The Netgear ReadyNAS 716X delivered the same flat-line results and the ASUSTOR AS5108T showed a little variation. With the SSD cache enabled, the TVS-863+ picked up the pace quite a bit and manage to increase performance until reaching peak. After the peak performance the IOPS drops due to the latency increasing for each additional IO.

The snake chart shows the latency and allows system administrators to tune the system for quality of service. Using the database test as an example, system administrators can use the cache system to get 5400 IOPS at roughly 3ms response time from the storage system. Any more IOPS past that point will have a negative impact on latency.

Conclusion

There is a lot of hype about the TVS-x63 series. Some of the hype is justified and some not so much. I don't think users will buy the system just for the AMD processor unit and very few will purchase the 8-bay model we tested today for the enhanced video for the Radeon GPU. The TVS-863+ with eight drive bays is a little too large for most home theater installations -- the TVS-463 (4-bay) and TVS-663 (6-bay) fit the home theater replacement a little better. The 8-bay model we tested today is a good fit for business use and offers powerful features at a great price point.

Starting with the primary function, storage. The TVS-863+ uses eight drive bays that support the latest in hard disk drive technology. With HGST's new He8 drives with 8TB density, users can easily store up to 64TB of data. After RAID 6 overhead, that comes out to about 48TB of usable space with dual disk failure redundancy.

Your data shares are no longer limited by wires or proximity. QNAP's cloud service gives your users access to permission-based files from anywhere in the world. The cloud service is one of the standout features for this model as it keeps the files local rather than sharing them to Google or Amazon. You control the content and who has access to it.

Other software add-on packages extend the range of useful business features as well. Virtualization and soon-to-come Docker support open the door for any operating system and feature you may want to add. Many features already exist, though, in the low-overhead QTS operating system via apps, even CPU- hungry databases run on the system.

MORE: All Storage ArticlesMORE: Latest Storage News
MORE: Storage in the Forums

Chris Ramseyer is a Contributing Editor for Tom's Hardware, covering Storage. Follow him on Twitter and on Facebook.

Follow Tom's Hardware on Twitter, Facebook and Google+.

Microsoft announced in a blog post that it's going to finally adopt the open source VP9 codec in Windows 10's Edge browser, as well as other open source audio and video codecs in the future.

Microsoft was initially reluctant to adopt the previous-generation VP8 codec, preferring the proprietary h.264 video codec instead. At the time, there was a battle between who's going to support what, and Google and Mozilla sided with VP8 while Apple and Microsoft chose h.264. Apple usually tends to go its own way, regardless of what everyone else does, but if Microsoft had adopted VP8 back then as well, VP8 and VP9 may have already had wide hardware support.

For now, Microsoft will support software decoding for VP9, but where the hardware exists (typically on some mobile chips), hardware decoding will also be supported for more efficient video playback. For software decoding, Microsoft recommends more powerful PCs. This is also why, for now, the VP9 support will be enabled only as an experimental flag in the Edge browser.

Microsoft Edge will support VP9's adaptive streaming using Media Source Extensions in the initial implementation, which will be detectable using the MediaSource.isTypeSupported() API. Microsoft said it is also working on future support for VP9 for media tags and local playback.

The company is also considering support for other codecs such as OGG, Vorbis and Opus. Opus, especially, tends to be paired with VP9 in the WebM container format, which YouTube currently uses by default.

Microsoft recently became part of the Alliance for Open Media, where companies such as Mozilla, Google, Cisco, Intel, Amazon and others will work on improving upon VP9, Daala and Thor to create a single open source and royalty-free video codec that everyone can use without limits.

The VP9 video codec will be supported in the Edge browser in the Windows 10 Insider Preview builds soon and will later come to a more stable build, as well.

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

Google announced today that YouTube users have already watched over 25 billion hours of VP9-encoded videos during the past 12 months. VP9 is Google's own open source video codec and alternative to HEVC. It's also the successor to VP8, the codec Google received by acquiring On2 Technologies (and later made open source).

The VP8 codec didn't manage to become particularly popular in part because it arrived late, and its main competitor, the AVC/h.264 codec, was already becoming entrenched on the video market. The VP8 also wasn't competitive enough performance-wise.

Likewise, the VP9 codec arrived slightly later than HEVC, but this time Google pushed it much more aggressively to its hardware partners and via its own products and services. Although Google promised to give up on h.264 and go all-in with VP9 in Chrome back in 2011, that never happened.

Now, Google is using VP9 as its main codec for YouTube videos. According to the company, a significant portion of those 25 billion hours of videos played in the last year couldn't have been watched in HD quality without the VP9 codec.

“This new format bumps everybody one notch closer to our goal of instant, high-quality, buffer-free videos. That means that if your Internet connection used to only play up to 480p without buffering on YouTube, it can now play silky smooth 720p with VP9," said Google in a blog post.

Mobile users, or those who live in countries where data plans are expensive, can also save money or bandwidth by watching VP9-encoded videos instead of h.264-encoded ones. At the same image quality level, VP9 videos usually have a bitrate of roughly half that of h.264 videos, which gives significant bandwidth savings.

VP9 was also designed from the beginning with high resolution video content in mind. At larger video sizes, the VP9 codec becomes even more efficient compared to h.264. Therefore, thanks to VP9, YouTube is serving uninterrupted 4K streaming to a growing audience as even lower-speed connections can now play 4K videos. In the past year, the number of 4K videos being uploaded to YouTube have tripled.

While Google and the MPEG-LA (HEVC's creator) are both currently pushing their own video codecs, Xiph, Mozilla, and the IETF are working on the next-generation NETVC (former Daala) codec. NETVC is supposed to go beyond the performance that the VP9 and HEVC codecs have today (roughly half the bandwidth of h.264), while at the same time not being patent-encumbered.

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

After the widely adopted h.264/AVC video codec and its open source VP8 competitor from Google, we're seeing the next-generation h.265/HEVC and VP9 video codecs adopted in more hardware. Two years ago, though, work began on what was at the time a "next next-generation" open source video codec called Daala.

Daala has already been under development by Mozilla and Xiph, the non-profit organization behind the Ogg container format and the Vorbis audio codec. Xiph also worked on what later became the Opus codec, which was standardized by the IETF (Internet Engineering Task Force). Opus got implemented in all browsers that support the WebRTC protocol for live video conferences over the web; thus, Xiph and the IETF already have some experience in getting codecs widely adopted.

Just as the IETF took Google's SPDY protocol and turned it into the HTTP/2 standard, IETF seems to be trying to do the same with Daala and turn it into the NETVC video codec standard. Much like Daala, NETVC seems to have largely the same main goals. The IETF wants it to be:

Optimized for real-time communications over the public InternetCompetitive with or superior to existing modern codecsViewed as having IPR licensing terms that allow for wide implementation and deployment

The web has been beholden for quite some time to proprietary patent-encumbered technologies such as h.264. This hasn't been as big of an issue as it could have been, because MPEG-LA, the group that owns the h.264 and HEVC patents, has only asked browser vendors and hardware companies to pay royalties, but not regular Internet users or small sites that put videos up.

Even so, the web has always evolved at a faster rate when using open patent-free technologies, and it would be ideal if the same spirit was maintained for any future web technologies.

One of the ways IETF wants NETVC to go beyond even HEVC in terms of efficiency and performance is by making sure that it's highly parallel and works well, not just with multiple CPU cores, but also with SIMD/GPU hardware.

Although the codec will eventually be supported in hardware, it can take many years to achieve ubiquity. Therefore, the IETF wants the codec to work well enough even on existing hardware in the software mode, until the dedicated hardware decoders arrive in future chips.

The Daala project was already showing great progress, but it was far from finished. Now that the IETF is taking over, it's likely going to take at least another year or two until the codec is ready to be pushed into the market. If by then NETVC manages to be so far ahead of other video codecs in terms of efficiency, then the video codec will see more adoption by browser vendors and hardware makers on its merit, without the IETF having to do too much convincing.

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

Fraunhofer, a German research company working on improving audio technology, together with 11 other companies including Ericsson, Nokia, Qualcomm and Samsung, announced that they've developed a new Enhanced Voice Services (EVS) codec that can boost audio quality significantly.

The companies started working on the new wide-band audio codec in order to bring the same level of quality to voice calls that you hear in other media. The GSM standard uses the AMR codec, which is a speech-only codec that is limited to the human voice and shuts out everything else, including other background sounds or music. Most of the current phone services remove at least three quarters of the audible spectrum, according to Fraunhofer.

The AMR codec has a bandwidth of 300 Hz-3.4 KHz, which has recently been doubled to 7 KHz (also called "HD Voice"). The new EVS codec has a bandwidth from 14 KHz (superwideband audio) and up to 20 KHz (full band audio) and has an audio quality comparable to streaming music. EVS was designed from the ground up to work with packet-switched services such as VoLTE (Voice over LTE). Fraunhofer calls it "Full-HD Voice."

The EVS codec allows bit rates ranging from 5.9 Kbit/s to 128 Kbit/s, which allows phone service providers to optimize their networks as needed. Bit rates for narrowband and wideband start at 5.9 Kbit/s, while superwideband Full-HD Voice bit rates start at 9.6 Kbit/s.

The EVS codec includes features such as a multi-rate speech and audio codec, source controlled variable bit-rate adaptation, a voice/sound activity detector, comfort noise generation, an error concealment mechanism, a channel-aware mode, and powerful jitter buffer management.

Fraunhofer said that the audio quality of the new codec is so good that reporters could go "live-to-air" from a remote location by simply using their phones. For regular consumers, the new codec comes with benefits such as being able to listen to music from a distance, such as from a concert they're not attending, with no degradation in quality.

According to the company, the new codec could start arriving in phones within 18 months. The codec can be implemented currently in software, but it's likely to arrive in hardware in the DSP in the future in order to achieve higher performance and efficiency.

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

A Guide to Professional Video Editing Software

Video Editing Software Buyer's Guide:

1. Video Editing Hardware: What You’ll Need To Get Started
2. A Guide To Free Video Editing Software
3. A Guide To Inexpensive Video Editing Software
4. A Guide To Professional Video Editing Software

Ready for the big-bucks big time?

There’s no way, in a single article, to fairly describe or compare the depth and breadth of these powerful, professional appliances. And, of course, each of you is coming at this exercise from differing needs and goals. So, I’ll vector toward striking the high points – and singing the sour notes – of each toolkit in an abbreviated effort to give you a sense of their flavors. That, at least, will save you some time when you embark upon your own deep diving expeditions…

Shodan Edit Ninja — Advanced Video Editing Software

The quanta will now leap. You aren’t satisfied with the features of basic editing packages. You want professional results. You need advanced utensils. May I assume that you’re intending to make serious money with them?

Caution: it’s one thing to work on your own projects at you own pace, but quite another to service the needs of clients. Your customers will want all three sides of the creative triangle: Fine Quality/High Speed/Low Cost. You’re going to have to learn to fly on your NLE. And it has to be good enough to keep you aloft.

You may need to bump your hardware up too. Pop over to our recommended hardware configurations to see how much horsepower you’ll require. Check out the hardware discussion in our Video Editing: what Do You Need? Part 1.

Get an even bigger disk. You may be working on longer form shows, a gravitational effect of convergence/democratization. Some “pro-sumer” price-point cameras are spitting out true 4K video now.

Learn to love RAID arrays. It merely sucks when you lose a week of work on your personal project. It’s a total catastrophe if you blow up a client’s job. And if you need to keep projects for later revision, you must have a media management strategy. This is the least fun—but most necessary—part of high-end editing.

Consider doubling up on your GPU. Maybe the most fun of having a capable editor driving a powerful platform is dropping in high-end visual effects plug-ins like those from BorisFx. Or get your kicks “round-tripping” elements out to a parallel universe of specialty tools as one can on Planet Adobe between Premiere Pro, After Effects, Photoshop and Illustrator.

Professional video tools and ever-faster processors confer great creative power. We can do wonders on our own, in seclusion, like monks masterfully illuminating a manuscript.

But something has gotten lost on the way to the edit room in this last decade: the producer in the room. Several of my broadcast editor-friends complain of the growing gap between the editor and the show; a physical and psychic disconnect from the other creative trades on projects. Written in our DNA, a time when we used to invent, share, embellish and, yes, edit stories around the fire. As recently as the turn of the 21st century, we huddled like a family in an edit suite. Now, we DropBox low-res reference cuts to tiny phone screens to wait for notes and approvals from distracted clients. No real-time creative interaction.

Avid Media Composer

Avid Media Composer (Best in Class)

(Mac + Windows)Street price: “…it’s complicated.”

Yeah, I gotta go with Avid Media Composer. This is the default workhorse tool of television and film. Its long evolution has been torturous and expensive for video editors. It has been agonizing and exasperating for Avid’s internal developers. The company has edged near death more than once. But the software is now mature, solid, stable and (dare I say) a little sexy in its own stodgy way.

With your purchase of Media Composer you, today, stand tall upon a huge pile of mangled souls—bruised professional users and wounded former Avid employees—who have fought for a quarter-century (often against one another) to fix bugs and forge the ragged frontier of non-linear editing. You benefit from their years of toil and frustration. Shed a tear. Say a prayer. Salute them.

Now go edit your little tush off. Because, with Media Composer, there's hardly anything you can't do.

But sometimes Media Composer’s way of getting you there is Byzantine. It takes a while to turn a massive ship at sea. Avid has been lagging behind smaller, more nimble companies in several innovations. There’s no lack of engineering acumen at Avid; it’s more about the need to ensure any new feature is compatible with a much larger user-base than most competitors. When those customers are enterprise-class pros (most Avid users are), there’s great demand for stable performance on existing (“sunk cost”) hardware.

Avid’s competitors were quicker to let you work directly with native media formats. Avid began to address this with Media Composer 7’s Avid Media Access (AMA) protocol. But AMA’s implementation wandered down bit of a bumpy road. In Media Composer 8, AMA percolates mixed resolutions much more smoothly.

You can drag and drop media files into a bin—which initiates an Import, transcoding and cloning as needed. Holding down modifier keys (Alt on Windows/Option on Mac) instead conjures an AMA link, essentially creating an Alias in your bin and accessing the media without importing. You can color-code such clips, so a glance at your timeline will tell you what’s native in the session and what is linked. All of this “AMA-ing” (yes, it’s now a verb) can speed your workflow, as long as the needed media drives are connected during your final Export.

Avid now also lets you transcode in the deep background, meaning you can set a list of mixed-media clips to wash, dry and fold themselves any time the CPU is active, even if you quit Media Composer entirely. It reminds me of the old school days of batch-process code compiling, but now with terabytes of 4K, 3D and HDR.

But the speed and crashworthiness of your workflow is still inversely proportional to the amount of transcoding needed. If you can set your project up to ingest a single format, Avid will fly like the industrial strength power plant you have a right to expect. Mixing resolutions puts a tax on your hardware; usually not a big deal, but it can get seriously squishy if you’re on an older CPU.

One great reason to work with 2K or 4K sources is Media Composer’s ability to re-frame (crop) radically in the edit, while still maintaining full 1080p HD resolution. It’s called Frame Flex. It saves you save time in the shoot and preserves your creative options because you can do your critical framing in post. If an actor threw down an inspired (but very subtle) expression, you can get the close-up needed to sell it. Not to mention the product shot for the agency producer to whom no brand mark is ever big enough…

But let's say you want to stay in 4K or UHD all the way. The v.8.3.1 release of Media Composer will support you. You can either work with these super-heavy media directly on the timeline, or cut using proxy-files and only call the native camera files on export. Newer cameras with huge sensors and high frame-rates make fat files. A RED 4K cranking at 59.94p will spit uncompressed video at more than 700MB per second. Even with ProRes compression, your NLE is fielding a stream of over 125MB/sec. if you have enough hardware horsepower, Media Composer will cut it.

Directors of Photography are staring to work with Look Up Tables (LUTs) right there on-set. This lets them capture full-range (but flat looking) files, yet see what proper profile color grading will do to them in post. Thus, the DP, the Director, the Producer, the Talent can all see the shot in the finished look of the film right there, without having to commit to anything.

Editors develop their own style of cutting. Like playing a musical instrument, pros get their best performances when wailing on their own axes. Media Composer’s highly customizable User Profiles feature lets you set up your bench the way you need it. And you can take it with you onto any Media composer system. I can’t stress how important this is for those who do this day-in, day-out. When it’s right, the interface melts away and you feel like you’re sculpting movies with your fingers around the images, like the best moments of cooking, molding modeling clay and making love, all rolled together. Well, something like that anyway.

In days gone by, a complex Media Composer session looked like a train wreck of overlapping Bin windows and Tool palettes. Since v.6, the software has gone to a tabbed navigation look that allows you to step through bins with drop-down mini-menus. You may also nest folders hierarchically inside of Bins. Doing so makes a big session manageable, even on a laptop’s limited screen real estate.

Looking at a Bin in Frame View lights up your visual cognition circuits. Thumbnails of clips appear. You can play each clip right in the window. And, yes, you can set the thumbnail frame. The flow of the show immediately starts coming together in your mind. You can drag clips around to create a quick and dirty Storyboard, which is handy for getting approvals from bosses above or for assigning “scut-work” to colleagues below.

Switching that same Bin to Script View lets you type or paste as much text as you want into an annotation bar with each clip. Dramatic script, documentary transcript, take-selection, director’s notes; whatever you need to see the narrative flow of a great show take shape before you.

Audio in Media Composer is taking a ridiculously long time to become fully integrated with ProTools, Avid’s industry standard audio DAW driver. A bit of company history perhaps explains this. Avid (East Coast) acquired the Digidesign (West Coast)—creator of ProTools—two decades ago. One was sample-based, the other frame-based. Culturally, Avid has long been a house divided.

But the matrimony is finally flowing. The mixer in Media Composer looks and feels like a toddler-level ProTools. The ability to set audio Clip Gain right on the time line, event by event, grew into Media Composer (and Avid’s ProTools 11). This is handy, letting you optimize each bit of sound as you work, building your mix as you go, without writing fader automation. And Media Composer has finally added a Master Fader to its mixer, at last replicating the most obvious feature of all analog audio consoles since the 1930s. Better late than never, Avid.

You can drop RTAS plug-ins across Avid tracks, but you may have to render them out to hear them. And sometimes they will invisibly hang; if your ears aren’t wary, you’ll miss the fail. 

It took Avid coders many years, but they’ve finally got over acting like stuck-up, self-centered computer nerds and learned to speak some actual television. And film. Along the way, the Avid product marketing team realized how bright their user-base actually was. Avid unbundled its systems (though way too late) to let users jockey their own hardware configs.

Font-work, for a long time, was a no-win choice of the puerile Avid Title Tool or the quirky Avid Marquee Tool. But now Avid FX is ratcheting up to its competitors' market level in the craft of text-over-video. Avid has also opened itself to the third-party specialist world; New Blue’s Titler Pro and VisTitle (Windows-only) are the go-to choices. And, of course, you can wormhole your way out of the Avid universe to the venerable Adobe After Effects. If you need custom 3D, you’ll need to bounce outside of Avid (into something like Blender), then import your final result.

Media Composer is industrial strength software, battle tested in high-pressure broadcast shops and tweaked in response to finicky film directors. Along the journey, Avid developed powerful metadata search tools to help you organize your artistic endeavors and access your assets amid your growing library.

Pushed by Adobe into the realm of subscription software, Avid now offers Media Composer for $49.99/month. Or you can buy it. If you already own it (I do) you can "lock in your perpetual license for $299.00 and get access to every software update plus standard Avid support for the next year—or subscribe for $39.99 per month for the first year." A cynical unpacking of the qualifiers and hidden gotchas behind that bit of marketing-speak suggests Avid would really, really like you to slip from privileged owner to addicted renter status, though it doesn't have the fortitude to insist upon your servitude. At least not just yet.

A less cynical perspective on Avid’s pricing reveals that it lets you grow or shrink your commitment to Media Composer, based on the changing needs of your enterprise. If you suddenly expand to need, say, 20 editors to cut your reality show footage, you’ll want to transition to a floating server license deal. If you are building down after a long-running production has sunset, Avid will step down with you to manage your sunk costs. 

More pertinent to most of us, the Avid subscription model will likely reduce the cycle time between upgrades and bug fixes. If you own Media Composer now, you should spend the $299 in 2014 (or by the end of your current support plan). If not, you will probably find yourself re-buying the whole package next year (it’s $1299 now).

Of course, there’s nothing wrong with freezing your platform at its current level of capability and running it until you can no longer replace the hardware or patch the software. You’d be driving down a one-way street into an evolutionary cul-de-sac. But not everybody likes to live downtown.

Avid support is—and has always been—surprisingly annoying. Though you may pay into the (exorbitant) “Support” assurance plan, prepare yourself for the bum’s rush if you are not a broadcast network or large post-production house.

I recently had the experience of waiting 28 minutes for an Avid support rep to answer his phone. Then I got to watch him remote-access my desktop and witness my issue first-hand, only to have him send me e-mail four days later demanding that I take screenshots of the problem he saw in real-time. And I’m still waiting for the fix. Unconscionable!

In what is perhaps an effort to act with more egalitarian morality, Avid is apparently pivoting to a “Googlesque” model of customer self-help. But Avid lacks Google’s deep pockets to resource the development of the needed online infrastructure. Until it gets there, if you want to solve a problem related to Avid’s architecture, you pay them to fix it.

We are focused on editing tools in this article. But know that Avid is a strong player in media management, enterprise-class workflow, distribution and play-out technologies; the stuff that comes before, behind and subsequent to your picture cutting wizardry.

If you are delivering to television, you won’t find better export compatibility. Right now, chances are your export will be transcoded at least one more time for play-out to air (by something like Grass Valley’s STRATUS cloud). But it is theoretically possible to play out of Avid, direct-to-air, via Avid’s Unity ISIS (Infinitely Scalable Intelligent Storage) system. If, say, you were starting your own news channel, you’d want to digitize once—at the camera—and be able to whip that asset to air, live or on-demand, with added value or simply raw. You could do it all in Avid-land.

As the media world spins into cloud creation, no work of art is ever truly new or wholly finished. But Avid can be there, powering your process, at every step past the lens.

Adobe Premiere Pro Creative Cloud

Adobe Premiere Pro Creative Cloud

(Mac + Windows)Street price: ”Confusing…”

Global warming is causing faster uptake of worldwide water into cloud-mass. So, too, has the suck-rate of video tools into cloud computing increased. And, as you might expect with such a major shift of basic data crunching, some unfortunate side effects come raining down in bug form.

“Pioneers get the arrows,” it is said. Adobe pushed deep into the frontier of “computing as a service” when it ported Premiere, along with its software siblings from Creative Suite, into the Creative Cloud. If you had a license to run CS, you can start shelling out $49.99 per month to pull updates from CC. If you’re a new customer, pay $69.99 per month or $839.88 per year “prepaid.”

Internet-resident software is seductive to companies. In white hat mode, they can offer seamless, rapid upgrades and fixes to their consumers. With their black hats on (or, at least, with their green eyeshades), they can shakedown addicted patrons (you), extracting an ascending rate structure of fees at will.

Adobe has been far more benevolent than predatory in its execution of Premiere Pro CC. Cruise the various fora. You’ll see very few horror stories, with most incidents exemplary of good-natured bungling rather than covert evil-doing.

Upon launching Creative Cloud, a few things immediately broke for many users. Examples include multi-cam editing, gamma curve replication, addressing older video hardware and others. But, to the company’s credit, Adobe seems to have addressed these and most similar issues quickly, if not completely.

As a mostly Media Composer editor, I love that Premiere Pro can import Avid timelines. You may have to monkey around with re-linking when you restart before all of your imported Avid clips find their way through the woods. But it works!

The “great attractor” of Adobe, for visual artists, is that close family of tools that is Adobe Creative Suite/Cloud. Dynamic Linking is how you pass your works of art between Premiere Pro other Creative Cloud toolkits, particularly After Effects. This teaming effectively replicates on your laptop what a half-million-dollar roomful of Silicon Graphics/Flame workstation could do a decade and a half ago. Dynamic Linking might be the single biggest raison d'être to go Adobe Premiere Pro over Avid’s Media Composer.

Color grading in Pro, on its own, is neither brilliant nor particularly subtle. But dynamically link your project out to SpeedGrade and you can paint with cunning delicacy or wild abandon. SpeedGrade’s interface is a hierarchical tree of color wheels. If you grew up on curve bending, this will take some getting used to. But it is enormously powerful. Adobe recently wired it so you can tweak your master clips. Of course, you can also venture outside the Creative Cloud to the black magic of DaVinci Resolve for extremely critical tint-work (for fussy clients). But to shade and grade fast, right on the timeline, many pros turn to Red Giant’s Colorista.

Adobe speaks metadata fluently. Type a search term into Pro’s project panel and assets dutifully present themselves to you. Premiere’s Speech Search can associate spoken dialog with text, for those increasingly rare projects that are actually scripted. In fact, you can mark/trim clips based on spoken words right in the Speech Analysis pane (very cool for documentaries). Face Detection can navigate you to recurring characters in your footage (handy for entertainment shows).

As good as its metadata manipulations may be, Premiere Pro seems to lose track of assets more often than Media Composer. On the other hand, Adobe’s reconnection tools work faster than Avid’s re-linking functions.

Are you Transitional? I am, with respect to video. Yes, the “cut” is still the best and most elegant from of video transition. But—when playing in appropriate visual-music languages—I do use wipes, generally softening edges quite a bit and producing a “shaped-dissolve” to direct the eye. In Premiere Pro CC 2014, Adobe kills much of its formerly rich palette of transition effects. Yes, a case can be made that any such contrivance is better constructed in After Effects and imported. Still, it was nice to do it quickly in the editor.

More than ever before, it’s become important for each of your creations to have a consistent and topically motivated “look.” Pro’s Adjustment Layer permits you to audition and apply global treatments to stacks of tracks, without having to nest them or composite them. And, importantly, you can tweak the effects while loop-playing sections or the whole. It’s a powerful hack for subtle manipulations (or rapacious violations) of your cut footage. With sufficient hardware, you can watch the changes in real time without rendering.

Premiere leverages your GPU to accelerate rendering, particularly of special effects. It brands this bit of kit as the Mercury Playback Engine. How much load Mercury hands your GPU depends on your graphics board. Adobe first began to task Nvidia GPUs with the release of CS5 (through the CUDA API). With CS6, Adobe went to OpenCL, which can speak to a select set of AMD GPUs. The approach continues under Creative Cloud. Driving the GPU doesn’t accelerate your encoding/ingesting, but it can speed up certain tasks. De-interlacing, effect-renders, some blurs and chores like scaling come out faster/better for having been offloaded from the CPU. For really intense color grading and animation rendering, think strongly about doubling up your GPU. Yes, play two-card Monte with your extra slots. See this excellent account of problem solving with “Dual ‘CUDA’s” by Ryan Ashford working at Matell (sign in through LinkedIn).

Over in the audio domain, Premiere Pro tries to keep you out of trouble. Drag a sound byte or music cue into the timeline and Pro spawns a new track for it. But the software tends not to use standard audio-engineer speak. Adobe can run the wide creed of VST audio effects and modifier plug-ins, but not universally so. From your colleagues, you will hear tales of inexplicable errors (especially on Windows platforms).

Premiere Pro’s interface is easy to operate. The “panels” (work windows) size and tile smoothly with a nice stick function. Your workspace always looks organized and runs intuitively.

I certainly like Premiere Pro’s slip and slide timeline sizing. It’s very easy to zip around and down into my cut—even with a very long-form show. Perfect for that phase of the edit when you have a handful of those “gotta-use-it” shots and are searching for the exact right places to surface them.

Dynamic trimming is another workhorse technique made easy in Pro. This lets you shift edit points on the fly, while watching a loop centered on whatever you’ve set as your default transition point. There's no better way to feel the visual rhythm of your cut.

Like many actions in Pro (and other top-line NLEs), there are keyboard commands well worth learning for speedy workflow. You can also key in values for Numeric Editing. I tend to think musically when editing. Once having defined a beat by frame count, I like to edit on multiples of that frame numeral—like beats in bars of music.

Another sort of bar, one that is set entirely too low across the industry, is customer service. As with Avid, Adobe seems to be forcing customers into the Google model of self-help. It works for Google because its product is entirely on-line, Google is wealthy and Google is a virtual monopoly. Adobe has competitors, but it is large and deeply infused into its markets. It could set the industry bar for customer success much higher if it stepped up. Are you listening, San Jose?

Yes, it’s completely appropriate for Adobe to present online customers a tree of “Top Issues.” But, if your issue isn’t among the handful of common questions, Adobe’s escalation path resembles corporate IT: “Open a ticket and pray.” Also, Adobe’s support is heavily weighted toward selling you more Adobe products. Don’t get me wrong, I like Adobe products. I buy Adobe products. But to slip barefaced marketing under the skirt of support is a perversion of the customer’s expectation.

There has also been a lot of recent user pushback about Adobe new site design. It is clearly optimized for mobile platforms. But is that really a good thing? Or just a trying to be chill thing? How often do you actually edit video on your phone? Many pro users are on workstations, or at least laptops, and do not appreciate this.

Additionally, Adobe’s internal forum functionality attracts a lot of negative critique. The fora seem curated by bot, which is to say: not. On the other hand, there’s a superb one on LinkedIn. If you like to read and are the motivated introvert type (self-starting hermit, like me) you can use Adobe’s excellent tutorial library to totally self-educate. But it’s not like having a knowledgeable someone at the other end of the phone…

EditShare Lightworks PRO 11.5

EditShare Lightworks PRO 11.5

(Windows + Linux; Mac version in “Public Beta” at the time of this writing)Street price: ~$60/year + some one-time codec charges

I envy feature film and documentary editors. They get to sink their creative teeth into deep projects. Most of the time, they have the luxury of concentration on subtlety. They are tasked with striving for elegance. They can worry sequences for weeks. They can learn their subjects’ rhythms. They can inhabit and marinate in a project’s culture. They can be artists, not hacks like me.

Lightworks PRO, long a filmmaker’s friend, is similarly subtle and elegant and beautifully organized. It’s just exquisite. It’s inclusive. And it’s inexpensive. I love it.

But I can’t use it.

I inhabit a world of assembly-line editorial. Short-form projects pass through us like sorority sisters at a frat party. Barely time to dance and do the deed, then on to the next romp. So demands the short attention span mass media market. In our case, it’s a Media Composer/ProTools planet.

Lightworks excels at long-form. It is simple, straightforward and intuitive. Yet it also rocks very advanced trim types. PRO is also quite fast and facile in Multicam mode: As many sources of as many formats as you could want. You can bang it like a live switcher, then step in and refine a cut as needed.

Like blocking a film, Lightworks PRO's Storyboard mode lets you push clip-tiles around to string out scenes. You can trim each clip directly. Then flip the whole progression to an edit sequence with a tap of your finger.

Lightworks’ Project Browser is intentionally minimalist. No menus. No nav bar. No “Save!” It stores everything you do, with apparently limitless Undo/Redo. There's just a toolbar to steer your edit by, like Lightworks doesn’t want you to be distracted by anything other than your footage. Simple export or play-out to a real-time recorder. It’s all very Zen.

Audio can follow video as you cut. Drag and drop shots to the timeline and their sound will follow—or not, if you prefer—with PRO’s Audio Linking function.

Beyond its filmic heritage, Lightworks PRO has bright potential in the open-source future, where footage can originate anywhere, shot by anyone, on every format. Lightworks will suck in almost anything (15 container types and growing), at any resolution, and cut it—without transcoding—right on the timeline.

Color timing is solid, addressing your GPU’s horsepower for fast realization. Lightworks recently signed on to the bandwagon of presets for various looks. Expect a growing catalog of these, but it’s not fully there yet. If you’re the type who shades by warping a curve in color-space, this latest version of Lightworks now supports you. But for special effects work, you will need to push your sequence beyond Lightworks’ borders. If, say, you’re conforming an entire movie to an unusual chromatic universe (think Gattaca, for example), Lightworks looks like it expects you to send your digital inter-negative out to a colorist for final finishing.

Tracing back to its Steenbeck film heritage, Lightworks offers a special purpose peripheral console modeled after flatbed ergonomics. It puts hardware function buttons and a jog-shuttle/scrubber wheel under your hand. But prepare for sticker-shock: at $2800, it’s for true professionals only. Lightworks will sell you its color-coded, function icon-adorned USB Keyboard for $145. If you’re a hunt-and-peck typist (like me), it’s a good value. Other brands—Avid MC, Adobe Premiere, Apple FCX, Grass EDIUS and Sony Vegas—rely on LogicKeyboard to supply dedicated pianos.

Lightworks Pro is only available on a yearly license basis ($60). Some special codecs—Avid’s DNxHD, for example—attract a one-time fee.

Sony Vegas Pro 13

Sony Vegas Pro 13

Street price: $400 for video + audio only / $550 with Blu-ray authoring  / $650 for “Suite”, which adds Sound Forge

The problem I have with Vegas (the town) is the prevalence of truly deadbeat, destitute gambling addicts haunting the slots when they should be home sleeping it off. Or, better yet, out contributing to society.

The problem I have with gambling on Sony Vegas Pro software is that it’s like driving up a dead-end. You invest serious time away from your loved ones on this platform, only to find yourself addicted—like a wanton wagerer—to a tool that has a questionable future. It can only do so much. That much is not quite enough for most users amidst today’s confusing convergence. And Sony’s track record of timely development is not encouraging.

And yet, if you like a plain vanilla workspace, have a pile of different camera formats to cut together and do Windows, Vegas might seduce you.

Now, if you are a big show, shooting a lot of cameras in XDCAM format, you might even be in for a Vegas wedding. Here’s why: you can upload low-res proxies to your private server, moments after the shots are shot, wirelessly (special hardware required). That lets your loggers and editors start working immediately. Your team can also log and annotate footage on the Vegas Pro Connect iPad app. These notes show up on the Vegas Pro timeline. Your team will conform the full-res footage later. It’s a blast from the production past, but it works. It’s just hard to see anything less than a small army needing this arcane workflow.

But even small projects these days seem to attract a swarm of media formats from a candy store of cameras: REDs, GoPros, DSLRs, mirror-less minis, smarter phones, digital cinema cams. The good news is Vegas Pro hosts most of these with ease on its ecumenical timeline.

And Vegas edits 3D stereo with easy auto-pairing in the timeline. This may be the last, best hope for this package: Sony introduced—then discontinued—its professional 3D XDCAM (TD300) 3D video camera. The company seems to have retreated to its consumer 3D Handycam. Perhaps it'd consider a bundling deal which more appropriately prices Vegas Pro?

Native effects and filters in Vegas Pro are limited. But Sony committed to speaking OpenFX, the protocol running under plug-ins from several market leaders. Vegas Pro is bundled with plugs and presets from NewBlueFX.

Sony headed down the path of companion programs—notably Sound Forge, for audio sweetening and sound design—but got lost in the desert.

Vegas Pro will, however, pick-up projects started in other editors. And it exports to a few as well. Adobe’s Premiere Pro and FCPX (via XML) are surprisingly well supported.

If you are familiar with another major NLE—and you want to take the time—you can re-map your keyboard to put your most often used commands where your muscle memory believes them to be. But since that other toolkit is undoubtedly deeper and wider than Vegas, why would you bother?

Despite the “Pro” in Sony Vegas, this toolset seems caught below the threshold of what most professionals now need, and it sells above a price -point where it can compete with amateur/hobbyist products like those reviewed in part two of our series.

Grass Valley (Belden) EDIUS Pro/Elite7

Grass Valley (Belden) EDIUS Pro/Elite7

(Windows 7 [64-bit])Street price: $700

As Apollo 8 sailed around the Moon in the tumultuous year of 1968, a small California company brought to market its first video switcher. And for the next three decades, Grass Valley more or less dominated the workflow of weaving linear video streams, whether live, off tape, from film-chains or in combination.

As that icy jerk Grand Moff Tarkin orders the destruction of Alderaan by the Death Star’s correlated-beam weapon, it is the T-handle and color-lit buttons upon the dusty deck of a Grass Valley 1600 “vision mixer” that is propped up to do the killing.

As Grass Valley grew in California, another small company named Canopus launched in Kobe, Japan in 1983. It was an early maker of video cards and soon saw the need for simple editing software to go with them.

Video editing passed through a bridge period during the 1990s; not yet non-linear, but not quite offline. And Grass Valley was there. We Cro-Magnons (Homo editus) would ingest low-res versions of clips into Avids or other platforms to make our cut decisions and plan effects. But all they spit out was an Edit Decision List (EDL), to be brought into a tape-based edit suite and “conformed.” Wipes, picture insets, composites, graphics and fonts were brought to the party through the mix effects amplifiers of Grass (and other) boards.

In 2002, Grass Valley Group was bought by Thomson Multimedia of France (later to be renamed Technicolor); in 2005 Thomson acquired Canopus, bringing a non-linear editor to the grand Grass hardware heritage. And EDIUS was born.

It is at home in broadcast and high-end post-production environments. When integrated with Grass Valley’s STRATUS media management tools, EDIUS probably offers the fastest inter-format editing solution around.

There have never been more different video formats than right now. EDIUS excels at scaling, sizing and cropping. This includes working at 4K. When they say “Edit Anything,” they mean it.

The simplicity of the interface is another plus for EDIUS. In a duel between EDIUS and Lightworks for which boasts the least complicated pro-level NLE, it’d be hard to pick a winner. Its ergonomics are intuitive.  

Similarly, EDIUS has one of the most straightforward multi-cam interfaces around, likely another outflow of the Grass Valley switcher birthright. But it takes things literally. If you try to edit very high-res multi-cam (say, four or five 4K sources), EDIUS will grind slower than Treebeard the Ent on Thorazine on all but the most hot-rodded hardware. The workaround is to ask EDIUS to build proxy clips—which will drop frames—and go have lunch while it does them. But don’t make it a liquid lunch because when you come back, EDIUS will be ready to cut cameras as quick as you want. Just be sure to check that the proxy parade has fully passed in the Background Jobs pane.

Ironically, in light of the Canopus DNA as a video card maker, EDIUS is building its bridge to the future by friending external hardware makers: Matrox (long a card-resident graphics slinger), Blackmagic Design (the Han Solo of the video business) and, just recently, AJA (ironically located in Grass Valley, CA).

EDIUS still retains flavors of the linear and hardware workflows of the past. Not at all a bad thing. For example, today’s EDIUS can still pass an EDL around, notably to Blackmagic’s DaVinci Resolve color grader. But its vibe can seem stodgy to those born on the right side of the digital divide.

Stepping up from EDIUS Pro to Elite yields only a few more tricks, and they are very specific to the professional broadcast setting. You can pass and edit multiplexed (encoded) Dolby AC3 audio and you can network via STRATUS for workflow and play-out to air.

Final Cut Pro X

Final Cut Pro X

(Mac OS)Street price: ~$300

There’s a lot to love—and hate—about FCPX. Many have weighed in on this fight. I’ll try not to stir it all up again. Let me get the bad stuff out of the way first. Apple fanatics (I was one once), skip down. Or bear with a few paragraphs of bruising critique.

Final Cut could have had it all. A decade ago, we caught a glimpse of a synchronous, end-to-end, creation-to-distribution audio-visual solution coming from the inscrutable Apple. Create in Final Cut. Sweeten audio in Logic. Author in DVD Studio. Version in Compressor. Distribute through iTunes…

Now, there’s a train wreck where a shining shaft of crystalline independent video genius once shimmered. Can FCPX make it back to truly “Pro?” Or does Apple so believe its own rhetoric of media democratization that it intends to try and kill the professional video editorial class entirely?

Yogi Berra famously malapropped: “When you come to a fork in the road, take it.” Apple tried precisely that, careening through a sharp turn to the south, from Final Cut Pro 7 off into the weeds of FCPX. The two tools are, at root, fundamentally different architectures with respect to picture flow.

Thus, project migration from the legacy to the offspring was thrown off the table and under the bus. It was a brave choice. But it gives lie to the moniker of “Pro.” To the professional editor or producer, this looks like archetypal Apple arrogance: “Where we lead, you shall follow.”

No professional would chuck her entire toolbox on the promise of a new holy-land, especially one whose landscape is suddenly barren of features she depended upon last week. No video client wants to pay a latter-day Hernán Cortés to mellifluously and melodiously incite their passion for adventure while burning their editorial boats behind them. They just want their freakin’ show cut, finely and finally. That’s what Final Cut was supposed to do.

It willdo so, if your show is brand new and you want to learn a brand new interface. If you need to recover old FCP projects, your only path leads through third-party translation-ware (Intelligent Assistance’s 7toX plug).

There’s also a bottleneck at the exit sign. Apple doesn’t call your attention to this, but FCPX lives by the maxim: “Lascucarachas entran, pero nopueden salir.” The program is a diode. Works beautifully, if you stay forever in the House of Apple. But the house rules are very Hotel California. If you need to export an incomplete sequence to another platform, your options are limited.

But it’s not impossible to transfer. If you are going out to the industry-leading Avid, Boris FX does make a pair of plug-ins called Transfer FCP that can short out the diode. The outbound path creates an AAF file-set that Avid understands. One especially good reason to do this is to finish audio on Avid ProTools.

You can get some of the elements from an FCPX project out to After Effects. XML transfers used to work going into Premiere Pro from FCP7, but FCPX is another story.

On its initial product launch, FCPX ran aground on the sharp rocks of professional editorial opinion. Broadcast cutters were suddenly left high and dry without multi-cam and up the creek without their color-calibrated monitors. Apple seems to have heard the howls and rolled out patches. But I worry about FCPX’s future.

Given Apple’s laser-focus (and huge success) in selling consumer mobile devices, I’m hard-pressed to believe the company will invest engineering and support commitment to a declining number of high-end craftspeople (you and me). I imagine a sales data-driven discussion, behind the Iron Curtain in Cupertino, around how many more “dark tower” Mac Pros would be sold if Final Cut is allowed to survive. The answer, sadly, will be only a comparative handful.

I hope Tim and Jony prove me totally wrong, but I worry Final Cut may finally get cut.

Enough angst against Apple. The bright side of Final Cut Pro X is brilliant. It’s a sleek and modern editor that has made a clean break with most of the baggage of its past.

If Lightworks was considered the genius at rendering in the background, Final Cut had traditionally been the dolt. Up through version 7, FCP was cold-molasses sluggish. It needed to stop, then think about rendering. Then render. Then stop again. The render some more. Often. Maybe the marketing team shrewdly figured that long, frequent renders stimulate long side trips to the App Store.

Not. Any. More.

FCPX is screamin’ fast by comparison, especially if you can afford ~$5000 to rock the home-court advantage of a six-core “New” Mac Pro with its dual GPUs. Now running in fully native 64-bit, FCPX operation rams right through the former 4GB RAM barrier. Apple boasts of rolling up to 16 streams of ProRes 4K. Complex composites, interpolated slo-mo, long transcodes and high-res exports are now much faster. Many former “stop-and-wait” operations now tick away in the background with a completion-predictor display happily keeping you company as you merrily edit other bits.

Outside the box, Thunderbolt will connect you with AJA, Matrox, Blackmagic and other brands of I/O, plus some newer cameras, at transfer rates that can only be described as sick. Native/proprietary formats from superstar cameras like RED and ARRI can be sucked in with ease.

A single window now lords over all manner of imports from a large universe of source types. It preserves most metadata, including rich XML wrappers. You can work in your source’s native format or ask FCPX to create ProRes (Apple’s own). FCPX’s Content Auto-Analysis works as your AI assistant to identify media types, separate shots and flag that there’s a human in the frame. It can stabilize footage, if you wish. And it can find and fix certain audio issues like 60-cycle hum as it’s importing. The AI can look at overall color balance with an eye toward globally shading shots by source, which is handy when, say, your white balances differ from cam to cam. Smart Collections, basic keyword tagging and other librarian tools can make organizing your media more fun (was that task ever actually fun?).

So now it’s time to edit. In its re-imagining of video editorial workflow, FCPX did away with the time-honored notion of tracks in favor of its spooky Magnetic Timeline. It hasn’t really killed the track concept, just made it more fluid. They’re now called “lanes.” More like driving lanes than swimming lanes though: clips snap together or bump one another along, depending on where you hover your driving cursor. Why should you have to bog down your creative flow keeping track of the relationships between clips when a bit of machine smarts can do it for you?

There’s still a timeline, though. And familiar source versus program monitor panes. If you can get past the ego-challenging feeling that you're editing in some iMovie-on-steroids toy, you can edit like a pro: slipping and sliding and rolling and rippling.

Confound it, compound it! OK, so you’ve got a complex multi-layered sequence built, tweaked and happy. But now you need to drag it elsewhere. Or use it again. This used to be a migraine waiting to happen. Now FCPX let’s you “compound” combinations of clips to easily relocate them. Whatever else Apple may have screwed up on the jump from Final Cut Pro to X, this it executed eloquently.

Likewise color-keying. Color Selection in FCPX refines what is possible, letting you create conditions in your key that compensate for imperfections in your shot with a great number of handles for problems with edges, shadows, spills, rim-light bloom, highlights and other bugaboos.

A lot of pros bemoan the now-dead Apple Color tool. But a few editors I know love FCPX’s color correction. Part of their ardor is that the waveform monitor and vectorscope now actually function (unlike those in Final Cut 7 and earlier), which is to say each has finally attained a level of resolution that can accurately model the ballistics of analog.

Trouble is, you cannot bring up more than one scope at a time. If you want to, say, saturate your chroma but take care that doing so doesn’t push your white-point beyond broadcast legal, you need to click back and forth repeatedly while your footage plays. All the while, you’re trying to integrate what each scope — plus the program monitor — is telling you. Pass the Fioricet, please.

Setting that weakness aside, the correction palette itself is a powerhouse of primary and secondary strength. FCPX now starts you off with a “Looks” effect category. Every editor on every show is always in search of a unique vibe that puts gentle bumpers around the audience’s range of emotional response. Color-casting is one way to do that: cold looks, warm looks, duo-tones, distressed, hyped, historic, futuristic, day-for-night and more, FCPX starts you down many paths. Each preset gives you maybe 70 percent of the shading and grading in a millisecond. It’s rather like the visual version of an audio producer choosing a pre-convolved ambient/reverberant space for her music track. Match Looks will let you sample the look of a clip, and then impart that vibe to any or all clips in your project.

When you’ve locked your look, FCPX can render the corrections in the background. Where Final Cut formerly failed it now succeeds brilliantly.

It's wonderful — and terrifying — how the human hardware universe can be radically rolled by the dedicated actions of a single one of us.

Please permit me an homage at this point: down deep in Final Cut’s DNA, there are the marks of common ancestry — or at least co-evolution — with Adobe Premiere. No discussion of non-linear video editors can be complete without a reverent nod to one of the field’s principal pioneers: Randy Ubillos. This artful technologist essentially invented Adobe Premiere (originally called “Reeltime”), and then led the team that grew it through its v.4 release. That team found another home at Macromedia where they assembled KeyGrip, which, upon acquisition by Apple and further development, released as Final Cut in 1998, which, when buckled to IEEE 1394 (FireWire) digital video I/O turned “Pro” in 1999. Then Aperture. And next Randy’s posse created iMovie.

Conclusion

The Moral Here

So, as you go forth into the valley of the shadow of looming video deadline, here’s what I’d like you to take with you:

It’s not about which video editing software is “the best.” It’s about which one fits your hand, your eye, your sense of timing and your way of organizing. Factor in your prior experiences, plus the learning curve to get comfortable with a new toolset. If you’ve invested, say, more than hundred hours on any decent platform, you’re probably best off staying there.

At least, until the brand dies and you’re forced to learn to worship in a new church.

But don’t underestimate your power to change the prescribed canons of worship or the dogma of editing practice. It is in the nature of NLE to be always either in-development or dying. A lot of what the manufacturers try to hand you—bad support, buggy whiz-bangs, half-baked features—is unacceptable. Say so. Don’t let them BS you. Make them work harder for your dollar. Always remember that you are the apex predator of this biosphere.

Where No Editor Has Gone Before: Into The Clouds

Advanced video editing hereafter is all about the cloud. Or, more accurately, clouds plural.

But this does not mean that you will be actually editing in the cloud just yet. Yes, with ubiquitous big-pipe broadband, you could someday be editing a proxy on your tablet from your kayak or igloo. Your actual 4K files will be screaming around Amazon’s data farm. And your rendered Final will directly play-out to viewers from your client’s head-end—or your YouTube channel—without you even having to export: truly on-line editing. But that day shall not come as soon as Avid and Adobe want you to believe.

Between then and now you will essentially be renting your editing tools from a cloud-based mother ship. But you’ll be carving with them locally on your home-heating, utility bill-killing CPU; all your pictures stored on local drives, with only yourself to blame when the array fails with no back-up. In this brave new world, the worker (you) controls the means of production. But you won’t ever own them.

Yes, it’s the old Soviet model, complete with a small committee of corrupt fat cat execs cashing your checks, surrounded by a cloud of dependent, sycophantic and largely incompetent apparatchiks (called customer service representatives). They will eavesdrop and surreptitiously record your forum chats. Complain too much and they will fire you, even though you are the customer. Mark my words, folks, this will suck. But it's the new normal. For now.

I know. You hate having to pay the lease each month for a car you may not drive that often, or that far. But, having spent the last few thousand words beating up the manufacturers, I’m now going to ask you think about this from their perspective.

For software shops to remain competitive, they must be nimble, delivering updates and refinements and bug fixes as soon as they can. Going, going, gone is the business model of selling an expensive box of video toolkit but having to constantly offer patches and updates for free.

So you pay a modest—but constant—fee to edit. And you get the latest toolset, updated and supported automatically in the background (mostly). Adobe Premiere Pro Creative Cloud is already here. Avid Everywhere is gathering on the horizon. You’ll find yourself monogamously married to one mob or another. You can pick your gang, but you “can’t not” be a gang-banger. For you have drunk the Kool-Aid and subscribed.

Here's how the seduction is structured: you date a non-linear editor because you like the look and feel of its timeline and sculpting tools. But you soon find that some other guy's compositing program is better than the one offered by the NLE you chose. It proves to be damnably difficult to move all your elements and half-realized cut cross-platform. And then you find you need to create and import an object (or worse, a scene full of them) from a 3D modeler from yet another company. You have a client—for whom you want to do award-winning work—breathing heavily in your ear about the budget and the deadline. You're basically hosed.

If you have to make a living doing this, you simply don't have time to become a fully vetted proficient citizen of many different townships, each with their own customs, currency, interface, terminology and keyboard shortcuts. You're hosed again.

Thus, the clusters of "intra-networked" programs become island universes. You could think of them as navies at war on the sea of visual content, or like different planets with mutually exclusive biota. But they are really much more like Organized Religions.

Thus you choose to marry within one faith or another. And you stay there. For years. And your working culture becomes ever-entwined with that brand.

So, when selecting a video edit package, be sure to pick one whose high priests return your e-mails.

How to learn the prayers and rituals of any new software package? Start with: Lynda.com. Lynda Weinman began as a special effects animator before founding the training company with her partner illustrator/designer Bruce Heavin. They get visual production. They know how to school us.

Epilogue

Better days lie ahead. I’m not yet a believer in all that Singularity stuff (Nerd Nirvana). But, looking at innovations like Apple’s Content Auto-Analysis, it’s pretty clear that fuzzy-logical neural-networked contextual software will save our sorry video editorial butts, someday.

Full interoperability between tool-sets from different creators (religions) can be made possible given sufficient cloud computing power and connectivity. We will eventually see (mostly) seamless “translation-in-the-background.” This will render the platform pedigree irrelevant. The economic winners: the third-party plug-in makers. And you and I, the content creators.

Yes, you could edit and deliver from your catamaran. You may even be able to speak to your Editing AI in esthetic terms: “Move that swarm of bees (cluster of related procedurally animated object-events) from near the diving board (locus by context) to over by the guy with the watermelon (face + object recognition and key-frame definition), in one and a half seconds (transition time irrespective of frame-rate); let me see a gentle swoop (3D Bezier motion path).”

You will finally graduate from button-pusher to full-time creative director, reducing the producer to being just the banker instead of your boss!

Until then, buy more RAM (can you ever have too much?). And hang in there.

Happy cutting.

“Check the gate.”

Toshiba's Japanese website now lists many new details about its Project Ara modules.

Last month, we covered details about Toshiba's Project Ara cameras modules, and since then Toshiba has increased the resolution for two of them. The lowest-end camera now offers 2.1 MP, up from the previously-reported 2 MP; although this seems like a minor change, it does make the camera capable of full HD recording at 60fps.

The mid-range camera gets a bigger improvement, leaping from 5 MP to 8 MP, and it's capable of recording at 30 fps at its max resolution. This new camera also boasts a more energy efficient design that the company claims will reduce power consumption by around 15 percent compared to other 8 MP cameras.

In addition to the camera modules, Toshiba is working on a media bar module, which utilizes the 8 MP camera mentioned above and includes an audio codec and a speaker.

There's also a display module in the works that will work with LCD and OLED displays and contains an audio codec with microphone and headphone connections.

Probably the most interesting modules being developed relates to connectivity. Utilizing both RF and IF, Toshiba has a module already in production that could potentially control your entire house, including TV, air conditioning and security.

Toshiba is currently producing Wi-Fi modules, too. Although details on which 802.11 standards will be supported are still unknown, the top-tier module will have NFC technology built in. 

Toshiba has big plans for NFC. A video produced by the company demonstrated the use of NFC using a prototype module and two tablets.

Companies have the ability to produce cheap NFC cards that can be placed inside of posters and other items. By simply pressing the NFC reader against the card inside the poster, users could load webpages. Using two devices together, users could easily send a website from one device to the other.

Using a smartphone and TV with NFC technology, for example, you could begin watching a video on your phone and then continue playback on the TV. All you would have to do is touch the two devices together.

Toshiba indicated that many of these modules are currently in production, which means we could see them in upcoming Project Ara devices in the near future.

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

A Guide To Inexpensive Video Editing Software

Video Editing Software Buyer's Guide:

1. Video Editing Hardware: What You’ll Need To Get Started
2. A Guide To Free Video Editing Software
3. A Guide To Inexpensive Video Editing Software
4. A Guide To Professional Video Editing Software

What You’ll Get/What You Won’t

Tools for consumers or prosumers, under or around the $100 mark

Know this at the outset: the ecosystem of video editing software is mature enough that you generally get what you pay for. Frankly, if you need highly watchable color-correction, subtle transitions, artfully conceived effects, very precise keying, deep track-stacking (nesting), uncompressed 4K export or actually useful audio mixing tools, just skip this entire application class and make the jump to the pro-level editors listed in part four of this series.

Yes, that’s harsh and will upset a number of developers. We're sorry.

If, on the other hand, it’s most important to get stuff done, out and up to social media (de-facto highly compressed), the basic editors are better than good; they’re good enough.

This level (consumer and “first-time prosumer”) is pretty much the domain of Windows. Only a couple of titles are written for Mac OS. There was a time when creative types needed Macs and productivity workers needed Windows. That ended eight years ago as Apple put Intel inside (even as Steve refused to put Intel Inside stickers on the outside of his products). We each have two brain hemispheres; we all need both linear, hierarchical, calculated organization and instinctive, associative intuition. There’s no one right computer for video editing.

Several NLE makers are starting to roll-out apps for tablets. This is about as sensible as making a sex tape. It only seems like a good idea for a few minutes. Then you think it through.

There are lots of screens out there now. They come in five basic types: phone, tablet/laptop, TV, computer monitor and theater (big screen). Unless you know you’ll be uploading to a content delivery network (CDN) that generates a brace of renditions in various formats and resolutions, you need to think about your editor’s ability to output for a big range of codecs (encode/decode software to squash the bits that make up your pretty picture streams into formats that your audience can play-out). Direct export to social networks has become a must-have feature of many editor titles.

Finally, be aware you are not limited to what’s in the box (or download). There’s a budding ecology of third-party plug-ins, a few of which (like New Blue) cross the brand-divides, appearing in several manufacturer's products.

CyberLink PowerDirector

PowerDirector from CyberLink is solid, superb and a little bit sexy. It’s quick to get around and remarkably speedy to render. It repays honest value. It’ll let you get your feet wet with 4K cutting.

But PowerDirector has a long-standing reputation for crashing and there’s lot’s of negative angst in the user community around CyberLink’s commonly crappy customer service.

A capable tool is nothing but a pain if it isn’t easy to use. PowerDirector is easy. It wants to make friends. It recognizes people (or at least faces) in your footage. It works generally the way your instincts tell you it should. Its tools fit your hand. Mostly.

At launch, the product gives you three tiers: a simple Slideshow Creator, a fast Easy Editor and its Full Feature Editor. If you like—and I know you will—you can ask PowerDirector to default to the Full Feature set. But try the Easy Editor a few times, even if just to watch the Themes feature at work. It’s a lot like iMovie in the Mac realm.

CyberLink has this marketing thing for magic: MagicMovie, MagicCut (auto-kills boring footage), MagicFix (guesses at lighting repair while stabilizing video and normalizing audio). I dislike the word in this context; it’s a little insulting to those of us who deal in evidence and engineering and science. But the stuff works well for your quick slash and burn editing jobs.

Stepping into the adult (non-magic) Full Editor, you get an appealing timeline that’s easy to get around. In most editors, the scaling and scrolling of your workspace, to show what you’re actually working on, is frequently a bit dicey. But PowerDirector does this extremely well. It's better, even, than a few big pro foundries like Avid Media Composer.

You’ll be keen on being able to drop multiple Ins and Outs on each source clip; very cool for documentarians laboring to stitch-up a narrative from fly-on-the-wall footage or extended interviews. Likewise, you can ask PowerDirector to Analyze Content for bits of action you may want to exclude (camera bumps and lighting blips) or specifically include (expressive people and dramatic pans or zooms). It’s not artificially intelligent editing, but in this you can grab a glimpse of that future.

We old-fart editors tend to be very protective of our treasured timeline paradigm. But it’s not the only way to crochet a storyline. PowerDirector lets you edit using an alternate Storyboard mode. It’s as simple as dragging and dropping clips depicted by thumbnails. The trouble is, it’s only thumbnails, butt-cut (where the end of one clip abuts the start of the next) together. I'm guessing you’ll grow tired of that quickly and jump back on the timeline.

And on that timeline, you can wrangle up to 100 tracks (not that you’ll use them all). Drag to restack them. PowerDirector recently added support for MultiCam editing. Up to four sources, each on its own track, synchronize based on their audio. You can also use the more old-school approach of timecode synch, or even time-of-day if your cams are set-up accurately.

You can choose from among 150+ transitions and 125+ effects. A lot of these lack a professional level of subtlety and are missing some desirable control handles (although a few auto-calculate how much footage lap is need to make them work well). They are not bad at all for the price, but you can’t search them efficiently. Some were apparently written outside CyberLink and ported in, though there’s not (yet) an open plug-in architecture in PowerDirector.

The keyer works, if you’ve lit your green screen halfway well. But it won’t save your bacon if you haven’t.

There must have been a few sharp code-jockeys at work on PowerDirector. Effects and transitions preview very quickly, even on modestly-powered CPUs. You can see stabilization in close to actual time, so you can fuss with the cropping (to ensure you include all of your clip’s goodness). Just how CyberLink pulls this off is over my programming head. Much more importantly, it seems to be over the competition’s head as well. CyberLink marketing claims “the world fastest video editing engine” and, hardware equality assumed, I don’t doubt it.

I like to layer a lot and am frequently invoking picture-in-picture. Among other uses, this lets you use SD and HD footage at full resolution within a 4K project (important in this time of format transition). PowerDirector has some pic-in-pic game with a positioning coordinate system and the ability to animate PiP objects—you can even add motion blur. These will smoothly visualize without time-sucking render passes. That's more evidence of code elegance under the hood.

But that elegance was hard won. Many longtime users of PowerDirector have gotten used to feeling like crash test dummies. The title traditionally had a nasty habit of blowing up. Rev. 12, however, seems to be much more stable and, reportedly, a lot faster. Then again, the war stories I hear come from folks who are running it native in 64-bit on newer hardware with oodles of RAM.

Audio is handled by a price-tiered structure of companion routines and supplementary programs: Audio Room, Voice-Over Room, WaveEditor and (for a significant extra spend) the “round-trip-able” but standalone-priced digital audio workstation package AudioDirector. For scoring, MagicMusic churns out pre-licensed cues, but it too costs extra.

PowerDirector was an early facilitator of 3D stereoscopic video editing, in the game well before MAGIX or Sony.  For the current release, CyberLink baked up some fresh 3D transitions effects and key-able font/title templates. 3D film-making has been around since 1922, but stuttered and sputtered often. Audience interest jumps in fits and starts with long stretches of ho-hum in between. There’s still a long way to go towards practicability and artistic subtlety in the 3D genre. That’s true in stereo for PowerDirector’s handling of 3D as well. But if you insist on poking people in the eye with sharp pointed sticks, please remember: working in 3D implies you need to more than double your drive capacity, and rendering generally slows down by at least half.

There’s a decent dojo of online ninja training at CyberLink, and more in the wild on YouTube. But if you still need answers, CyberLink supports you with the Director Zone, template sharing service and a user forum.

Like Starbucks, which offers four named-sizes of beverage each of which translates as “large,” CyberLink offers PowerDirector in Deluxe, Ultra, Ultimate and Ultimate Suite at ascending price/feature points.

Bottom line: If you can stomach Windows, you won't do better than PowerDirector, until you make the jump to all-pro programs

Adobe Premiere Elements

Adobe Premiere Elements lets you make excellent movies—for cheap—while you learn and grow. But Adobe is betting you’ll want to pay your way up to full Premiere Pro as part of the company's Creative Cloud. It’s a similar marketing strategy to that of Lightworks (and heroin dealers worldwide): an entry-level platform to get you addicted to the interface, leading to a high-priced habit for the endorphin reward of knowing you have the latest and (in theory) best version.

Adobe doesn’t think like television; it thinks like the Internet. Before you get to edit anything, you are prompted to build a cluster of metadata around all of your elements. Ingesting your video is preferentially done through Adobe’s Organizer, a separate routine from the Editor. You can, instead, “Get Media” from within the Editor, but you feel like you’re swimming upstream against the Adobe River flow.

But Premiere Elements recently did learn something from film and TV editors with their language of shots/clips placed in bins: Elements now gives you an “Assets” window somewhat similar to what you’ll find in the big professional flagships like Lightworks Pro and Avid Media Composer.

Adobe is hanging back from 4K for now. I have no doubt that when high pixel-count begins flowing through Premiere Elements, it will look spectacular. Perhaps the company’s calculus is based on the realization that mobile pipelines are too narrow for practical extreme-res production. More likely, Adobe marketers figure that if you really desire 4K, you’re not so cost-sensitive and they can easily up-sell you into the pricey Premiere Pro product. If so, Adobe has not woken up to the vibe of the GoPro crowd.

Similarly, you cannot work in 3D stereoscopic with Elements. But who really does that? Amazing how most of us wouldn’t consider listening to music that isn’t in stereo, but hardly anyone feels passionate about two channels of video.

Operationally, Elements steps you from an elegantly simple Quick mode (I just want to make my movie now), through a Guided zone (earn while I learn), up to the Expert level.

Color correction in Premiere Elements has some serious game. Adobe’s Photoshop pedigree may be at work here. As with pro-level editors, Elements has a tri-level color-grader so you can separately address areas in shadow, in highlight and in-between. If you need more subtle tweakage, you could port a clip's worth of frames out to Photoshop Elements, fix one, apply the tweak-settings to a stack of them, and return them to Premiere. But who—other than astrophotography buffs making eclipse movies—has time for that?

In fact, any color correction takes time, attention and perspective. It’s easy to fool yourself and lose reference. Fortunately, there’s a shortcut called Auto Smart Tone. Park on a frame that averages the scene. Auto Smart Tone applies a range of algorithmically-concocted fixes, displaying them in four panes placed around your original. Move your cursor towards the corner or side showing the look you want, and Elements grades the whole clip towards your choice. Happy day.

To make your movie positively jolly with emotional vibe, you can try FilmLooks. These are canned thematic packages, each using color, noise, gain, grain, motion simulation, distress, contrast and other treatments to give you an overall feel. You can have, say, an old home movie, or newsreel, or cartoon look. FlimLooks overrides any previous light bending you’ve applied to your clip. But Adobe gives you various levers to pull on the FilmLooks treatment. Caution: it’s way easy to overdo this.

Big props to Premiere Elements for its prowess with titles! I’m not enough of a code-kid to tell if some of Adobe Illustrator’s genes have been spliced in, but it wouldn’t surprise me. You get useful templates for common font work across a series of themes. And you can step-in, design your own title looks, and animate them in interesting design-y ways. Fashionable!

For handheld shots, or those taken from aboard a bumpy ride, Elements offers a Stabilize treatment. But it’s not wonderful—not as good as the parallel feature in Apple’s free iMovie. You can end up trading shaky-cam for what one old editor friend describes as bad-acid-cam.

Premiere Elements attempts rudimentary position-path following (motion tracking). This lets you mark a bounding box around an object moving within a frame, and then assign an element (perhaps a font line, a graphic or a picture-in-picture video frame) to that object. It’s OK for simple paths and cartoon-y esthetics. But Adobe has a ways to go before this becomes a seriously graceful feature.

For anything beyond standard-def video, Premiere Elements starts to huff and puff. Remember: video is a pig. Adobe tries to slaughter this data hog in the common way. You work in reduced resolution (“proxies”) until you can’t stand it anymore, whereupon you hit render and go make yourself a latte. You'll sip that slowly while your computer ka-chunks along, merrily burning dinosaurs and heating the planet. In a rendering race, Elements will probably get there ahead of most of the pack, but it doesn't come close to beating CyberLink’s PowerDirector.

Once rendered final, Elements does a fine job of outputting to a great range of codecs, resolutions and bitrates. The Publish+Share badge brings up built-in hacks built for a bunch of choice destinations in social sharing, or for media burning.

At this final point in your workflow, Adobe would like to connect your production output with your asset input. Wait, what? Adobe is building a cloud called Revel. You get a free account (with subscription fee-based upgrades), into which you launch videos, photos, graphics. These become raw materials; synchronized fodder for your next opus—accessible from Premiere Elements’ Organizer. They also get transcoded by Revel for “view-ability” on any platform (remember those five screen sizes).

If you know Photoshop well, or if you see the handwriting on the wall and picture yourself as a Creative Cloud customer, Premiere Elements is the right entry door NLE for you.

Corel VideoStudio X7 Pro And Ultimate

Corel users tend to be ferociously loyal bunch. So I imagine I’ll be on the pointy end of more than a few barbed comments for this dismissive snark: VideoStudio is no toy, but neither is it much of a creative instrument.

VideoStudio is most at home in the corporate presentation quadrant. Its interface is much more AV Dude than Filmmaker. But, in fairness, let’s recall that are a lot of employed AV Dudes and a lot of starving filmmakers.

Corel gives you a solid template set for slideshows and easily facilitates computer screen capture. And, if you check your artistic pretensions at the splash-screen, you may be laughing your way to the bank by the time you close out of the program.

A 64-bit native architecture is much less of a big deal than it was 18 months ago. Still, we should recognize that Corel forded the 64-bit river earlier than most. With support for increased memory capacity, VideoStudio can slash through 4K renders and meld stacks of HD tracks (up to about 20 of them).

Like others in the space, Corel gives you a FastFlick quick-and-dirty edit mode that helps get a visual story together, well, fast. 

My snooty-attitude notwithstanding, Corel VideoStudio does have a cute back door wormhole into the filmmaking world. If you own a Canon DSLR, VideoStudio Pro can likely drive it to create stop motion animations.

If you’re married to the Windows Mob, give VideoStudio a try. Even if you hate it (and you probably won’t), at $80, you won’t have gotten stung too badly.

Cough up an additional $20 for the “Ultimate” package and you get extra tools for shot-stabilization that can fix some of the “jello-cam” (rolling-shutter) effect that happens when you pan with a DSLR.

Corel VideoStudio X7 Ultimate also gives you tweaky color correction/alteration (called ColorFast), a rudimentary pen-on-screen “tele-strator”-style light-doodler (RotoPen), a title-animating kit that simulates handwriting (Handscript), a useful set of title/graphic animation templates by Boris (Graffiti 6) and some less-than-understated transitions and “light-splashy” effects (VitaScene). Hey, sometimes you need garish.

One standout feature is Corel’s entry into motion tracking. This is useful tech previously available only from pro-level programs. It lets you do tricks like identifiers/explainers that follow objects around the screen, “greek-ing” things you don’t want your audience to see (brand names, genitalia and so on), follow-the-ball sports action and more.

The premium level VideoStudio Ultimate drops a hundredweight of picture element-shifting gizmos (Motion Effects) on you. It also lets the third-party plug-in makers in the door. One standout: the Mercalli SE plug offers tweaky, tweezy stabilization. It takes too many steps, but only because it offers very precise control and creative choice.

If you spent all your money on a 4K camera, Corel gives you the least costly editing tool to cut your footage. But it will be slow unless you have monster hardware, in which case you probably didn’t need to skimp on the software after all.

If you’re going direct to the Web in an HTML5 environment, VideoStudio lets you code interactivity into your video pages. You can prompt users to play videos without downloading and shepherding a raft of plug-ins. Your movie can have multiple plots and user-dependent endings. Just be sure to set your project up from the get-go to export .html.

And, like other packages in this class, you can spit out your final versions to social-sharing directly.

Finally, Corel’s Discovery Center is one of the most organized free training sites around. It’s a nice way to get a sense of VideoStudio before you buy it. And a vital way to run it once you have.

Sony Movie Studio

Apparently, what happened in Vegas didn’t stay in Vegas this time. Sony Movie Studio is the new name for Vegas, a product the company has been selling for more than 12 years. But it’s the same workbench. This version is apparently not much more than another de-featured adaptation of the venerable Vegas Pro. Sony didn’t write it, but acquired the code-base from creator Sonic Foundry.

Vegas/Movie Studio began life as an audio cutting tool, to which video was later added. So, it still feels like Digital Audio Workstation (DAW) software, except it has no MIDI (Musical Instrument Digital Interface) chops. Movie Studio applies visual effects with a plug-in paradigm that evokes an audio mixing desk with a patch-bay.

The latest Movie Studio (version 13) architecture is natively 64-bit, and it’s a bit faster than Vegas was. It does handle 4K media astonishingly fast compared to others in this price range.

You can customize your workspace quite a bit. But some of the functions one sees in peers are missing. Others are there, but you wouldn’t know it from the interface. For example, you can change the play-out rate of a clip to fit a hole in your timeline with a keyboard hack that “rubberizes” the media file, but there’s no obvious button or switch to get you into the mode. Clip stabilization is hiding somewhere there too. Confusing.

I find the interface perplexing and pokey, decidedly old school, but not in a good way. Images and source videos live in film-style bins, but you can’t drop other types of media or effects there. Most of the switches and levers look like an alien spaceship. You’ll soon find yourself scratching your head, muttering, “What’s this do?” about some control. You can toggle your cursor to give you a pop-up explanation window or—in some cases—hover over icons to get a pop-up cheater. What you’ll discover is that several of the widgets appear to do the same things.

Even with the explainers, operating the editor feels like swimming in honey. Or like that dream where you flick your black coat-tails over the piano bench at Carnegie Hall as you sit to perform the Rachmaninoff Concerto #3, only to discover your hands have morphed into permanent boxing gloves. This thing is sluggish.

It’s a timeline-only editor. And getting around in the timeline takes some getting used to. There’s no thumbnail/storyboard view. You stay in one terminal view; you'll find no separate windows for ingest or export. That’s counter-intuitive. It seems like a very linear workflow for such a specifically nonlinear utensil. Sony has updated its Trim/Split method, but that’s hardly a breakthrough.

Got two eyes? MovieStudio gives you 3D stereoscopic video editing at this low price point. But so does MAGIX Movie Edit. Plus, CyberLink’s PowerDirector isn’t really all that much more costly, and it runs 3D circles around MovieStudio.

You can wrangle up to 20 tracks. That’s probably enough (it’s certainly a decent value for the price). More expensive competitors offer hundreds, though. Audio “Voices” display waveforms, but you can’t really use this to visually edit audio very well. You’ll have to hunt around to turn on deeper audio-sweetening gizmos, but they are there. Many of these are derived from Sonic Foundry Sound Forge.

When you’ve locked your cut (or given up fighting with it), you can export via the “Make Movie” function to YouTube, or to attached storage. Sony experimented with housing its own “Pixelcast” sharing service, but bagged it this past April after less than two years in the game. At least you can still handily burn a DVD or Blu-ray.

Sony hosts a marketing department-feeling “forum” to supposedly service users. But it looks like they are more interested in up-selling customers than helping them.

If you buy Sony Movie Studio, buy a crash helmet. In other words, save your work early and often. The package is prone to system hangs. Be sure to reach out frequently to the Sony Creative site to patch your system. At least that means your software is supported and backed by working developers. That’s no small thing. A few lesser competitors are just riding out old investments, trying to squeeze a few last shekels before folding their tents. Sony still cares about brand reputation. It’s still Vegas, baby.

MAGIX Movie Edit Pro 2014

Windows users who need to cut quick (and, yes, a little dirty) video should take a look at MAGIX Movie Edit Pro. It’s not the fastest, most comprehensive or most cost-effective NLE. But it seems to be purpose-built.

Let’s call it “sportsmanlike.” Maybe you have a new 4K extreme-sports camera and a dream to produce your own version of The Endless Summer. Or you bought into 3D stereoscopic and yearn to really pop open people’s eyes (Plus and Premium levels only). MAGIX Movie Edit seems to understand your worldview. Like a few competitors, it edits in a proxy mode for data-gobbling high-resolution formats. As you intercut action shots, you can feel the fluidity of motion without choking your platform. Then, sit back and see it all done (after rendering) in UHD.

Even at full-res, the software seems to have some smart-tricky load balancing coded in. It makes notably efficient use of your CPU, and of your GPU if you’re running Windows 7 or above.

If you come back with a couple of angles on the action, there’s a rudimentary multi-cam edit mode. Movie Edit synchs each stream for you if the audio is clearly recorded on both.

If the available light on your shoot wasn’t great, there’s a two-level color corrector to pull more goodness from your images. New in this release is an HDR effect mode, which tries to pull more contrast in picture details for that “hyped-life” look.

If you edit in Europe, you probably know of MAGIX. If you’re working in the U.S., the brand may be unfamiliar. MAGIX Movie Edit has been in the game a long time. Maybe too long. Except for its 4K chops, the software is on the stale side of current capabilities from competitors.

Like most of those other companies, Movie Edit has a quick-up template mode. You pop clips into pre-visualized stencil-like buckets and out spits your cinematic experience. You can create a hybrid of slide show and video. Do this tastefully and it's effective. Fail to bring esthetics (or a concept) to the party and you’ll get something like poorly animated cartoon chart-junk littering a bad Power Point.

MAGIX clearly studied Avid’s industry-flagship interface. There are multiple ways to do things through two-dozen icon-driven mode toggles. You can shave clips with various razor blade and barber’s trim widgets. If your intent is to impress a client (or a date) with your mastery of a starship’s worth of arcane buttons to push, you’ll be happy with this interface.  

Movie Edit thinks of every transition as a fade. It’s a little hard to find ones you want. Why? In part, because it’s a multi-level product—meaning the developers built a full-up version with professional aspirations, then watered it down across two lower-function, cheaper tiers.

For scoring, there’s a companion audio suite called MAGIX Music Maker. Naturally, it's sold separately.

In general, there’s a lot of “download this, authorize that” required to run Movie Edit. The feature tiers are close together in price point. I say go right for Premium level or go elsewhere. I wouldn’t be surprised to see MAGIX Movie Edit bundled with some extreme sports camera sometime soon. But that’s pure speculation…

Pinnacle Studio 17.5 Family

Last on our list is the Pinnacle Family. It isn’t the pinnacle of anything. Pinnacle is slow.

Pinnacle does a few things more and goes a (little) deeper than basic freeware editors. But I just don’t see a differentiator that would make you choose it over its very worthy competitors.

And some of those competitors are blood relatives. Pinnacle has been bounced around the industry. In some ways a similar story to Grass Valley EDIUS (see the professional section in Part 4), Pinnacle editing software began as a driver for Pinnacle video ingest hardware. In an inexplicable move, Avid Technologies bought out Pinnacle in 2005, only to divest it out to Corel seven years later. Corel’s stewardship has cleaned up Pinnacle’s user experience a bit. But this product is clearly cruising on the goodwill of its past.

Track count is weirdly limited. You only get three video and three audio tracks at the Studio price point. That clearly seems like a marketing ploy to get you to jump for the 12+12-track Plus level, or the unlimited track count of Ultimate, limited only by your hardware.

Pinnacle claims support for 4K video, and it can edit respectably in the (“Playback Optimized”) proxy mode. But leave lots of time for the proxies to be generated. If you have one hour of footage, allow three hours of cooking time before you’re ready to munch and crunch. It’s also 4K flavor-finicky; your choice of camera will impact Pinnacle’s ability to juggle the frames. Of course, no video edit matters if an audience doesn’t see it. Pinnacle is severely challenged in its 4K export capabilities. 

Audience interest in 3D comes and goes with an inscrutable historical rhythm. But if you’re a dedicated stereoscopic fan (maybe you own a ganged pair of GoPros?), Pinnacle Plus lets you cut movies, add 3D effects and transitions, optimize them and export them in user-friendly formats that run smoothly on graphics cards, 3D TVs, PlayStations and more.

Need music? All levels of Pinnacle come with a rights-pre-paid “Scorefitter” library. It’s facile for tailoring to time. But it sounds a bit like “imitation music.” I'm not talking about song covers; it's just not like real organic, heartfelt music that began in someone’s soul. On the bright side, music rights can easily cost much more that you’ll pay for the entire Pinnacle package, so you’re getting reasonable value. There are a few audio enhancement widgets, such as Music & Speech Cleaner, a port from iZotope, but nothing that moves my VU meter needle.

Pinnacle is effects-rich for the price point, with more than 1500 templates in the toolkit. You get lots of bundles and “portings” from outside the walls of Pinnacle itself. Au-courant production styles, however, are getting less effect-y. So, no extra points for Gryffindor there.

Newsflash: Pinnacle Goes Portable; Reviewer Remains Skeptical. Our now-wireless world lets us edit anything anywhere, right? Uh, not so fast.

Pinnacle will be happy to sell you “Studio for iPad” so you can make cuts on the bus (or toilet). Pinnacle does offer 50GB of cloud vault (provided by Box). But your mobilized productions are decidedly tied down by connectivity and transfer rates. It’s frustrating to watch your needed HD video assets drip ponderously down through a cellular connection. It’s even worse to finish your show, but miss your deadline as you endure the tortuously slow upload to your CDN (content delivery network). And forget 4K, even if you have 4G.

This is, of course, not a Pinnacle-only predicament.

But there are plenty of Pinnacle problems. It’s crash-prone. It’s slow to render. Titles are cruddy. You can’t roll multiple synchronized sources, so no multi-cam.

It would not shock me to wake up one morning and find that Corel unified the product line under its other NLE brand, VideoStudio.

Wrap-Up: Basic Video Editors Under Or Around $100

Each of these — every editor, in fact — speaks something of its own language. Capture, ingest, import and digitize can mean the same thing, or not. One dude’s Export may be another dudine's Make Movie. At a certain point, you just have to choose a horse and ride.

There are a few other software titles in this space: Nero Video, AVS Video Editor, Roxio Creator, MoviePlus, Honestech Video Editor. Our sister site, Top Ten Reviews, looked at each. Most of what these do can be covered by the freeware titles mentioned in Part 2, so I steered clear of soiling my delicate editorial digits with them. 

When it comes right down to it, the editor you want is the one you’re most comfortable with. But beware of staying loyal to an inferior tool simply because you’ve grown accustomed to it. 

Truthfully, if you fuss around in any one package long enough, you can map a path around a lot of the shortcomings of any editor. But ask yourself: “What is my time worth?” If you are serious about making beautiful movies, eventually you will be dissatisfied with any of the NLE’s we’ve looked at so far.

Now, are you ready to move uptown?

In Part four, we beam up to the enterprise class of non-linear video editing software. Check your Bitcoin balance and hang on tight...

When touchscreen smartphones began playing a bigger role in video-watching activities, VideoLAN had to start embracing the mobile world as well and port its apps to the popular mobile platforms. Even though there were already good alternatives to VLC, such as the free (but ad-based) MX Player, VLC tends to have better codec and format support than most players out there, offering MKV, MP4, AVI, Ogg, MOV, FLAC ,TS, M2TS and AAC. It's also open source and free.

Today, we got the news that VLC 1.0 has finally been released as stable, losing the beta tag. VLC will continue to keep a beta branch for the braver users who want to get all the latest updates first and help the group behind the open source organization, VideoLAN, to discover and fix the bugs in the new versions.

Despite being an open source app, VLC was first launched for iOS rather than Android (the open source OS out of the two), presumably because it was easier to port most of the source code from the desktop version. The iOS mobile operating system was also more popular than Android at the time, so the group may have simply tried to answer to the platform that had the biggest demand.

Unfortunately for VideoLAN and the iOS users waiting for the app, Apple rejected VLC from the iTunes App Store. Apparently, there was a conflict between VLC's GPL open source license, which is incompatible with any sort of DRM-like restrictions, and Apple's mandatory DRM for all iOS apps.

After the rejection, VideoLAN switched its focus to the Android platform and has been in the process of porting and optimizing its application to all Android devices for the past couple of years.

From the changelog in the Play Store is the following statement:

"This release fixes ARMv8 processors, Android 5.0 crashes and minor improvements. The 0.9.x series is major release with hardware decoding and a new interface available in dark or white colors. It integrates DVD iso and menu support, an equalizer, playlist management, Widi [sic] screens support and updated SD cards detection. Hardware acceleration is now enabled by default on 4.3+ and has better subtitles support. Software decoding has been accelerated too."

VLC 1.0 for Android can be downloaded from the Play Store now.

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