LLVM Clang Merges -mcpu=gb10 Support For NVIDIA GB10 Superchip
- Reference: 0001559270
- News link: https://www.phoronix.com/news/LLVM-Clang-NVIDIA-GB10
- Source link:
The NVIDIA GB10 superchip coming in devices like the ASUS Ascent GX10 AI supercomputer, DGX Spark, and other devices for developers / data scientists / researchers now has optimized support available within the LLVM Clang C/C++ compiler.
[1]This patch was merged to LLVM Git today for LLVM/Clang 21 in enabling -mcpu=gb10 support for the NVIDIA GB10 Grace Blackwell Superchip. The patch sums up:
[AArch64] Add support for -mcpu=gb10. (#146515)
This patch adds support for -mcpu=gb10 (NVIDIA GB10). This is a big.LITTLE cluster of Cortex-X925 and Cortex-A725 cores. The appropriate MIDR numbers are added to detect them in -mcpu=native.
We did not add an -mcpu=cortex-x925.cortex-a725 option because GB10 does include the crypto instructions which we want on by default, and the current convention is to not enable such extensions for Arm Cortex cores in -mcpu where they are optional in the IP.
The NVIDIA GB10 consists of 20 Arm cores between ten Cortex-X925 cores and ten smaller Cortex-A725 cores.
LLVM 21 stable release is expected around September.
[1] https://github.com/llvm/llvm-project/commit/84e54515bc4e9dd4938121f4df7cc27bb89a0a43
sophisticles