Stories

  1. Image: TechPowerUp

    AMD's desktop CPU share falls to 29.9% in Q3 2026 as Intel regains ground, Mercury Research says

    AMD's share of desktop CPU shipments dropped to 29.9% in the third quarter of 2026, down from 34.9% in Q2, according to Mercury Research data, while Intel rose to 70.1%. It is AMD's lowest desktop figure since early 2025, when it held about 28%, and compares with 33.6% a year earlier. Mercury is unsure of the cause but suggests weaker laptop demand let Intel move more chips to desktops during the seasonal Q3 bump, while AMD's reliance on PC gaming, hit by gaming GPU supply constraints, weighed on it.

  2. Image: TechPowerUp

    Noctua releases NT-CP1 Carbice carbon nanotube thermal pad for AMD AM5 and AM4

    Noctua has introduced the NT-CP1 AM5/4 thermal pad for AMD AM5 and AM4 processors, using Carbice's carbon nanotube technology originally developed for aerospace. The pad sells for $24.95 in the US and €24.95 in Europe. Noctua says its performance improves over the first few hundred to few thousand thermal cycles as the nanotubes conform to the heatspreader surface, then stays stable for years, avoiding the pump-out and cracking that degrade pastes.

  3. ASRock releases version 1.0.6 update after 1.0.5 bricked some Radeon RX 7900 XT and XTX cards

    ASRock has quietly posted a version 1.0.6 update for its Radeon RX 7000-series graphics cards, following reports that the earlier 1.0.5 release bricked some RX 7900 XT and RX 7900 XTX models. Accounts differ on whether 1.0.6 is a new VBIOS or a new release of the ASRVBFlash flashing utility. ASRock has published no changelog. The original update was meant to apply AMD's VBIOS security patch, and affected users had restored older VBIOS images to recover systems that failed to boot.

  4. Image: Tom's Hardware

    AMD executives say Zen 6 Venice EPYC chips were not purpose-built for agentic AI, rebuking Nvidia and Arm

    AMD corporate VP Madhu Rangarajan and chief Zen architect Mike Clark said in separate talks that the company did not design its Zen 6 EPYC 9006 Venice server CPUs specifically for agentic AI, a jab at Nvidia and Arm, which market their latest data center CPUs as agentic. Rangarajan still called EPYC 9006 the best CPU for the agentic era, arguing agentic workflows need compute for diverse workloads. Clark said AMD did not foresee agentic computing but always expected CPUs to keep a role in AI software.

  5. Image: Chips and Cheese

    AMD launches 6th Gen EPYC Zen 6 'Venice' server CPUs and EPYC Embedded 9006 series

    AMD announced general availability of its 6th Gen EPYC SP7 server processors based on the Zen 6 architecture, with systems shipping from leading server vendors and Microsoft Azure and OCI instances coming soon. It also announced the EPYC Embedded 9006 Series with up to 256 cores, 96 PCIe Gen 6 lanes, CXL 3.1 and sixteen DDR5 channels; the 128-core part ships November 15. AMD claims up to 1.8x the throughput of the EPYC 9005 Series. Zen 6 uses TSMC's N2P node, and the high-frequency EPYC 9686F offers 96 cores at up to 5.3GHz.

  6. Image: Phoronix

    AMD posts GFX13 support patches for AMDGPU Linux driver, likely for RDNA 5

    AMD has posted patches adding GFX13 graphics IP support to the open-source AMDGPU Linux kernel driver, with GFX13 widely expected to underpin RDNA 5. The eight-patch series from maintainer Alex Deucher adds over 12,600 lines of code excluding headers, introducing version 13 of the GFX, MES and IMU blocks plus initial GC 13 compute support in AMDKFD. It references GFX 13.0.1 and GFX 13.0.4 targets and follows earlier GFX13 work in the AMDGPU LLVM back-end.

  7. Image: TechPowerUp

    Google adds TPU 8tx with AMD EPYC Turin hosts and confidential computing, and TPU 8i Max for inference

    Google has extended its Tensor Processing Unit line with two new systems, the TPU 8tx and the TPU 8i Max. The TPU 8tx pairs AMD EPYC Turin host processors with confidential computing, while the TPU 8i Max doubles memory bandwidth for inference workloads. Google said the systems respond to external customers asking for accelerators that can reason across large AI model context windows, and to highly regulated customers seeking specialized encryption. The company tied the launch to recent PyTorch work aimed at making TPUs more accessible to developers and Cloud customers.

  8. Image: Tom's Hardware

    AMD acknowledges Radeon External Events Utility bug causing continuous SSD writes

    AMD has acknowledged a bug in its External Events Utility, installed alongside Radeon drivers, that users say writes data to SSDs continuously even when PCs are idle. One user reported sustained writes of about 5MB/s at idle, and another said SSD health dropped about 1% per month over 16 months. In an October 21 forum reply, an AMD representative said the company aims to resolve the issue in a future driver release; the problem drew renewed attention on November 8.

  9. Image: The Register

    Huawei launches Qingyun HM650 desktop it says uses only Chinese-made components

    Huawei has launched the Qingyun HM650 desktop PC, which it claims is built entirely from components manufactured in China. It runs on Huawei's 10-core, 20-thread Kirin X90 processor and an enterprise version of HarmonyOS, with 16 GB of LPDDR5 memory and a 256GB or 512GB SSD. One Chinese government agency has reportedly ordered 600 units as Beijing pushes government bodies away from Wintel machines. It remains unclear whether all the components use Chinese designs, and the move threatens Intel, AMD and Qualcomm sales.

  10. Image: The Next Platform

    Oak Ridge unveils IBM-AMD platform for quantum error-correction research

    Oak Ridge National Laboratory’s Quantum Science Center has unveiled a co-design platform for researching quantum error correction, combining IBM quantum processors with a high-performance computing system based on AMD silicon. The platform brings quantum and conventional computing together to study how to protect fragile quantum information from environmental noise. QSC director Travis Humble said solving error correction challenges is essential to making quantum computing useful for scientific discovery. IBM’s Jay Gambetta said the focus is now on accelerating and optimizing error correction.

  11. AMD FSR Upscaling 4.1.2 arrives on Xbox Series X|S

    AMD has made FSR Upscaling 4.1.2 available for Xbox Series X|S through the October 2026 Game Development Kit. The release includes a full-scale INT8 model and a more efficient Lite INT8 model, both supported on the consoles, with developers able to choose which to implement. AMD also added support through FSR SDK 2.4 for Ryzen AI Z2 Extreme and Ryzen Z2 A processors used in ROG Xbox Ally handhelds. The Xbox implementation uses the same FSR SDK API as PC.

  12. Image: Ars Technica

    AMD reports record $13.4 billion third-quarter revenue as data center sales nearly double

    AMD reported record third-quarter revenue of $13.4 billion, up 45 percent year over year, as its data center business nearly doubled. The segment brought in a record $8.5 billion, a 97 percent increase, which AMD attributed to demand for EPYC server CPUs and the launch of Instinct MI450 GPUs. The company forecast fourth-quarter revenue of about $17 billion, plus or minus 3 percent, with growth primarily driven by data center demand. Its gaming segment declined, while Ryzen mobile chip sales supported growth in the client business.

  13. Image: TechPowerUp

    Nvidia RTX 50-series cards face price increases of up to $500

    Nvidia’s GeForce RTX 50-series graphics cards are expected to become more expensive this week, with increases varying by model. One report says the RTX 5090 DV2 could rise by as much as 3,300 RMB, nearly $500, while the RTX 5060 could increase by about $30. The changes are attributed to demand outstripping supply; consumer prices may take a week or two to reflect the new listings. Another report also notes that Intel and AMD have not disclosed pricing plans.

  14. Image: Phoronix

    System76 unveils compact Thelio Prime Linux desktop

    System76 has unveiled the Thelio Prime, an 11-liter Linux desktop built at its Denver facility, and says the AMD-powered system will begin shipping in November. Configurations offer AMD Ryzen 9000 processors, up to a Ryzen 7 9850X3D, optional AMD or NVIDIA graphics up to an RTX 5070, as much as 96 GB of DDR5 memory and up to 8 TB of storage. The compact chassis uses two thermal chambers to separate the CPU and GPU, with networking that includes 2.5 GbE, Wi-Fi 6E and Bluetooth 5.2.

  15. Image: The Guardian Technology

    Meta plans $95 billion stock offering to fund AI expansion

    Meta announced plans to raise $95 billion through a stock offering to fund its AI expansion. The company is investing multibillion-dollar sums in data centers, and the fundraising is intended to support AI infrastructure and superintelligence. Nvidia, AMD and Broadcom are eyeing up to $5 billion in shares, while Elon Musk has warned that funding for smaller AI companies is drying up.

  16. Image: Phoronix

    AMD Posts Linux Kernel Patches for Ryzen Overclocking Driver

    AMD has posted Linux kernel patches for its AMD Overclocking Driver (AOD), an interface for tuning Ryzen CPUs through firmware and sysfs. The initial patches expose PPT, TDC and EDC power and current limits, their supported ranges, platform-level SMT toggling and a readout of memory clock. The power-limit changes take effect only after a reboot, and the patches do not bring the Ryzen Master application to Linux. The driver provides a kernel-side interface for Linux tools to use.

  17. Image: Phoronix

    AMD and Intel are combining CPU optimizations for PyTorch in oneDNN

    AMD and Intel are combining their CPU optimizations for PyTorch in the oneDNN library, creating a shared path for tuning software on both companies’ processors. AMD’s ZenDNN, a downstream oneDNN version tailored to its Zen processors, is moving into the open-source project. AMD also joined the UXL Foundation as a steering member, with an annual contribution of $20,000. The companies said the integration was recommended by the x86 Ecosystem Advisory Group, established in 2024.

  18. Image: TechPowerUp

    AMD brings FSR 4.1 to Radeon RX 6000 GPUs and Xbox platforms

    AMD is expanding its FSR 4.1 upscaling technology to Radeon RX 6000-series graphics cards, Ryzen-powered devices and Xbox consoles, extending it beyond Radeon RX 9000 GPUs. The technology is available on RX 6000 cards and some Ryzen AI devices through Adrenalin Edition 26.10.1 drivers, while Xbox Series X and S support is coming soon. AMD says FSR 4.1 can improve performance in more than 300 supported games; its handheld implementation uses a smaller model, and the update adapts the technology for older hardware.

  19. Image: HPCwire

    OCP Global Summit highlights infrastructure announcements and CXL memory demonstrations

    The OCP Global Summit in San Jose featured infrastructure announcements and demonstrations, including AMD’s Helion rack design and XCENA’s CXL-based memory solutions for AI inference. The event ran from October 12 to 15 at the San Jose McEnery Convention Center. XCENA said its demonstrations address the memory requirements and data-movement burden of AI inference.

  20. Image: The Register

    AMD unveils Helion open rack design for agentic AI infrastructure

    AMD introduced Helion, an open rack-scale reference design for agentic AI infrastructure, at the OCP Global Summit. Its two configurations include a CPU-only rack with up to 32,768 Zen 6 cores and an enterprise design combining EPYC processors with up to 16 Instinct MI350P GPUs. AMD says the blueprint builds on its Helios rack-scale platform and is based on an Open Compute Project design. Partners are expected to offer CPU-only and mixed CPU/GPU configurations in November.

Topics

Publishers