GIGABYTE Technology: Giga Computing, a subsidiary of GIGABYTE and an industry leader in high-performance servers, and IT infrastructure, today announced the GIGABYTE G383-R80 for the AMD Instinct MI300A APU and two GIGABYTE G593 series servers for the AMD Instinct MI300X GPU and AMD EPYC 9004 Series processor. As a testament to the performance of AMD Instinct MI300 Series family of products, the El Capitan supercomputer at Lawrence Livermore National Laboratory uses the MI300A APU to power exascale computing. And these new GIGABYTE servers are the ideal platform to propel discoveries in HPC & AI at exascale.⁠

Marrying of a CPU & GPU: G383-R80
For incredible advancements in HPC there is the GIGABYTE G383-R80 that houses four LGA6096 sockets for MI300A APUs. This chip integrates a CPU that has twenty-four AMD Zen 4 cores with a powerful GPU built with AMD CDNA 3 GPU cores. And the chiplet design shares 128 GB of unified HBM3 memory for impressive performance for large AI models. The G383 server has lots of expansion slots for networking, storage, or other accelerators, with a total of twelve PCIe Gen 5 slots. And in the front of the chassis are eight 2.5" Gen 5 NVMe bays to handle heavy workloads such as real-time big data analytics and latency-sensitive workloads in finance and telecom. ⁠

Microsoft is preparing a big update for its Windows operating system. Currently at version 11, the company is gearing up for the launch of Windows 12, which is supposed to bring a monumental shift in the tectonic plates of the regular PC user experience. Enhanced by AI, the Windows 12 OS should utilize many features like generative AI, large language models, some GPT integration, and many other tools that could benefit AI, like photo editors. The confirmation for the Windows 12 launch coming in 2024 is sourced from the Taiwanese Commercial Times, which analyzed comments from Barry Lam, the founder and chairman of PC contract manufacturer Quanta, and Junsheng (Jason) Chen, the chairman and chief executive of Acer.

Both of them underscored the importance of AI and that AI PCs are coming with the next version of Windows. Supposedly, the launch date for Windows 12 is set for June 2024. In that timeframe, hardware vendors should roll out their SoCs embedding AI processing elements at every silicon block. Qualcomm is set to debut its Snapdragon Elite X SoCs in mid-2024, aligning with the alleged release schedule of Windows 12. With more players like NVIDIA, AMD, and others planning to utilize an Arm instruction set for their next-generation PC chips, we expect to see Windows 12 get full-fledged support for Arm ISA and treat it like a first-class citizen in the OS.

AMD today released the latest version of AMD Software Adrenalin Edition. Version 23.12.1 WHQL comes with optimization for "Avatar: Frontiers of Pandora," and "The Finals." The AMD Software UI has undergone a redesign for its Home and Gaming tabs, which make it easier to enable HYPR-RX profiles on a global or per-game basis. With this release, AMD also introduced HYPR-RX Eco, a performance enhancement that prioritizes energy savings, and should benefit gaming notebooks.

The Radeon Software Adrenalin 23.12.1 WHQL drivers now extend FRTC, Radeon Super Resolution, Overlay, Record and Stream features to OpenGL applications. The drivers also finally re-enable Hardware Accelerated GPU Scheduling, but for Radeon RX 7000 series GPUs only. The drivers also enable addtional Vulkan API extensions. Among the issues fixed with this release are high idle power in dual monitor setups with mixed resolutions and high refresh rates; intermittent stuttering seen in Chromium-based browsers (Chrome, Edge), on a machine in which a secondary display is connected to a Radeon RX 7000 series GPU, on machines with Ryzen 7000 series processors.

DOWNLOAD: AMD Software Adrenalin 23.12.1 WHQL

FinalWire Ltd. today announced the immediate availability of AIDA64 Extreme 7.00 software, a streamlined diagnostic and benchmarking tool for home users; the immediate availability of AIDA64 Engineer 7.00 software, a professional diagnostic and benchmarking solution for corporate IT technicians and engineers; the immediate availability of AIDA64 Business 7.00 software, an essential network management solution for small and medium scale enterprises; and the immediate availability of AIDA64 Network Audit 7.00 software, a dedicated network audit toolset to collect and manage corporate network inventories.

The new AIDA64 update introduces a revamped user interface with a configurable toolbar, as well as AVX-512 accelerated benchmarks for AMD Threadripper 7000 processors, AVX2 optimized benchmarks for Intel Meteor Lake processors, and supports the latest AMD and Intel CPU platforms as well as the new graphics and GPGPU computing technologies by AMD, Intel and NVIDIA.

DOWNLOAD: FinalWire AIDA64 Extreme v7.0

BIOSTAR, a leading manufacturer of motherboards, graphics cards, and storage devices today showcases the B650MT motherboard, featuring the AMD Ryzen 5 PRO 7645 processor, the best workstation setup ideal for office users and businesses. The BIOSTAR B650MT motherboard delivers exceptional performance, efficiency, and security to meet the demands of the modern workplace; with support for up to 96 GB of DDR5 memory, it ensures seamless multitasking and exceptional system responsiveness. Offering a versatile storage solution with an M.2 slot and four SATA III ports, this motherboard provides lightning-fast data access and ample storage capacity. BIOSTAR's signature features, including Super Durable Solid Caps, Super Durable Inductor, and Super Hyper PWM, guarantee a dependable and long-lasting system.

With the latest AMD Ryzen 5 PRO 7645 processor, BIOSTAR's B650MT motherboard sets up an ideal platform for business and casual use. The AMD Ryzen 5 PRO 7645 processor offers a compelling blend of top-notch performance, cutting-edge security features, and energy efficiency, making it an excellent fit for the modern-day business sector. Powered by the groundbreaking AMD Zen 4 architecture, this desktop processor comes equipped with integrated AMD Radeon Graphics, simplifying integration, management, and administration tasks. With six cores and twelve threads, fast clock speeds reach up to 5.10 GHz, making it ideal for multitasking and demanding tasks.

A somewhat expected result of the November 17th export ban on RTX 4090 GPUs to China has been that AMD's top offerings; recently rumored to soon suffer the same fate, have been selling faster than AMD or its partners can make them. Predicting the worst, some OEMs such as Dell have allegedly already set their own self-imposed restrictions on the export of both Radeon and Instinct cards to China. Meanwhile insiders in board channels have not yet received any official warning that they can no longer sell RX 7900 XTXs or XTs to DIY markets, however vendors and consumers alike sense an oncoming storm and unable to get RTX 4090s are buying up as many of the high end Radeons as they can.

It's said that AMD's production capacity for this surge in sales has been underwhelming, and supply to Chinese board partners has been barely trickling in. This lack of incoming supply of the Radeon RX 7900 XTX and XT, restriction or not, is creating a shortage of cards in China that is expected to impact the remainder of Q4 2023 and much of Q1 2024. With no immediate restrictions expected for AMD's top cards it's possible that they could react and ramp up exports to cover the demand, however that could be risky if AMD is trying to avoid being targeted.

Gearbox Publishing and Blackbird Interactive have officially unveiled the system requirements for Homeworld 3, the highly anticipated next installment in the GOTY-winning sci-fi RTS series, launching in 2024. Blackbird Interactive continues its work to optimize the game to ensure the fidelity, scale and immersiveness of Homeworld 3's gameplay can be experienced across a wide array of PCs.

Players with the latest graphics cards and processors can enjoy the game in 4K with all of the settings turned up while players with older components can still experience Homeworld 3's hallmark 3D strategy experience with dynamic space battles and incredible visuals.

OneXPlayer, a trailblazer in portable gaming tech, proudly introduces its latest innovation - the OneXGPU. Redefining the gaming landscape, this cutting-edge external graphics processing unit is set to revolutionize portable computer performance. The highly anticipated OneXGPU, the world's first eGPU with integrated storage, will launch on Indiegogo at 00:00 on November 29th (PST November 28th, 08:00 AM). The introductory pricing for the first 96 hours of launch is set at $599. Additionally, users can opt for upgrade options, including the Oculink Cable and various SSD capacities, to customize their experience.

Designed for ultimate portability without compromising on power, the OneXGPU boasts the AMD Radeon RX 7600M XT. With a magnetic back cover allowing swift access to the M.2-2280 SSD slot and versatile interface support via Oculink, it's a game-changer. The RGB lighting adds a stylish touch to its aesthetic. This breakthrough innovation signifies more than an external graphics card. The OneXGPU marks a new era in gaming, delivering unparalleled performance anytime, anywhere. Its integration of SSD PCIe support significantly enhances both storage capacity and performance, promising seamless computing and unmatched speed.

Visit the OneXGPU Campaign on Indiegogo.

The market for mini PCs has been blooming recently, and GEEKOM has been working to put themselves at the forefront with exceptionally well priced and decently performing machines mainly benefiting from their partnership with ASUS. We don't have much info about this upcoming mini PC except that it will offer configurations based on either the Ryzen 7 7840HS or Ryzen 9 7940HS and departs from the ASUS PN series styling for an aluminium chassis that takes clear inspiration from Apple's Mac Mini. Whether this is due to a vendor change or the efforts of internal R&D we'll likely learn when it launches. Aside from the rounded corners, bead blasted aluminium finish, and rear I/O laid out on a black accent fascia plate little else about the machine compares to the Mini except the tiny size at a mere 112.4 x 112.4 mm (4.43 in). Despite this tiny frame the I/O compliment is decent; three 10 Gbps USB 3.2 Gen 2 Type A ports, a 5 Gbps USB 3.2 Gen 1 Type A, a 40 Gbps and a 10 Gbps Type C that can each handle DisplayPort out, two HDMI ports, a 2.5G Ethernet jack, SD Card reader on the side, and one 3.5 mm combo audio jack at the front.

GEEKOM is expected to launch this new mini PC model in Asia within the next month. Worldwide availability should follow soon after. Competitor offerings with similar hardware configurations have been available for a few months by now and have been seen discounted down to around $600 USD in recent weeks, with barebone unit prices going even lower. GEEKOM has quite the challenge ahead of them to offer competitive value.

AMD's 3D V-Cache technology utilizes blocks of SRAM stacked on top of the CPU logic die, where CPU cores reside, and allows the processor to access massive pools of cache for applications. However, using this extra level 3 (L3) cache as a RAM disk appears possible, where the L3 SRAM behaves similarly to a storage drive. A big disclaimer here is that this is only possible by exposing the L3 to the CrystalDiskMark benchmark, and no real-world applications can do it in a way that CrystalDiskMark. According to X/Twitter user Nemez (@GPUsAreMagic), the steps to replicate this procedure are: Obtaining an AMD Ryzen CPU with 3D V-Cache, installing OSFMount and creating a FAT32 formatted RAM disk, and running CrystalDiskMark, with values set to values to SEQ 256 KB, Queue Depth 1, Threads 16, and data fill to 0s instead of random.

The results of this experiment? Well, they appear to be rather stunning as the nature of L3 SRAM is that the memory is tiny but very fast and accessible to the CPU, so it can help load data locally before going to the system RAM. With AMD Ryzen 7 5800X3D, the speeds of this RAM disk are over 182 GB/s for reading and over 175 GB/s for writing. In another test, shared by Albert Thomas (@ultrawide219), we managed to see RAM disk based on AMD Ryzen 7800X3D V-Cache, which scores a little less with over 178 GB/s read and over 163 GB/s write speeds. Again, CrystalDiskMark only performed these tests on small allocations varying between 16 MiB and 32 MiB, so no real-world workloads are yet able to utilize this.

TrendForce's latest research into the HBM market indicates that NVIDIA plans to diversify its HBM suppliers for more robust and efficient supply chain management. Samsung's HBM3 (24 GB) is anticipated to complete verification with NVIDIA by December this year. The progress of HBM3e, as outlined in the timeline below, shows that Micron provided its 8hi (24 GB) samples to NVIDIA by the end of July, SK hynix in mid-August, and Samsung in early October.

Given the intricacy of the HBM verification process—estimated to take two quarters—TrendForce expects that some manufacturers might learn preliminary HBM3e results by the end of 2023. However, it's generally anticipated that major manufacturers will have definite results by 1Q24. Notably, the outcomes will influence NVIDIA's procurement decisions for 2024, as final evaluations are still underway.

Uptake on AMD's latest generation mobile offerings has been slow and steady to put it mildly, but today TUXEDO Computers, a specialist in Linux notebooks at a range of performance and pricing tiers, has announced pre-orders for their new Sirius 16 gaming laptop. This machine combines AMD's latest generation Ryzen 7 7840HS "Phoenix" APU with a Radeon RX 7600M XT RDNA3 GPU inside a sleek all aluminium chassis design that strives to remain understated while still providing a "sleek gamer look" via programmable RGB keys. The Sirius 16 is TUXEDO's first go at an all-AMD configuration and they've held very little back, choosing to allow the full TDP rating of the Phoenix APU at 54 W sustained (or 80 W CPU-only turbo) as well as keeping the RDNA3 GPU at its rated 120 W TGP under full CPU+GPU loads. The Sirius 16 features venting out of both sides as well as the rear of the chassis, and roughly half of the bottom panel is open intake for the dual-fan cooling system.

Powering everything is a 230 W power brick and an 80 Wh replaceable battery bolted inside the chassis. TUXEDO claims up to 10 hours of battery life at minimum display brightness with wireless disabled and without any programmable lighting enabled, or a more realistic 6 hours at medium brightness with wireless enabled and under minimal "office work" load.

NVIDIA's most recent FY3Q24 financial reports reveal record-high revenue coming from its data center segment, driven by escalating demand for AI servers from major North American CSPs. However, TrendForce points out that recent US government sanctions targeting China have impacted NVIDIA's business in the region. Despite strong shipments of NVIDIA's high-end GPUs—and the rapid introduction of compliant products such as the H20, L20, and L2—Chinese cloud operators are still in the testing phase, making substantial revenue contributions to NVIDIA unlikely in Q4. Gradual shipments increases are expected from the first quarter of 2024.

The US ban continues to influence China's foundry market as Chinese CSPs' high-end AI server shipments potentially drop below 4% next year
TrendForce reports that North American CSPs like Microsoft, Google, and AWS will remain key drivers of high-end AI servers (including those with NVIDIA, AMD, or other high-end ASIC chips) from 2023 to 2024. Their estimated shipments are expected to be 24%, 18.6%, and 16.3%, respectively, for 2024. Chinese CSPs such as ByteDance, Baidu, Alibaba, and Tencent (BBAT) are projected to have a combined shipment share of approximately 6.3% in 2023. However, this could decrease to less than 4% in 2024, considering the current and potential future impacts of the ban.

AMD's lineup of Radeon and Instinct GPUs, including the flagship RX 7900 XTX/XT, the professional-grade PRO W7900, and the upcoming Instinct MI300, are facing sales prohibitions in China, according to an alleged sales advisory guide from Dell. This restriction mirrors the earlier ban on NVIDIA's RTX 4090, underscoring the increasing export limitations U.S.-based companies face for high-end semiconductor products that could be repurposed for military and strategic applications. Notably, Dell's report lists several AMD Instinct accelerators, which are integral to data center infrastructure, and Radeon GPUs, which are widely used in PCs, indicating the broad impact of the advisory.

The ban includes discrete GPUs like AMD's Radeon RX 7900 XTX and 7900 XT, which, despite their data-center potential, may still be sold under specific "NEC" eligibility. This status allows for continued sales in restricted regions like sales of NVIDIA's RTX 4090. However, the process to secure NEC eligibility is lengthy, potentially leading to supply shortages and increased GPU prices—a trend already observed with the RX 7900 XTX in China, where it's become a high-end alternative in light of the RTX 4090's scarcity and inflated pricing. The Dell sales advisory also lists that sales of the aforementioned products are banned in 22 countries, including Russia, Iran, Iraq, and others listed below.

AMD is working hard on delivering next-generation products, and today, its Linux team has submitted a few interesting patches that made a subtle appearance through recent GitHub patches for GFX12 targets, as reported by Phoronix. These patches have introduced two new discrete GPUs into the LLVM compiler for Linux, fueling speculation that these will be the first iterations of the RDNA4 graphics architecture, potentially being a part of the Radeon RX 8000 series of desktop graphics cards. The naming scheme for these new targets, GFX1200 and GFX1201, suggests a continuation of AMD's logical progression through graphics architectures, considering the company's history of associating RDNA1 with GFX10 and following suit with subsequent generations, like RDNA2 was GFX10.2 and RDNA3 was GFX11.

The development of these new GPUs is still in the early stages, indicated by the lack of detailed information about the upcoming graphics ISA or its features within the patches. Currently, the new GFX12 targets are set to be treated akin to GFX11 as the patch notes that "For now they behave identically to GFX11," implying that AMD is keeping the specifics under wraps until closer to release. The patch that defines target names and ELF numbers for new GFX12 targets GFX1200 and GFX1201 is needed in order to enable timely support for AMD ROCm compute stack, the AMDVLK Vulkan driver, and the RadeonSI Gallium3D driver.

ASRock, the leading global motherboard, graphics card, and mini PC manufacturer, today announced the ASRock Radeon RX 7900 XT Phantom Gaming White 20 GB OC. This graphics card is powered by the AMD Radeon RX 7900 XT GPU and built on 5 nm process. The new design pattern, featuring a white color scheme, is for enthusiasts seeking a white-themed PC. ASRock Radeon RX 7900 XT Phantom Gaming White 20 GB OC Graphics card has 84 compute units and boasts 5376 stream processors with a Boost Clock of up to 2450 MHz and a Game Clock of up to 2075 MHz. With 20 GB of GDDR6 high-speed memory clocked at 20 Gbps Effective and 80 MB of Infinity Cache, delivering 4K gaming at the highest settings.

ASRock Radeon RX 7900 XT Phantom Gaming White 20 GB OC graphics cards use the triple-fan Phantom Gaming 3X cooling system with the customized 100/100/100 mm Striped Ring Fans, Air Deflecting Fin, and Ultra-fit Heatpipe to provide superior cooling performance. ARGB LEDs that support Polychrome SYNC allow users to customize lighting effects, while the metal frame and backplate prevent the PCB from bending. The Super Alloy Graphics Card components provide stability for long-term operation. With its great performance and rich features, it can fully satisfy the needs of gamers.

Intel is slowly preparing to launch its next-generation Meteor Lake mobile processor family, dropping the Core i brand name in favor of Core Ultra. Today, we are witnessing some early Geekbench v6 benchmarks with the latest leak of the Core Ultra 7 155H processor, boasting an integrated Arc GPU featuring 8 Xe-Cores—the complete configuration expected in the GPU tile. This tile is also projected to be a part of the more potent Core 9 Ultra 185H CPU. The Intel Core Ultra 7 155H processor has been benchmarked in the new ASUS Zenbook 14, which houses a 16-core and 22-thread hybrid CPU configuration capable of boosting up to 4.8 GHz. Paired with 32 GB of memory, the configuration was well equipped to supply CPU and GPU with sufficient memory space.

Perhaps the most interesting information from the submission was the OpenCL score of the GPU. Clocking in at 33948 points in Geekbench v6, the GPU is running over AMD's Radeon 780M GPU found in APU solutions like AMD Ryzen 9 7940HS and Ryzen 9 7940U, which scored 30585 and 27345 points in the same benchmark, respectively. The GPU tile is millimeters away from closing the gap between itself and the desktop Intel Arc A380 discrete GPU, which scored 37105 points for less than a 10% difference. The Xe-LPG GPU version is bringing some interesting performance points for the integrated GPU platform, which means that Intel's Meteor Lake SKUs will bring more performance/watt than ever.

MAINGEAR, the leader in premium-quality, high-performance, custom PCs, today announced the launch of its latest lineup of Pro Series Workstation PCs, meticulously engineered and configurable with the industry's most powerful components, to cater to the diverse needs of professionals across multiple industries.

Ideal for game developers, photo editors, graphics designers, videographers, 3D rendering artists, music producers, CAD engineers, data scientists, and AI/Machine Learning developers, the MAINGEAR ProWS Series introduces a range of desktop workstations crafted to crush the most intensive tasks, elevate productivity and streamline workflow.

ASUS today announced that an international group of elite overclockers has set nine world records and achieved sixteen first-place finishes in a variety of performance benchmarks with ASUS Pro WS WRX90E-SAGE SE and ASUS Pro WS TRX50-SAGE WIFI workstation motherboards. The motherboards leveraged AMD Ryzen Threadripper PRO 7000 WX-Series and AMD Ryzen Threadripper 7000 series processors which featured up to 96 cores and 192 threads. The group of overclockers included Der8auer, Elmor, Massman, OGS, Safedisk, Seby and Shamino.

Ready for CPU and memory overclocking
For AMD Threadripper 7000 series workstations and high-end desktop (HEDT) models, AMD offers CPU and DRAM overclocking options. To fully leverage the performance of the AMD processors, ASUS has incorporated an innovative dual power supply design on Pro WS WRX90E-SAGE SE and Pro WS TRX50-SAGE WIFI motherboards. Along with robust power delivery and advanced cooling modules, these boards exhibit rock-solid stability during overclocking.

G.SKILL International Enterprise Co., Ltd., the world's leading brand of performance overclock memory and PC components, is excited to announce the Zeta R5 Neo series overclocked DDR5 R-DIMM memory kits for the latest AMD Ryzen Threadripper 7000 series and AMD Ryzen Threadripper Pro 7000 WX-series processors with AMD TRX50 chipset motherboards. This is the first R-DIMM platform from AMD that officially supports memory overclocking, so the Zeta R5 Neo features AMD EXPO memory overclocking profiles for easy memory overclocking. Built with hand-screened ICs and will be available in DDR5-6400 CL32 16 GB x4 and 32 GB x4 kit specifications at launch, Zeta R5 Neo series is the ideal R-DIMM memory solution to build an overclocked performance AMD workstation.

Overclocked DDR5-6400 R-DIMM Kits Designed for Quad-Channel AMD TRX50 Platform
With the launch of the AMD TRX50 workstation platform that supports DDR5 R-DIMM overclocking, G.SKILL is bringing AMD EXPO memory overclocking profile support to its R-DIMM memory lineup under the Zeta R5 Neo series. Available in quad-channel configurations of DDR5-6400 CL32-39-39-102 64 GB (16 GB x4) and 128 GB (32 GB x4) specifications, visit the G.SKILL product pages for more product information.

Leading global motherboard manufacturer, ASRock, today launched ASRock AI QuickSet software tool. ASRock AI QuickSet software tool can help users quickly download, install and set up artificial intelligence (AI) applications. The first version is launched based on the Microsoft Windows 11 64-bit platform and utilizes the powerful computing performance of ASRock's own AMD Radeon RX 7000 series graphics cards to optimize the operating performance of two well-known open source artificial intelligence (AI) drawing applications, Shark and Stable Diffusion web UI, so that interested users can quickly experience the fun of artificial intelligence (AI) at their fingertips.

Today at Microsoft Ignite, AMD and Microsoft featured how AMD products, including the upcoming AMD Instinct MI300X accelerator, AMD EPYC CPUs and AMD Ryzen CPUs with AI engines, are enabling new services and compute capabilities across cloud and generative AI, Confidential Computing, Cloud Computing and smarter, more intelligent PCs.

"AMD is fostering AI everywhere - from the cloud, to the enterprise and end point devices - all powered by our CPUs, GPUs, accelerators and AI engines," said Vamsi Boppana, Senior Vice President, AI, AMD. "Together with Microsoft and a rapidly growing ecosystem of software and hardware partners, AMD is accelerating innovation to bring the benefits of AI to a broad portfolio of compute engines, with expanding software capabilities."

TYAN, an industry leader in server platform design and a subsidiary of MiTAC Computing Technology Corporation, debuts its new server line-up for 4th Gen AMD EPYC & AMD Ryzen Processors at SC23, Booth #1917, in the Colorado Convention Center, Denver, CO, November 13-16.

AMD EPYC 9004 processor features leadership performance and is optimized for a wide range of HPC, cloud-native computing and Generative AI workloads
TYAN offers server platforms supporting the AMD EPYC 9004 processors that provide up to 128 Zen 4C cores and 256 MB of L3 Cache for dynamic cloud-native applications with high performance, density, energy efficiency, and compatibility.

During its Ignite conference, Microsoft introduced a duo of custom-designed silicon made to accelerate AI and excel in cloud workloads. First of the two is Microsoft's Azure Cobalt 100 CPU, a 128-core design that features a 64-bit Armv9 instruction set, implemented in a cloud-native design that is set to become a part of Microsoft's offerings. While there aren't many details regarding the configuration, the company claims that the performance target is up to 40% when compared to the current generation of Arm servers running on Azure cloud. The SoC has used Arm's Neoverse CSS platform customized for Microsoft, with presumably Arm Neoverse N2 cores.

The next and hottest topic in the server space is AI acceleration, which is needed for running today's large language models. Microsoft hosts OpenAI's ChatGPT, Microsoft's Copilot, and many other AI services. To help make them run as fast as possible, Microsoft's project Athena now has the name of Maia 100 AI accelerator, which is manufactured on TSMC's 5 nm process. It features 105 billion transistors and supports various MX data formats, even those smaller than 8-bit bit, for maximum performance. Currently tested on GPT 3.5 Turbo, we have yet to see performance figures and comparisons with competing hardware from NVIDIA, like H100/H200 and AMD, with MI300X. The Maia 100 has an aggregate bandwidth of 4.8 Terabits per accelerator, which uses a custom Ethernet-based networking protocol for scaling. These chips are expected to appear in Microsoft data centers early next year, and we hope to get some performance numbers soon.

Researchers at Graz University of Technology and the Helmholtz Center for Information Security have released their paper on CacheWarp—the latest vulnerability affecting some of the prior generation AMD EPYC CPUs. Titled CVE-2023-20592, the exploit targets first-generation EPYC Naples, second-generation EPYC Rome, and third-generation EPYC Milan. CacheWarp operates by exploiting a vulnerability in AMD's Secure Encrypted Virtualization (SEV) technology, specifically targeting the SEV-ES (Encrypted State) and SEV-SNP (Secure Nested Paging) versions. The attack is a software-based fault injection technique that manipulates the cache memory of a virtual machine (VM) running under SEV. It cleverly forces modified cache lines of the guest VM to revert to their previous state. This action circumvents the integrity checks that SEV-SNP is designed to enforce, allowing the attacker to inject faults without being detected.

Unlike attacks that rely on specific guest VM vulnerabilities, CacheWarp is more versatile and dangerous because it does not depend on the characteristics of the targeted VM. It exploits the underlying architectural weaknesses of AMD SEV, making it a broad threat to systems relying on this technology for security. The CacheWarp attack can bypass robust security measures like encrypted virtualization, posing a significant risk to data confidentiality and integrity in secure computing environments. AMD has issued an update for EPYC Milan with a hot-loadable microcode patch and updated the firmware image without any expected performance degradation. And for the remaining generations, AMD states that no mitigation is available for the first or second generations of EPYC processor (Naples and Rome) since the SEV and SEV-ES features are not designed to protect guest VM memory integrity, and the SEV-SNP is not available.