OpenBLAS 0.3.24 Released With Intel Sapphire Rapids Improvements, Apple M2 Detection

OpenBLAS 0.3.24 is now available for this latest open-source BLAS and LAPACK implementation known for its advanced CPU optimizations and extensive tuning for providing for very speedy linear algebra kernels.

OpenBLAS 0.3.24 was released on Sunday with many fixes, improved NVIDIA HPC compiler support, LLVM 17 compilation support including for its Flang Fortran compiler, CPU model detection for Intel Alder Lake N, various Intel Sapphire Rapids fixes/improvements, CPU model auto-detection for Apple M2 processors, various ARMv8 fixes, ABI detection for LoongArch64, and various RISC-V fixes. The NVIDIA HPC compiler work includes support within OpenBLAS' CMake builds, support for building AVX-512 kernels with this NVIDIA compiler, and also support for building Arm SVE kernels with this compiler.

Intel Xeon Scalable Sapphire Rapids processors now enjoy activation of the AMX tile for the SBGEMM kernel, run-time detection of Sapphire Rapids in the DYNAMIC_ARCH code path, and more.

The lengthy list of fixes and new features with the OpenBLAS 0.3.24 release can be found via GitHub as well as for downloading this leading open-source BLAS/LAPACK implementation for high performance computing.