Portable and architecture-tuned GEMM microkernels evaluating RISC-V Vector (RVV 1.0 on SpacemiT X100, X60 & A100), x86 AVX2/FMA & AVX512/FMA (Intel Meteor Lake / AMD Zen4) & ARM Neon (Cortex-A76, Apple Sillicon Firestorm) via MIPPv2
hpc simd high-performance-computing avx2 vectorization avx512 gemm arm-neon dgemm micro-benchmarks firestorm rvv meteorlake cortex-a76 spacemit mipp zen4 riscv-vector
-
Updated
Sep 21, 2026 - C++