The OpenCL Working Group today published a draft extension, cl_khr_cooperative_matrix, which introduces cooperative matrix operations to OpenCL. Developed in collaboration with Arm, Intel, and Qualcom...
Ollama has released version 0.23.1, introducing Gemma 4 Multi-token Processing (MTP) speculative decoding for its MLX runner. This update, primarily benefiting macOS users with Apple Silicon, can prov...
The Khronos OpenCL Working Group has today released OpenCL 3.1, moving several previously optional capabilities into the core specification. Conformant implementations are now required to consume SPIR...
A Khronos contributed blog post published today, April 6, 2026, details how FFmpeg leverages Vulkan Compute shaders to accelerate video encoding and decoding. This method enables significant GPU compu...
AMD has recently released ROCm 7.2.1, an update to its open-source GPU compute stack designed for Radeon and Instinct GPUs. This release builds upon the ROCm 7.2 version, which was initially made avai...
AMD has released ROCm 7.2 today for Linux, following discussions about its features at CES earlier this month. This update to AMD's open-source GPU compute stack officially extends support to more Rad...
Phoronix has published a benchmark comparison evaluating the Dell Pro Max with NVIDIA GB10 against AMD's Ryzen AI Max+ 395 "Strix Halo" within a Framework Desktop SFF PC. The Dell Pro Max, which integ...
Canonical today announced plans to package and distribute the AMD ROCm software stack directly within the Ubuntu 26.04 LTS repositories. This integration, set for next year's long-term support release...
Ollama has recently released version 0.13.2-rc2, bringing several updates focused on AI model performance and GPU compatibility. A significant change is the default enablement of flash attention for v...
AMD GPUOpen recently announced AMD Schola v2, a significant update to its open-source reinforcement learning (RL) plugin for Unreal Engine 5. Schola v2 is described as a "next-generation reinforcement...
NVIDIA today released CUDA 13.1, which the company claims is "the largest and most comprehensive update to the CUDA platform since it was invented two decades ago." The most notable addition in this r...
The Khronos SYCL Working Group today released Revision 11 of the SYCL 2020 Specification. This update introduces eight new extensions, alongside several specification clarifications. Khronos stated th...