Thanks to a recent report from VideoCardz, we learned that AMD has introduced a series of prebuilt desktop computers in China featuring their new Ryzen 7 9800X3D processor. The systems are being sold through AMD's store on JD.com rather than through traditional computer manufacturers. The Ryzen 7 9800X3D processor, which launches November 7 at $479, will be the centerpiece of these systems. For the components, AMD partnered with several manufacturers: ASUS supplies motherboards and graphics cards, Thermalright provides the 360 mm 360 Frozen Infinity AIO liquid cooling system, and LOONGTR makes the computer cases, which include Ryzen branding. The company is offering seven different configurations. Each system includes 32 GB of DDR5-6000 memory and uses an ASUS TUF Gaming B650M-Plus motherboard with 1 TB storage, except for one model with a Prime X670E-Pro motherboard with 2 TB storage.

Prices range from $1,000 for the base model without a graphics card to $2,100 for the version with AMD's RX 7900 XTX graphics card. Some models include NVIDIA graphics cards instead of AMD's own, specifically the RTX 4070 Super and RTX 4070 Ti Super. The computers will be available starting November 7, just before Singles' Day on November 11, a major shopping event in China. It overlaps with the Ryzen 7 9800X3D launch, given that it is the heart of the system. AMD already sells other prebuilt systems in the Chinese market, including models with various Ryzen processors and graphics cards, such as the China-only RX 6750 GRE.

ASUS today announced the launch of the exclusive Warhammer 40,000: Space Marine 2 Collector's Edition graphics card, developed in collaboration with AMD and Focus Entertainment and officially unveiled at the ROG press event during Gamescom on August 20. At the Gamescom press event, ASUS revealed that this card is based on the TUF Gaming Radeon RX 7900 XTX and is custom-built for a global giveaway for gamers and fans of Space Marine 2 and the Warhammer 40K universe.

To celebrate the game's launch, ASUS and AMD collaborated on a live stage event at the Gamescom Hall on August 23, where three randomly-chosen audience members competed to win one of the exclusive Space Marine 2 graphics cards. After several rounds of quickfire questions centered around Warhammer 40,000 lore and Space Marine 2, one of the participants received the first Space Marine 2 graphics card prize.

ASUS has introduced a new limited edition graphics card: the Radeon RX 7900 XTX Warhammer 40,000 Space Marine 2 Edition. The company will produce only 20 units of this specially designed card, making it one of the rarest graphics cards on the market. The new card is based on the Radeon RX 7900 XTX, which uses AMD's RDNA 3 architecture. It maintains the same technical specifications as the standard model, including its gaming capabilities. The primary difference lies in its aesthetic design, which features artwork and elements inspired by the Warhammer 40,000 franchise, specifically referencing the Space Marines from the upcoming game Space Marine 2. ASUS developed this product in partnership with the developers of Space Marine 2. Each card will be individually numbered, potentially increasing its value for collectors. Despite its unique appearance, the card is expected to perform similarly to the standard RX 7900 XTX in terms of gaming performance.

Distribution of these graphics cards will not occur through typical retail channels. Instead, ASUS plans to organize a giveaway contest on August 23. The company has yet to release full details about how interested parties can enter the contest, and that information will be available through their official communication channels shortly. While the card offers no performance improvements over the standard model, its scarcity and unique design may appeal to enthusiasts of both computer hardware and the Warhammer franchise. Potential contestants should note that, as with any limited edition product, demand is likely to far exceed the available supply. ASUS has not indicated whether similar collaborations or limited editions might be planned for the future.

ASUS is readying a limited edition TUF Gaming Radeon RX 7000 series graphics card themed after Warhammer 40000: Space Marine 2. AMD last week announced the Q3-2024 edition of its "This is Why We Game" bundle that sees the company give away the latest Warhammer 40000 title and Unknown 9 Awakening with select RX 7000 series graphics cards, select Ryzen 7000 series desktop processors, and prebuilts. The ASUS TUF Gaming Warhammer 40000: Space Marine 2 Radeon RX 7000 sees the card draped in colors of the Space Marine armor, including the backplate, and cooler shroud. This isn't a sticker job, but actual metal embellishments. Even the fan impellers come in Space Marine blue, and bear insignia on the impeller hubcaps. At this point we don't know which exact GPU model ASUS is using—our guess looking at the level of detail into the board design, is the very top-end RX 7900 XTX, so ASUS can get the most margins out of it to cover the product design costs. ASUS will launch this graphics card (which will include the game), on August 20, 2024.The teaser trailer follows.

ASRock, the global leading manufacturer of motherboards, graphics cards, mini PCs, and gaming monitors, today launched the first passive series graphics cards -- ASRock AMD Radeon RX 7900 XTX Passive 24 GB and ASRock AMD Radeon RX 7900 XT Passive 20 GB graphics cards.

ASRock Radeon RX 7900 Passive series graphics cards are powered by the AMD Radeon RX 7900 XTX and Radeon RX 7900 XT GPUs. Both of these two cards are supporting multi-GPU collaborative computing, and designed for multi-card parallel computing for better performance. ASRock AMD Radeon RX 7900 Passive series graphics cards are featuring a VAPOR-CHAMBER heatsink, efficiency aluminum cooling fins with v-shaped cutting. Furthermore, thanks to the single horizontal 12V-2x6 power connector, to install couple of ASRock AMD Radeon RX 7900 Passive series graphics cards become much easily due to less power cords.

ASRock, the leading global motherboard, graphics card and mini PC manufacturer, today launched their first blower series graphics cards -- ASRock Radeon RX 7900 XTX Creator 24 GB and ASRock Radeon RX 7900 XT Creator 20 GB graphics cards.

ASRock Radeon RX 7900 Creator series graphics cards are powered by the AMD Radeon RX 7900 XTX and Radeon RX7900 XT GPUs. Both of these two cards are supporting multi-GPU collaborative computing, and designed for multi-card parallel computing for better performance. Radeon RX 7900 Creator series graphics cards are featuring a VAPOR-CHAMBER heatsink, efficiency blower fan and 2-slot thickness. Furthermore, thanks to the single horizontal 12V-2x6 power connector, installing a couple of ASRock Radeon RX 7900 Creator series graphics cards becomes much easier due to fewer power cords.

Regulatory filings by PC OEM major Acer, with the Eurasian Economic Commission (EEC), drop some of the first references to three upcoming performance-segment graphics card SKUs by AMD. These include the Radeon RX 7900, the Radeon RX 7800, and the Radeon RX 7700. Acer is an AMD Radeon add-in board partner under its Acer Predator brand, which makes graphics cards not just for Predator and Nitro gaming PCs, but also graphics cards for the retail channel.

At this point, there are no specs known for the three, not even their memory sizes. It would be interesting to see how AMD slots the three into its product stack, particularly the RX 7900. The company currently has three RX 7900 series SKUs—the flagship RX 7900 XTX, and the second-best RX 7900 XT, but then the SKU next to this is the RX 7900 GRE, which saw a worldwide launch. Then there are the RX 7800 and RX 7700, which are pretty straightforward to predict. The RX 7800 will slot in between the RX 7700 XT and RX 7800 XT, and will probably feature a 256-bit wide memory interface with 16 GB of memory, much like the RX 7800 XT; whereas the RX 7700 will be positioned below the RX 7700 XT. It wouldn't surprise us if AMD further cuts down the memory interface to 160-bit (10 GB).

XFX has prepared its newest "Magnetic Air" series of Radeon RX 7000 series GPUs with hot-swappable fans for the US market, where they were only exclusive to the Chinese region. The company has informed us that the series will be called "Magnetic Air," instead of the previously believed "MagAir." Comprising the XFX Qicksilver Radeon RX 7800 XT Magnetic Air and the XFX Mercury Radeon RX 7900 XTX Magnetic Air, these GPUs are now available for purchase in the US, with an official launch date set for June 18. What sets the Magnetic Air series apart is its unique cooling design, featuring Honeywell PTM 7950 phase-changing thermal pads that offer superior thermal conductivity. The cooling system boasts 216 ultra-thin matrix fins strategically arranged in a high and low structure to optimize airflow, while the heatsink is attached to a vapor chamber for efficient heat dissipation. Additionally, the backplate is crafted from durable aluminium die-cast, ensuring long-lasting performance.

The XFX Qicksilver Radeon RX 7800 XT Magnetic Air is available in both white and black variants, priced at $529.99, while the XFX Mercury Radeon RX 7900 XTX Magnetic Air carries a price tag of $979.99. The most innovative feature of the Magnetic Air series is its swappable fan design, allowing users to customize their cooling setup to suit their specific needs. This level of flexibility is sure to appeal to enthusiasts and overclockers seeking optimal performance from their rigs. Currently, only the black variants of the 7900XTX and 7800XT are in stock, but all Magnetic Air models are expected to be available on the official launch date of June 18 for PC enthusiasts in the United States.

[Editor's note: Our in-depth review of the RX 7900 XTX Magnetic Air is now live]

At Computex 2024, we went hands-on with the only custom-design AMD Radeon RX 7900 XTX graphics card to implement the 16-pin 12V-2x6 power connector. We've written quite a bit about this card, but weren't expecting it to be this long. This isn't just the only RX 7900 XTX with a 12V-2x6, but also the only air-cooled RX 7900 XTX that's strictly 2 slots-thick. ASRock made this card for graphics rendering farms, or AI development/inferencing builds, in which a GPU server would have 4-6 of these packed like sardines. The lateral blower helps ensure proper ventilation. The backplate is recessed to ensure better ventilation to the neighboring card.

The card sticks to AMD reference clock speeds of 2270 MHz Game clock, and 20 Gbps (GDDR6-effective) memory speed. ASRock also gave us a technical deep-dive into the card's design. For one, it's confirmed that the power connector is 12V-2x6 (H++) and not 12VHPWR. The cooling solution consists of a large copper vapor-chamber plate that makes contact with the GPU, memory, and VRM; conveying heat to an aluminium channel heatsink, which is ventilated by a high-speed lateral blower. The cooler shroud and backplate are both made of aluminium.

GIGABYTE showed off a trio of new or recently announced graphics cards based on the GeForce RTX 40-series GPUs. The lineup is led by the RTX 4080 SUPER AORUS Xtreme Ice, a limited edition, bleeding edge product that the company has serial numbered each card with. The card features some very premium materials in its board design that both look good and are very sturdy. The card also features GIGABYTE's highest factory overclock for the RTX 4080 SUPER.

Next up, is the GeForce RTX 4070 Ti SUPER AI TOP. The company just announced the new AI TOP ecosystem of first-party generative AI tools, AI assistants, and hardware certified to accelerate AI, and this graphics card is one of them. It features a lateral-blower cooling solution. The idea behind this is so you could install many of these cards in a desktop or workstation to build a rendering or acceleration farm. The company also unveiled an AI TOP branded AMD TRX50 workstation motherboard, which should be a perfect pairing for this card. Lastly, there are a few recently announced cards, the RTX 4070 SUPER Eagle OC Ice, and the AMD Radeon RX 7900 XTX Gaming OC.

ASRock is ready with the first Radeon RX gaming graphics card to feature a modern 12V-2x6 power connector, replacing the up to three 8-pin PCIe power connectors it took, to power a Radeon RX 7900 series graphics card. The ASRock RX 7900 series WS graphics cards are also the first 2-slot RX 7900 series graphics cards. They target workstations and GPU rendering farms that stack multiple graphics cards into 4U or 5U rackmount cases, with no spacing between 2-slot graphics cards. ASRock is designing cards based on both the RX 7900 XT, and the flagship RX 7900 XTX.

The ASRock RX 7900 series WS graphics cards appear long and no more than 2 slots thick. To achieve these dimensions, a lateral-flow cooling solution is used, which combines a dense aluminium or copper channel heatsink with a lateral blower. Remember we said these cards are meant for workstations or rendering farms? So the noise output will be deafening, at least up to datacenter standards. The most striking aspect of these cards of course is their 12+4 pin ATX 12V-2x6 power input, which is capable of drawing 600 W of continuous power from a single cable. It's located at the card's tail-end, where it would have been an engineering challenge to put three 8-pin connectors.

ASRock, a global leader in motherboards, graphics cards, gaming monitors, and small form factor PCs, is excited to announce its participation in the upcoming COMPUTEX Taipei 2024. AI has been a hot topic in the IT industry in recent years. ASRock keeps up with the trends, showcasing the latest AI solutions. In addition, creating an impressive gaming environment is one of ASRock's commitments for gamers. Visitors can explore the latest products and AI computing solutions firsthand and experience the gaming simulator at ASRock booth L0818.

Full Firepower Display of Next Gen Motherboard Platform
ASRock will also be revealing its full range of next generation Intel & AMD motherboards including the new line up of Taichi & Phantom Gaming Series, the robust Steel Legend, newly designed Livemixer and mainstream Pro series. The high end Taichi brand will receive its major repositioning and reorganized to be ASRock' s flagship series by merging with existing OC Formula & AQUA product line. Alongside the next gen Phantom Gaming series, the full array of Taichi AQUA, Taichi OCF, Taichi and Taichi Lite motherboards will also be revealed during Computex 2024!

NVIDIA's upcoming GeForce RTX 5090 Founders Edition graphics card may implement a design closely resembling the cinder block product design the company readied for its RTX 4090 Ti graphics card that never materialized into a marketable product. This sees a 4-slot thick board design, with a slender main logic PCB arranged along the plane of the motherboard, on top of which the cooling solution is mounted perpendicular to the plane, as shown in the images below. This main logic board contains the GPU, memory, and VRM. There two additional PCBs—one has the display I/O, and the other has the PCIe interface. There is a fourth disaggregated component, the 12V-2x6 receptacle, located somewhere along the top of the cooling solution.

Confirmation of NVIDIA using the RTX 4090 Ti "cinder block" board design for the RTX 5090 comes from kopite7kimi, a reliable source with NVIDIA leaks. Kopite7kimi mentions a card that has a "Main Board, IO Rigid Board and a separate PCIE slot component (perhaps it should not be considered as the third PCB)," which perfectly describes with the RTX 4090 Ti. NVIDIA had completed the design phase of this card, which made it to its cooling solution OEM (which is likely where the images leaked out from). The company probably decided against launching this product because the AMD Radeon RX 7900 XTX fell significantly short of the performance proposition of the RTX 4090.

AMD earlier this month released documentation for the Micro Engine Scheduler (MES) firmware of its RDNA 3 GPUs. The MES is a hardware component that distributes graphics processing and general-purpose compute workloads among the main number-crunching machinery of the AMD GPU—the shader engines, which contain the compute units (CU), the indivisible SIMD muscle of the GPU with programmable shaders and certain kinds of specialized hardware, such as the AI Accelerator and Ray Accelerator. The MES is driven by a programmable firmware, which is what AMD released developer documentation of.

The decision to release MES documentation probably comes from the very top of AMD corporate management. In March, a controversy erupted when Tiny Corp, builders of AI compute servers and workstations, complained of issues testing the Radeon RX 7900 XTX graphics card with a ROCm compute stack, prompting a response from no less than CEO Dr Lisa Su. There were then calls to open-source the firmware, which AMD didn't agree to, probably since it treads on their core GPU hardware IP; but the company did the next best thing, by releasing detailed developer documentation for the MES firmware.

ASUS introduced the Radeon RX 7900 XTX and RX 7900 XT DUAL OC graphics cards. The two join ASUS's rather slim RX 7900 series custom-design lineup, which until now only included the TUF Gaming OC products for the RX 7900 XT and RX 7900 XTX. ASUS's common board design for the cards it's launching, features a large cooler shroud, a tall, yet triple-slot board design, and a cooling solution that uses an aluminium fin-stack heatsink that uses no more than two Axial-Tech fans, hence the name DUAL OC. We've seen this exact board design on some of the RTX 30-series "Ampere" DUAL OC products, so ASUS may probably be carrying over the design, with suitable changes for compatibility with the "Navi 31" GPU.

Both the RX 7900 XTX and RX 7900 XT DUAL OC cards feature a milder factory overclock compared to the company's TUF Gaming OC products. The RX 7900 XTX DUAL OC does 2455 MHz Game clock (compared to 2365 MHz reference); while the RX 7900 XT DUAL OC offers 2075 MHz Game clocks compared to 2025 MHz AMD reference. It's also interesting to note here, that the RX 7900 XTX/XT DUAL OC cards feature a significantly different board design than the RX 7900 GRE DUAL OC, which features a design closer to that of the RX 7800 XT DUAL OC. Both the RX 7900 XT DUAL OC and RX 7900 XTX DUAL OC feature triple 8-pin PCIe power connectors, and display I/O that includes three DisplayPort 2.1, and one HDMI 2.1. Besides the minimal RGB, the cards offer dual-BIOS, with the Q-BIOS running them at reference speeds, and with a tighter fan curve. The card measures 32.3 cm in length, 14.7 cm in height, and is no more than 3 slots thick. The company didn't reveal pricing.

XFX over the weekend rolled out the Radeon RX 7900 XTX Phoenix Nirvana graphics card, the company's new flagship custom-design product based on fastest AMD Radeon GPU you can buy. From the looks of it, the card is China-exclusive, as we haven't seen any updates from the European or North American arms of XFX showcase this card. The Phoenix Nirvana is noticeably larger than the company's current MERC 310, and features a 4-slot design, with a significantly heavier heatsink. The card is 34.6 cm long, 13 cm tall, and is 4 slots thick.

The star attraction with this card is its large aluminium fin-stack heatsink, which features 216 aluminium fins for a total of 62.586 mm² of surface area for heat dissipation; six copper heatpipes, a vapor-chamber base, and 15,000 W/mK Honeywell PTM 7950 phase-changing thermal pad between the vapor-chamber and the GPU. Each of the three 100 mm fans of the XFX RX 7900 XTX Phoenix Nirvana features a fluid-dynamic bearing, turns at speeds of up to 3,200 RPM, pushing 70.08 CFM of airflow, at 4.88 mm H₂O, each. The card is priced at ¥7,899 in China, which converts to roughly $1,090.

Today, we have the latest round of leaks that suggest that AMD's upcoming RDNA 4 graphics cards, codenamed the "RX 8000-series," might continue to rely on GDDR6 memory modules. According to Kepler on X, the next-generation GPUs from AMD are expected to feature 18 Gbps GDDR6 memory, marking the fourth consecutive RDNA architecture to employ this memory standard. While GDDR6 may not offer the same bandwidth capabilities as the newer GDDR7 standard, this decision does not necessarily imply that RDNA 4 GPUs will be slow performers. AMD's choice to stick with GDDR6 is likely driven by factors such as meeting specific memory bandwidth requirements and cost optimization for PCB designs. However, if the rumor of 18 Gbps GDDR6 memory proves accurate, it would represent a slight step back from the 18-20 Gbps GDDR6 memory used in AMD's current RDNA 3 offerings, such as the RX 7900 XT and RX 7900 XTX GPUs.

AMD's first generation RDNA used GDDR6 with 12-14 Gbps speeds, RDNA 2 came with GDDR6 at 14-18 Gbps, and the current RDNA 3 used 18-20 Gbps GDDR6. Without an increment in memory generation, speeds should stay the same at 18 Gbps. However, it is crucial to remember that leaks should be treated with skepticism, as AMD's final memory choices for RDNA 4 could change before the official launch. The decision to use GDDR6 versus GDDR7 could have significant implications in the upcoming battle between AMD, NVIDIA, and Intel's next-generation GPU architectures. If AMD indeed opts for GDDR6 while NVIDIA pivots to GDDR7 for its "Blackwell" GPUs, it could create a disparity in memory bandwidth performance between the competing products. All three major GPU manufacturers—AMD, NVIDIA, and Intel with its "Battlemage" architecture—are expected to unveil their next-generation offerings in the fall of this year. As we approach these highly anticipated releases, more concrete details on specifications and performance capabilities will emerge, providing a clearer picture of the competitive landscape.

Vastarmor's Radeon RX 7900 XTX Super Alloy custom model was revealed a while back (four months ago according to VideoCardz)—for whatever reason, it has taken a long time for finalized units to reach retail outlets in China. A newly discovered JD.com listing reveals an interesting price point of 6799 RMB (~$940)—given the Radeon RX 7900 XTX GPU's age at this point in time, Vastarmor has implemented quite a steep discount over the launch MSRP for Chinese markets (7999 RMB). The premium tier ARGB-appointed "Super Alloy" models sport substantial custom cooling solutions—clearly designed to temper higher boost clocks. VideoCardz has looked at the best Navi 31-based cards on the field: "(Vastarmor's latest) is actually among the fastest models on the market. Currently, the highest boost clock for RX 7900 XTX is 2680 MHz, and it can be found on models like ASRock Taichi, Aqua, PowerColor Liquid Devil, or Sapphire Nitro+."

The RX 7900 XTX Super Alloy is one of the largest high-end gaming graphics cards out there—it is a triple-slot, 330 mm x 134 mm x 69 mm design. Strangely, Vastarmor's reference specced RX 7900 XTX Starry Sky model also sports a similarly-proportioned cooling solution. ZOTAC's Prime Gamer Force (PGF) OC design remains triumphant in terms of stupendous dimensions—check out these digits: 381 mm (L) x 154 mm (W) x 74 mm (D)! ZOTAC PGF cards are Chinese market exclusives—much like Vastarmor's best offerings—only the most hardcore/deep-pocked enthusiasts outside of the PRC will be importing these vast gaming hardware delights.

George Hotz and his startup operation (Tiny Corporation) appeared ready to completely abandon AMD Radeon GPUs last week, after experiencing a period of firmware-related headaches. The original plan involved the development of a pre-orderable $15,000 TinyBox AI compute cluster that housed six XFX Speedster MERC310 RX 7900 XTX graphics cards, but software/driver issues prompted experimentation via alternative hardware routes. A lot of media coverage has focused on the unusual adoption of consumer-grade GPUs—Tiny Corp.'s struggles with RDNA 3 (rather than CDNA 3) were maneuvered further into public view, after top AMD brass pitched in.

The startup's social media feed is very transparent about showcasing everyday tasks, problem-solving and important decision-making. Several Acer Predator BiFrost Arc A770 OC cards were purchased and promptly integrated into a colorfully-lit TinyBox prototype, but Hotz & Co. swiftly moved onto Team Green pastures. Tiny Corp. has begrudgingly adopted NVIDIA GeForce RTX 4090 GPUs. Earlier today, it was announced that work on the AMD-based system has resumed—although customers were forewarned about anticipated teething problems. The surprising message arrived in the early hours: "a hard to find 'umr' repo has turned around the feasibility of the AMD TinyBox. It will be a journey, but it gives us an ability to debug. We're going to sell both, red for $15,000 and green for $25,000. When you realize your pre-order you'll choose your color. Website has been updated. If you like to tinker and feel pain, buy red. The driver still crashes the GPU and hangs sometimes, but we can work together to improve it."

George Hotz and his Tiny Corporation colleagues were pinning their hopes on AMD delivering some good news earlier this month. The development of a "TinyBox" AI compute cluster project hit some major roadblocks a couple of weeks ago—at the time, Radeon RX 7900 XTX GPU firmware was not gelling with Tiny Corp.'s setup. Hotz expressed "70% confidence" in AMD approving open-sourcing certain bits of firmware. At the time of writing this has not transpired—this week the Tiny Corp. social media account has, once again, switched to an "all guns blazing" mode. Hotz and Co. have publicly disclosed that they were dabbling with Intel Arc graphics cards, as of a few weeks ago. NVIDIA hardware is another possible route, according to freshly posted open thoughts.

Yesterday, it was confirmed that the young startup organization had paused its utilization of XFX Speedster MERC310 RX 7900 XTX graphics cards: "the driver is still very unstable, and when it crashes or hangs we have no way of debugging it. We have no way of dumping the state of a GPU. Apparently it isn't just the MES causing these issues, it's also the Command Processor (CP). After seeing how open Tenstorrent is, it's hard to deal with this. With Tenstorrent, I feel confident that if there's an issue, I can debug and fix it. With AMD, I don't." The $15,000 TinyBox system relies on "cheaper" gaming-oriented GPUs, rather than traditional enterprise solutions—this oddball approach has attracted a number of customers, but the latest announcements likely signal another delay. Yesterday's tweet continued to state: "we are exploring Intel, working on adding Level Zero support to tinygrad. We also added a $400 bounty for XMX support. We are also (sadly) exploring a 6x GeForce RTX 4090 GPU box. At least we know the software is good there. We will revisit AMD once we have an open and reproducible build process for the driver and firmware. We are willing to dive really deep into hardware to make it amazing. But without access, we can't."

AMD in a GPUOpen blog post showed off some "super early" performance numbers for a Radeon RX 7900 XTX GPU rendering a DirectX 12 workload using Work Graphs, instead of the traditional ExecuteIndirect method. Work Graphs is method by with GPUs enjoy greater autonomy in executing render and general purpose compute workloads, by vastly reducing the role of the CPU in the rendering pipeline. At the ongoing GDC 2024, AMD showed off a performance demo of a DirectX 12 rendering workload that implements Work Graphs, running in sync with Mesh Nodes, a feature that will process draw calls while the rest of the graph is executing. This is compared its render times to the traditional method. The differences are staggering.

It takes the traditional ExecuteIndirect method 64% longer to render a frame compared to Work Graphs, in other words, the new method is 39% faster. This has a direct impact on frame-rates for applications that implement Work Graphs. Although not part of the demo, AMD RDNA 3 also implement a silicon-level acceleration for Multi-draw indirect, another API-level feature that's underutilized. AMD's demo showcases a 3D scene without the HUD UI and skybox, being rendered on a single work graph dispatch. Work Graphs and Mesh Nodes are the next big feature addition to the DirectX 12 API feature-set, which will begin rolling out later this year. Both AMD and NVIDIA have ongoing implementation efforts to implement it.

Lately Tiny Corp. CEO—George Hotz—has used his company's social media account to publicly criticize AMD Radeon RX 7900 XTX GPU firmware. The creator of Tinybox, a pre-orderable $15,000 AI compute cluster, has not selected "traditional" hardware for his systems—it is possible that AMD's Instinct MI300X accelerator is quite difficult to acquire, especially for a young startup operation. The decision to utilize gaming-oriented XFX-branded RDNA 3.0 GPUs instead of purpose-built CDNA 3.0 platforms—for local model training and AI inference—is certainly a peculiar one. Hotz and his colleagues have encountered roadblocks in the development of their Tinybox system—recently, public attention was drawn to an "LLVM spilling bug." AMD President/CEO/Chair, Dr. Lisa Su, swiftly stepped in and promised a "good solution." Earlier in the week, Tiny Corp. reported satisfaction with a delivery of fixes—courtesy of Team Red's software engineering department. They also disclosed that they would be discussing matters with AMD directly, regarding the possibility of open-sourcing Radeon GPU MES firmware.

Subsequently, Hotz documented his interactions with Team Red representatives—he expressed 70% confidence in AMD approving open-sourcing certain bits of firmware in a week's time: "Call went pretty well. We are gating the commitment to 6x Radeon RX 7900 XTX on a public release of a roadmap to get the firmware open source. (and obviously the MLPerf training bug being fixed). We aren't open source purists, it doesn't matter to us if the HDCP stuff is open for example. But we need the scheduler and the memory hierarchy management to be open. This is what it takes to push the performance of neural networks. The Groq 500 T/s mixtral demo should be possible on a tinybox, but it requires god tier software and deep integration with the scheduler. We also advised that the build process for amdgpu-dkms should be more open. While the driver itself is open, we haven't found it easy to rebuild and install. Easy REPL cycle is a key driver for community open source. We want the firmware to be easy to rebuild and install also." Prior to this week's co-operations, Tiny Corp. hinted that it could move on from utilizing Radeon RX 7900 XTX, in favor of Intel Alchemist graphics hardware—if AMD's decision making does not favor them, Hotz & Co. could pivot to builds including Acer Predator BiFrost Arc A770 16 GB OC cards.

The TinyBox AI server system attracted plenty of media attention last week—its creator, George Hotz, decided to build with AMD RDNA 3.0 GPU hardware rather than the expected/traditional choice of CDNA 3.0. Tiny Corp. is a startup firm dealing in neural network frameworks—they currently "write and maintain tinygrad." Hotz & Co. are in the process of assembling rack-mounted 12U TinyBox systems for customers—an individual server houses an AMD EPYC 7532 processor and six XFX Speedster MERC310 Radeon RX 7900 XTX graphics cards. The Tiny Corp. social media account has engaged in numerous NVIDIA vs. AMD AI hardware debates/tirades—Hotz appears to favor the latter, as evidenced in his latest choice of components. ROCm support on Team Red AI Instinct accelerators is fairly mature at this point in time, but a much newer prospect on gaming-oriented graphics cards.

Tiny Corporation's unusual leveraging of Radeon RX 7900 XTX GPUs in a data center configuration has already hit a developmental roadblock. Yesterday, the company's social media account expressed driver-related frustrations in a public forum: "If AMD open sources their firmware, I'll fix their LLVM spilling bug and write a fuzzer for HSA. Otherwise, it's not worth putting tons of effort into fixing bugs on a platform you don't own." Hotz's latest complaint was taken onboard by AMD's top brass—Dr. Lisa Su responded with the following message: "Thanks for the collaboration and feedback. We are all in to get you a good solution. Team is on it." Her software engineers—within a few hours—managed to fling out a set of fixes in Tiny Corporation's direction. Hotz appreciated the quick turnaround, and proceeded to run a model without encountering major stability issues: "AMD sent me an updated set of firmware blobs to try. They are responsive, and there have been big strides in the driver in the last year. It will be good! This training run is almost 5 hours in, hasn't crashed yet." Tiny Corp. drummed up speculation about AMD open sourcing GPU MES firmware—Hotz disclosed that he will be talking (on the phone) to Team Red leadership.

Tiny Corp., a neural network framework specialist, has revealed intimate details about the ongoing development and building of its "tinybox" system: "I don't think there's much value in secrecy. We have the parts to build 12 boxes and a case that's pretty close to final. Beating back all the PCI-E AER errors was hard, as anyone knows who has tried to build a system like this. Our BOM cost is around $10k, and we are selling them for $15k. We've put a year of engineering into this, it's a lot harder than it first seemed. You are welcome to believe me or not, but unless you are building in huge quantity, you are getting a great deal for $15k." The startup has taken the unusual step of integrating Team Red's current flagship gaming GPU into its AI-crunching platform. Tiny Corp. founder—George Hotz—has documented his past rejections of NVIDIA AI hardware on social media, but TinyBox will not be running AMD's latest Instinct MI300X accelerators. RDNA 3.0 is seemingly favored over CDNA 3.0—perhaps due to growing industry demand for enterprise-grade GPUs.

The rack-mounted 12U TinyBox build houses an AMD EPYC 7532 processor with 128 GB of system memory. Five 1 TB SN850X SSDs take care of storage duties (4 in raid, 1 for boot), and an unoccupied 16x OCP 3.0 slot is designated for networking tasks Two 1600 W PSUs provide necessary electrical juice. The Tiny Corp. social media picture feed indicates that they have acquired a pile of XFX Speedster MERC310 RX 7900 XTX graphics cards—six units are hooked up inside of each TinyBox system. Hotz's young startup has ambitious plans: "The system image shipping with the box will be Ubuntu 22.04. It will only include tinygrad out of the box, but PyTorch and JAX support on AMD have come a long way, and your hardware is your hardware. We make money either way, you are welcome to buy it for any purpose. The goal of the tiny corp is to commoditize the petaflop, and we believe tinygrad is the best way to do it. Solving problems in software is cheaper than in hardware. tinygrad will elucidate the deep structure of what neural networks are. We have 583 preorders, and next week we'll place an order for 100 sets of parts. This is $1M in outlay. We will also ship five of the 12 boxes we have to a few early people who I've communicated with. For everyone else, they start shipping in April. The production line started running yesterday."

Bethesda on February 20 released the 1.9.67 path update for Starfield, which adds support for AMD FSR 3.0 performance enhancement, including frame generation. To celebrate this update, the game's developers are giving away an ultra-rare Collector's Edition bundle of Starfield-themed flagship AMD hardware. The bundle includes an AMD Radeon RX 7900 XTX with a special paint-job; and an AMD Ryzen 7 7800X3D, which remains the fastest desktop processor for gaming. The RX 7900 XTX Starfield Collector's Edition card features a special cooler shroud and backplate design with design elements from the game; including some anodized aluminium fins in its heatsink in the game's colors. The Giveaway is open to residents of the US, Canada (excluding Quebec), Mexico, the UK, the Netherlands, Belgium, and Luxembourg. You don't need to purchase the game or any AMD hardware to be eligible. Simply follow the Starfield Twitter account, and reply to the Giveaway announcement tweet. One winner will be randomly chosen. Find all Giveaway rules here.