AMD Upstreams AVX10_V2_AUX Support Into LLVM Clang 24

AMD has upstreamed initial support for AVX10_V2_AUX into LLVM/Clang, merged into LLVM 24 Git. The new instruction set extends AVX10 with data processing and format conversion instructions for AI workloads, including conversion to compressed formats like FP4 and FP6, OCP microscaling formats, and new rounding modes. It is exposed via a new -mavx10v2aux compiler flag and spans roughly 15,000 lines of code across 20 patches. This is part of the AI Compute Extensions (ACE) effort jointly developed by AMD and Intel under the x86 Ecosystem Advisory Group. AMD is also working on equivalent support for GCC. LLVM/Clang 24.1 stable is expected around next March.

Source: https://www.phoronix.com/news/LLVM-Clang-AVX10_V2_AUX. 8 Sync News only summarizes and links out; content copyright belongs to the authors and original sources.