Cisco submitted the industry's first multi-vendor accelerator entry to MLPerf Inference v6.1, running a distributed inference service across NVIDIA H200 and AMD Instinct MI350X systems over a Cisco Si…
Nvidia's GeForce RTX 60 series, built on the architecture codenamed Rubin, is now targeting a 2028 launch window, according to leakers Kopite7kimi and Kepler_L2, a slip from earlier projections of 202…
Europe is "entirely supply constrained" on digital compute infrastructure, General Tensor chief investment officer Victor Teixeira said at the 12th European Blockchain Convention in Barcelona, citing …
Modal, Fireworks AI, and Baseten are offering hosted inference for Moonshot AI's Kimi K3 at roughly one-tenth the cost of direct access, using Nvidia GB300 NVL72 systems and AMD MI350X and MI355X acce…
AMD will showcase physical AI and robotics demonstrations at ROSCon 2026, held Sept. 22-24 at the Sheraton Centre Toronto, as a platinum sponsor. AMD Fellow Dr. Giulio Corradi will join the opening ke…
AMD released ROCm Hyperloom, an open-source multi-agent harness that delivered a median 1.73× inference speedup on AMD Instinct GPUs, with gains ranging from 1.35× to 7.31× in unattended evaluation. T…
A Level1Techs forum user reported that running two AMD Radeon Pro AI R9700 GPUs at 210W with fans at roughly 33% and reduced clock speeds makes the cards silent at the cost of some prompt-processing p…
A developer released two open-source Python tools aimed at simplifying local LLM inference on AMD GPUs. ROCmFix automatically resolves HSA_OVERRIDE_GFX_VERSION issues by querying PCI IDs via the Windo…
AMD published a white paper claiming its 256-core 6th Gen EPYC 9006 "Venice" server CPU delivers 2.24x the platform-level performance of NVIDIA's 88-core Vera in SPECrate 2026 Integer, with a 96-core …
Taiwanese server and storage provider AIC is exhibiting server motherboards powered by AMD EPYC processors and high-availability storage servers at the AMD Embedded Summit Taipei 2026, extending a lon…
A developer added native XDNA compilation and experimental execution to the Loom compiler stack, then used it to run Doom with the game's world rasterized by ten resident NPU tiles on a Ryzen AI Max+ …
SemiAnalysis, the semiconductor and AI infrastructure newsletter founded by Dylan Patel in 2020, is projected to surpass $100 million in revenue this year, with the vast majority of the world's larges…
Honor has revealed the Tiangong AXB35, an AI workstation platform for enterprise use in a 2 L small-form-factor chassis measuring 70 × 192 × 203 mm, offered in Standard, Pro, and Ultra variants. The A…
Honor has revealed the Tiangong AXB35, a 2-liter small-form-factor AI workstation for enterprise use that ships in three variants: a Standard model with Intel's Core Ultra X7 358H CPU and Arc B390 iGP…
A user running llama.cpp in tensor split-mode on an ASUS ProArt X870E motherboard with two AMD R9700 GPUs reports that system suspend hangs instead of entering sleep mode, requiring a power cycle, eve…
A builder documented a dual-boot Ubuntu 24.04/Windows workstation running four NVIDIA RTX6000 Blackwell Pro Max-Q GPUs (96 GB each, 384 GB total) with an AMD Ryzen Threadripper PRO 7985WX and 512 GB o…
A developer released gemmcheck3, a self-checking rocBLAS GEMM stress test for AMD GPUs targeting RDNA3/gfx11 hardware such as the RX 7900 XTX. The tool runs five rocblas_gemm_ex calls in LLM layer sha…
Hugging Face's hardware list for its Fit Finder omits the AMD RX 5600 XT even though it includes the RX 5500 XT, RX 5700 and RX 5700 XT, a gap a forum user flagged as inconsistent. A respondent said t…
TSMC's CoWoS-L is expected to remain the mainstream advanced packaging solution for AI chips through 2028 despite competition from Intel's EMIB-T, according to TrendForce research, which cites CoWoS-L…
SemiAnalysis reported that DeepSeek's Engram architecture, which extends standard token embeddings with learned multi-token lookups, can lower required HBM capacity for models at the same quality by o…