Close monitoring of AMD engineering activities—around mid-February time—revealed the existence of a new set of patches for GNU Compiler Collection (GCC). At the time, news reports put spotlights on Team Red's "znver5" enablement—this target indicated that staffers were prepping Zen 5 processor microarchitecture with an expanded AVX instruction set (building on top of Zen 4's current capabilities). Phoronix's Michael Larabel has fretted over AMD's relative silence over the past month—regarding a possible merging of support prior to the stable release of GCC 14.

He was relieved to discover renewed activity earlier today: "AMD Zen 5 processor enablement has been merged to GCC Git in time for the GCC 14.1 stable release that will be out in the coming weeks. It was great seeing AMD getting their Zen 5 processor enablement upstreamed ahead of any Ryzen or EPYC product launches and being able to do so in time for the annual major GNU Compiler Collection feature release." Team Red is inching ever closer to the much anticipated 2024 rollout of next-gen Ryzen 9000 processors, please refer to a VideoCardz-authored timeline diagram (below)—"Granite Ridge" is an incoming AM5 desktop CPU family (reportedly utilizing Zen 5 and RDNA 2 tech), while "Strix Point" is scheduled to become a mobile APU series (Zen 5 + RDNA 3.5).

AMD's flagship desktop processor, the Ryzen 9 7950X3D, saw an impressive retailer-level price-cut, sending its street-price down to $565 on Newegg, or nearly $135 (19%) below its launch price of $700. The processor is listed at $595, but a $30 discount coupon sends the price down to $565. It's not just Newegg, even Amazon US has the chip listed at $591, with a $26 discount coupon sending it down to $565. The Ryzen 9 7950X3D is about 1% slower than the 7800X3D at gaming, but is still about as fast as the Core i9-14900K at gaming; and its 8 additional "Zen 4" cores helps shore up productivity performance over the 7800X3D.

Price-cuts to the Ryzen 9 7950X3D can be seen as a move to greet last week's launch of the Intel Core i9-14900KS, an enthusiast-magnet for its up to 6.20 GHz clock-speeds, and overclocking headroom; but in our testing, the new chip barely beats the Ryzen 7 7800X3D in gaming; although is 2% faster than the regular i9-14900K at both gaming- and productivity.

The mid-range AMD Radeon RX 5600 XT is not supposed to have 16 GB of video memory, but the same hardware modders from Brazil behind the recent GeForce RTX 2080 16 GB mod, had other ideas for the card. They have not only increased the memory size to 16 GB through memory chip replacement, but also succeeded in widening its memory bus to 256-bit. The RX 5600 XT was launched in 2018 with 6 GB of 14 Gbps GDDR6 memory over a 192-bit memory interface. The card is cut down from the 7 nm "Navi 10" silicon powering the RX 5700 series, by enabling 36 out of 40 compute units (the same count as the RX 5700), but with a truncated 192-bit memory bus wired to 6 GB of memory (and so 25% lower memory bandwidth).

Paulo Gomes and Ronaldo Buassali pulled off the daring Radeon RX 5600 XT memory mod, which involves not just increasing the memory size from 6 GB to 16 GB, but also widening the memory bus from 192-bit to 256-bit. Since the RX 5600 XT is based on the same "Navi 10" GPU as the RX 5700, custom-design graphics cards tend to reuse PCB designs from the RX 5700 series, and have two vacant memory pads that are sometimes exposed and even balled. The mod involves three key stages—to replace the six 8 Gbit GDDR6 memory chips with eight 16 Gbit ones; to add the required electrical SMDs and VRM components for the two additional memory chips; and lastly, to give the card a modified BIOS that can let it play with the new memory configuration. The "Navi 10" silicon also powers certain Radeon Pro graphics cards with 16 GB of memory using 16 Gbit memory chips, so that could be the starting point for the BIOS mod.

Acer introduced the Predator Bifrost Radeon RX 7900 GRE OC graphics card. This card is based on what appears to be an identical board design to its Predator Bifrost RX 7800 XT OC. Given that the TBP of both GPUs is similar to each other around the 260 W-mark, this shouldn't be a cause for concern. The card features a triple-slot cooling solution, with an aluminium fin-stack heatsink, and three Frostblade 3.0 fans with webbed impellers, designed to maximum axial airflow. The card draws power from a pair of 8-pin PCIe power connectors. Display outputs include three DisplayPort 2.1, and an HDMI 2.1. The card comes with a healthy factory OC of 2050 MHz Game clock, compared to 1880 MHz AMD reference for the RX 7900 GRE.

Based on the compacted version of the "Navi 31" chiplet GPU, the RX 7900 GRE is powered by the latest RDNA 3 graphics architecture, and features 5,120 stream processors across 80 compute units (CU); besides 160 AI accelerators, 80 Ray accelerators, 320 TMUs, and 160 ROPs. Only four of the six memory cache dies (MCDs) are enabled, so the GPU gets 64 MB of Infinity Cache, and a 256-bit wide GDDR6 memory interface, driving 16 GB of memory at 18 Gbps (576 GB/s bandwidth). Acer didn't reveal pricing.

Sony is giving final touches to a first-party super-resolution technology called PSSR (PlayStation Spectral Super Resolution), according to a sensational new leak by Moore's Law is Dead. The company plans to debut the tech with the new PlayStation 5 Pro console, this November. The tech is unlikely to make it to the current PlayStation 5 console due to the underlying graphics architecture. PSSR, from the looks of it, is closer in function to NVIDIA DLSS than it is to AMD FSR. The tech leverages the over 300 TOPS of AI inferencing power of the RDNA 3 GPU powering the PS5 Pro, to drive an AI-based reconstruction algorithm. The RDNA 2-based GPU of the current PS5 lacks AI accelerators. The biggest driving force behind the PSSR development isn't just this AI-based upscaling tech, but the impact of upscaling tech on frame-times and whole-system latencies. PSSR apparently makes the PS5 Pro capable of being not just a 4K-class game console, but also one that's ready to take on 8K. Sony is, after all, a television company, and would want to create use-cases for its latest 8K televisions.

A lot is also being speculated about the GPU driving the PlayStation 5 Pro. We've known from several older leaks that it is based on AMD's latest RDNA 3 graphics architecture, but we're now learning that the GPU will be a mix-match of several current and future graphics architectures from AMD's IP bouquet. It could have more advanced media and display engines than the current Radeon RX 7000 GPUs, but even the shader engines could incorporate certain elements from a future architecture, such as RDNA 4. The report speaks of a total AI inferencing performance of 300 TOPS, an FP16 throughput 67 TFLOPs, and an FP32 throughput of 33.5 TFLOPs. To put these into perspective, the GPU driving the Xbox Series X is rated for 12 TFLOPs FP32.

BIOSTAR, a leading manufacturer of motherboards, graphics cards, and storage devices today, unveils the brand new A620MS motherboard, designed for business and casual use. The A620MS motherboard is a versatile solution that seamlessly bridges the gap between professional efficiency and home entertainment. It's designed to meet the needs of office users, home enthusiasts, and system integrators, whether it's enhancing productivity in the workplace, enjoying online movies and gaming at home, or managing complex business processes and client data. The A620MS is a benchmark for reliability and adaptability, catering to diverse computing environments.

Built on the advanced Socket AM5, the A620MS motherboard offers comprehensive support for the latest AMD Ryzen 7000 and 8000 series processors, ensuring compatibility with cutting-edge CPU technology. It harnesses the power of AMD's A620 single-chip architecture to provide a stable and efficient computing base. With support for up to 96 GB of DDR5 memory across two DIMM slots, it offers ample capacity and superior speed for demanding applications, making it a powerhouse for both current and future computing tasks. Furthermore, It supports PCIe 4.0 M.2 (64 Gb/s) for advanced, high-speed storage solutions, ensuring a streamlined and efficient computing experience.

With the XMG FUSION 15 and CORE 15, XMG unveils its first 15.3-inch laptops in 16:10 format. The two compact devices, which are very similar except for a few details, differ primarily regarding their CPUs: the FUSION 15 is powered by Intel's Core i9-14900HX or i7-14650HX and either an NVIDIA GeForce RTX 4060 or 4070, while the CORE 15 combines an AMD Ryzen 7 8845HS with an RTX 4060. Apart from that, the gaming laptops, which weigh around 2 kg and are 21.9 mm high, offer a WQXGA display with a brightness of 500 nits and an almost entirely aluminium chassis with a clean and elegant design.

The design language of the XMG FUSION 15 (E24) and the XMG CORE 15 (M24) focuses on understated elegance. The laptops share an identical, black anodised and torsion-resistant aluminium chassis. At 342.5 x 245 x 21.9 mm, these two 15.3-inch devices are very compact, while a reduced weight of just 2.0 kg (XMG CORE 15) or 2.1 kg (XMG FUSION 15) offers good portability.

ASUS today announced BIOS updates that enable support for up to 256 GB of memory on its Intel 700 and 600 series motherboards that use DDR5 modules. Such models with four DIMM slots can now support up to 256 GB of memory, while such models with two DIMM slots can now support up to 128 GB. These enhancements significantly improve multitasking potential, ensuring smooth and seamless computing. AMD AM5 motherboards from ASUS do not require a BIOS update to enable support for up to 256 GB of DDR5 memory modules.

The BIOS updates can be accessed on the ASUS support pages for the models listed below.

AMD today released its first drivers to implement Microsoft's DirectX Agility SDK version 1.613, which introduces the new DirectX 12 Work Graphs 1.0 API. AMD has extensively worked on implementing the new technology, which among other things, significantly reduces the CPU's role in most common shader graphics workloads, and improve GPU shader thread saturation, as the GPU waits less on the CPU's share of shader workloads. The new AMD Software Adrenalin 23.40.14.01 drivers are off the main driver update channel, and is intended for developers and enthusiasts to start exploring GPU Work Graphs. GPU Upload Heaps, and certain features of Shader Model 6.8 on supported AMD Radeon GPUs. There are some known issues with the driver specific to AMD's implementation of GPU Work Graphs, and the latest version of the Agility SDK in general, which are listed below.

DOWNLOAD: AMD Software Adrenalin 23.40.14.01 for Agility SDK Support

AMD today released the latest version of the Ryzen Chipset Software. Version 6.02.07.2300 WHQL updates the driver versions of its various component drivers, and is applicable to AMD 300-series through 600-series desktop chipsets; and TRX and WRX series Ryzen Threadripper chipsets. The version also identifies a few issues that will be addressed in future updates, such as non-English driver names appearing in the English language selection; incorrect Uninstall log when uninstallation fails; and a failure to upgrade Ryzen PPKG.

The installer from AMD has now been fixed, and we are now hosting the 6.02.07.2300 driver.

DOWNLOAD: AMD Chipset Software 6.02.07.2300 WHQL

A new patch for Avatar: Frontiers of Pandora deployed on March 1, bringing more than 150 fixes and adjustments to the game. Title Update 3 includes technical, UI, balancing, main quest, and side quest improvements, plus additional bug fixes. To provide players with an improved experience, the development team, led by Massive Entertainment, listened to feedback from the community while working on Title Update 3.

An additional patch, Title Update 3.1, was deployed on March 7, adding additional fixes to the game. Check out the full list of improvements included in Title Update 3 & 3.1 here, and read on for the most notable improvements now available in Avatar: Frontiers of Pandora.

Update Mar 14th: TPU has received alerts regarding player feedback—Massive Entertainment's "Title Update 3" has reportedly broken their implementation of FSR 3 in Avatar: Frontiers of Pandora. We will keep an eye on official Ubisoft channels—so far they have not addressed these FSR-related problems.

TrendForce highlights the current landscape of the HBM market, which as of early 2024, is primarily focused on HBM3. NVIDIA's upcoming B100 or H200 models will incorporate advanced HBM3e, signaling the next step in memory technology. The challenge, however, is the supply bottleneck caused by both CoWoS packaging constraints and the inherently long production cycle of HBM—extending the timeline from wafer initiation to the final product beyond two quarters.

The current HBM3 supply for NVIDIA's H100 solution is primarily met by SK hynix, leading to a supply shortfall in meeting burgeoning AI market demands. Samsung's entry into NVIDIA's supply chain with its 1Znm HBM3 products in late 2023, though initially minor, signifies its breakthrough in this segment.

It looks like cryptocurrency mining is back in craze, as miners are firing up their old mining hardware from 2022 to cash in. Bitcoin is now north of $72,000, and is dragging up the value of several other cryptocurrencies, one such being Qubic (QBIC). Profitability calculators put 24 hours of Qubic mining on an AMD Ryzen 9 7950X 16-core processor at around $3, after subtracting energy costs involved in running the chip at its default 170 W TDP. "Zen 4" processors such as the 7950X tend to retain much of their performance with slight underclocking, and reducing their power limits; which is bound to hold or increase profitability, while also prolonging the life of the hardware.

And thus, the inevitable has happened—stocks of the AMD Ryzen 9 7950X have disappeared overnight across online retail. With the market presence of the 7950X3D and the Intel Core i9-14900K, the 7950X was typically found between $550-600, which would have added great value considering its low input costs. CPU-based cryptocurrency miners, including the QBIC miner, appear to be taking advantage of the AVX-512 instruction set. AMD "Zen 4" microarchitecture supports AVX-512 through its dual-pumped 256-bit FPU, and the upcoming "Zen 5" microarchitecture is rumored to double AVX-512 performance over "Zen 4." Meanwhile, Intel has deprecated what few client-relevant AVX-512 instructions its Core processors had since 12th Gen "Alder Lake," as it reportedly affected sales of Xeon processors. What about the 7950X3D? It's pricier, but mining doesn't benefit from the 3D V-cache, and the chip doesn't sustain the kind of CPU clocks the 7950X manages to do across all its 16 cores. It's only a matter of time before the 7950X3D disappears, too; followed by 12-core models such as the 65 W 7900, the 170 W 7900X, and the 7900X3D.

A 2019-signed export license has allowed Intel to supply laptop processors to Huawei, under an exclusive deal—this US Government approved arrangement was not viewed favorably by AMD. The rival chipmaker apparently missed out on the securing of a similar trade license back in 2021. According to a new Reuters report, Team Red and a handful of supporters have attempted to revoke Intel's license—worth hundreds of millions of dollars. Two anonymous sources allege that: "Intel has survived an effort to halt chip sales to Huawei...giving one of the world's largest chipmakers more time to sell to the heavily sanctioned Chinese telecoms company." Intel and Huawei's symbiosis is set to end later this year—folks on the inside reckon that the current US administration will not approve a renewal. Reports suggest that Qualcomm is not anticipating a renewal either—Huawei is an approved buyer of Snapdragon chips, but industry whispers indicate an eventual shift to in-house fare.

Intel, Huawei, US Commerce Department and the White House have declined to comment on the aforementioned scenario. Reuters also sent a query to AMD, but the publication did not receive a response. Earlier last year, a government official revealed that "Huawei's licensing policy" was under review, alongside a general push to scrap a number of trade deals. According to insiders, the same government official allegedly told companies—in private—that the US Commerce Department would fix "the licensing discrepancy." Another anonymous source believes that the agency shelved these plans late last year, for reasons unknown—they stressed that there is potential for a revival. Given the upcoming expiry of Intel and Huawei's arrangement—within the year—it makes little sense to implement a drastic change.

Thanks to Phoronix, we have discovered that AMD has recently released initial patches for performance monitoring and events related to their upcoming Zen 5 processors in the Linux kernel. These patches, sent out for review on the kernel mailing list, provide the necessary JSON files for PMU (Performance Monitoring Unit) events and metrics that will be exposed through the Linux perf tooling. As the patches consist of JSON additions and do not risk regressing existing hardware support, there is a possibility that they could be included in the upcoming Linux v6.9 kernel cycle. This would allow developers and enthusiasts to access detailed performance data for Zen 5 CPUs once they become available, helping with optimization and analysis of the next-generation processors.

The release of these patches follows AMD's publication of performance monitor counter documentation for AMD Family 1Ah Model 00h to 0Fh processors last week, confirming that these models represent the upcoming Zen 5 lineup. While Linux kernel 6.8 already includes some elements of Zen 5 CPU support, the upstream Linux enablement for these next-generation AMD processors is an ongoing process. Upon Phoronix examining the Zen 5 core and uncore events, as well as the metrics and mappings, it appears that they are mainly similar to those found in the current Zen 4 processors. This suggests that AMD has focused on refining and optimizing the performance monitoring capabilities of its new architecture rather than introducing significant changes. As the launch of Zen 5 CPUs draws closer, we await to see the performance and capabilities of these next-generation processors. With performance monitoring also getting a push, this could be a sign that Zen 5 launch is nearing.

Microsoft's DirectX department is scheduled to show off several innovations at this month's Game Developers Conference (GDC), although a late February preview has already spilled their DirectSR Super Resolution API's beans. Today, retail support for Shader Model 6.8 and Work Graphs has been introduced with an updated version of the company's Agility Software Development Kit. Program manager, Joshua Tucker, stated that these technologies will be showcased on-stage at GDC 2024—Shader Model 6.8 arrives with a "host of new features for shader developers, including Start Vertex/Instance Location, Wave Size Range, and Expanded Comparison Sampling." A linked supplementary article—D3D12 Work Graphs—provides an in-depth look into the cutting-edge API's underpinnings, best consumed if you have an hour or two to spare.

Tucker summarized the Work Graphs API: "(it) utilizes the full potential of your GPU. It's not just an upgrade to the existing models, but a whole new paradigm that enables more efficient, flexible, and creative game development. With Work Graphs, you can generate and schedule GPU work on the fly, without relying on the host. This means you can achieve higher performance, lower latency, and greater scalability for your games with tasks such as culling, binning, chaining of compute work, and much more." AMD and NVIDIA are offering driver support on day one. Team Red has discussed the launch of "Microsoft DirectX 12 Work Graphs 1.0 API" in a GPUOpen blog—they confirm that "a deep dive" into the API will happen during their Advanced Graphics Summit presentation. NVIDIA's Wessam Bahnassi has also discussed the significance of Work Graphs—check out his "Advancing GPU-driven rendering" article. Graham Wihlidal—of Epic Games—is excited about the latest development: "we have been advocating for something like this for a number of years, and it is very exciting to finally see the release of Work Graphs."

LG Electronics (LG) announced today pricing and availability of its 2024 UltraGear OLED gaming monitors, designed to provide superior gaming experiences. The lineup is led by the 45GS95QE-B—an immersive 45-inch curved LG OLED monitor with a 21:9 aspect ratio. The expanded lineup includes the highly anticipated CES 2024 Innovation Award-winning 32-inch 4K gaming monitor (model 32GS95UE) equipped with the new Dual Mode feature which allows gamers to switch between 4K (3,840 x 2,160) at 240 Hz and Full-HD (1,920 x 1,080) at 480 Hz with one simple click.

The company is also introducing its 39- and 34-inch models with ultra-wide curved screens: the 39GS95QE and 34GS95QE. All four models are now available for pre-order thru March 17, 2024 at LG.com (shipping expected in late March and throughout April) and available later at LG-authorized retailers. Customers who place a pre-order for the 45GS95QE-B, 39GS95QE or 34GS95QE UltraGear OLED gaming monitors via LG.com are eligible to receive a $200 Virtual Mastercard prepaid card. Terms and conditions apply.

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.

AMD has caught up with NVIDIA and Intel in the race to get a locally run AI chatbot up and running on its respective hardware. Team Red's community hub welcomed a new blog entry on Wednesday—AI staffers published a handy "How to run a Large Language Model (LLM) on your AMD Ryzen AI PC or Radeon Graphics Card" step-by-step guide. They recommend that interested parties are best served by downloading the correct version of LM Studio. Their CPU-bound Windows variant—designed for higher-end Phoenix and Hawk Point chips—compatible Ryzen AI PCs can deploy instances of a GPT based LLM-powered AI chatbot. The LM Studio ROCm technical preview functions similarly, but is reliant on Radeon RX 7000 graphics card ownership. Supported GPU targets include: gfx1100, gfx1101 and gfx1102.

AMD believes that: "AI assistants are quickly becoming essential resources to help increase productivity, efficiency or even brainstorm for ideas." Their blog also puts a spotlight on LM Studio's offline functionality: "Not only does the local AI chatbot on your machine not require an internet connection—but your conversations stay on your local machine." The six-step guide invites curious members to experiment with a handful of large language models—most notably Mistral 7b and LLAMA v2 7b. They thoroughly recommend that you select options with "Q4 K M" (AKA 4-bit quantization). You can learn about spooling up "your very own AI chatbot" here.

HP Inc. today announced the industry's largest portfolio of AI PCs leveraging the power of AI to enhance productivity, creativity, and user experiences in hybrid work settings.

In an ever-changing hybrid work landscape, workers are still struggling with disconnection and digital fatigue. HP's 2023 Work Relationship Index reveals that only 27% of knowledge workers have a healthy relationship with work, and 83% believe it's time to redefine our relationships with work. Most employees believe AI will open new opportunities to enjoy work and make their jobs easier, but they need the right AI tools and technology to succeed.

AMD Linux engineers have been working on "GFX1150" and "GFX1151" targets for a while—official references to "Strix 1/Strix Point" and "Strix Point Halo" have appeared several times on official development channels. Phoronix's head honcho—Michael Larabel—monitors these activities with keen interest, his latest finding indicates that Team Red is preparing open-source RadeonSI/RADV driver support for the GFX1151 IP. Their MESA 24.1 update merges in GPU enablement for possible high-end "Strix Point Halo" laptop processors—tech tipsters believe that these chiplet variants could sport up to sixteen Zen 5 CPU cores and forty RDNA 3.5 GPU cores.

AMD's enablement of the "GFX1150/Strix Point" GPU appeared online late last month—these monolithic laptop chips are alleged to sit below "Strix Point Halo" in Team Red's product hierarchy. Insiders suggest that the best configurations could house twelve Zen 5 CPU cores and sixteen RDNA 3.5 GPU cores. Phoronix posited that the "RDNA 3 refresh" graphics solution: "is just rumored for select APUs, while ultimately we'll see where this GFX 11.5.1 IP is found if for some further upgraded APU or something more special. In any event the open-source Linux driver support is coming together." According to official product roadmaps, the initial batch of "Strix Point" mobile chips are expected ship later this year—representing a proper next-gen upgrade over current "Hawk Point" offerings.

AMD in an announcement on Tuesday, announced changes to its AMD FreeSync display requirements. The updated requirements should vastly reduce the flood of 1080p and 1440p displays with lower refresh rates, such as 75 Hz, from boasting of FreeSync support. AMD has three tiers of FreeSync based on the refresh-rate and features such as HDR. The base tier FreeSync feature now calls for a desktop monitor or TV to come with a refresh rate of at least 144 Hz, if its horizontal resolution is under 3440 pixels. This would cover everything 2560 x 1600, to 2560 x 1440, to the popular 1920 x 1080, and certain ultrawide resolutions such as 2560 x 1080.

To qualify for the FreeSync Premium tier, which customers use to identify high refresh-rate displays; a desktop monitor or TV with a horizontal resolution under 3440 pixels now needs at least 200 Hz of refresh rate. Displays with over 3440 pixels (which should include 4K Ultra HD), now need at least 120 Hz to qualify for FreeSync Premium. AMD isn't clear on what refresh-rate it takes for a 4K UHD display to qualify for the base FreeSync tier. Moving on, the top FreeSync Premium Pro tier, which helps customers identify displays that not only have high refresh-rates but also HDR; nothing is changed—the display must feature HDR on top of meeting the new requirements for FreeSync Premium. The story is different for notebook displays. For the base FreeSync tier, a refresh-rate of 40-60 Hz will do; and for FreeSync Premium, you need at least 120 Hz. HDR on top of this gives you FreeSync Premium Pro. All displays will need FreeSync hardware support (which essentially builds on top of VESA Adaptive Sync).

AMD's 12-core/24-thread Ryzen 9 7900X3D processor that comes with 3D Vertical Cache technology, is seeing a slew of price-cuts. The SKU appears to be trapped between the 8-core 7800X3D, which remains the fastest desktop processor for gaming; and the 16-core 7950X3D, which is AMD's flagship. A Newegg listing at $408 with a $20 discount coupon see its effective retail price drop to as low as $389, which is just $20 more than the 7800X3D.

Would you pay $20 for four more cores? The choice is not so simple. While the 7900X3D is a 12-core processor, it features a dual-CCD design, with a 6+6 core arrangement between the two CCDs. Just like with the 7950X3D, only one of the two CCDs has the 64 MB 3D Vertical Cache, or 96 MB of L3 cache; while the other is a regular "Zen 4" CCD with 32 MB of on-die L3 cache. AMD's chipset drives use UEFI CPPC preferred-core flagging to guide gaming workloads to the CCD with the 3D V-cache. While we haven't had a chance to test the 7900X3D, this chip is tested by Tom's Hardware to still be faster than the Core i9-13900K at gaming.

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.

According to the latest report from Bloomberg, AMD has hit a roadblock in offering its top-of-the-line AI accelerator in the Chinese market. The newest AI chip is called Instinct MI309, a lower-performance Instinct MI300 variant tailored to meet the latest US export rules for selling advanced chips to China-based entities. However, the Instinct MI309 still appears too powerful to gain unconditional approval from the US Department of Commerce, leaving AMD in need of an export license. Originally, the US Department of Commerce made a rule: Total Processing Performance (TPP) score should not exceed 4800, effectively capping AI performance at 600 FP8 TFLOPS. This rule ensures that processors with slightly lower performance may still be sold to Chinese customers, provided their performance density (PD) is sufficiently low.

However, AMD's latest creation, Instinct MI309, is everything but slow. Based on the powerful Instinct MI300, AMD has not managed to bring it down to acceptable levels to acquire a US export license from the Department of Commerce. It is still unknown which Chinese customer was trying to acquire AMD's Instinct MI309; however, it could be one of the Chinese AI labs trying to get ahold of more training hardware for their domestic models. NVIDIA has employed a similar tactic, selling A800 and H800 chips to China, until the US also ended the export of these chips to China. AI labs located in China can only use domestic hardware, including accelerators from Alibaba, Huawei, and Baidu. Cloud services hosting GPUs in US can still be accessed by Chinese companies, but that is currently under US regulators watchlist.