issues
search
google
/
XNNPACK
High-efficiency floating-point neural network inference operators for mobile, server, and Web
Other
1.89k
stars
376
forks
source link
F32-GEMM AVX512 generate up to 16x64
#7423
Closed
copybara-service[bot]
closed
1 week ago
copybara-service[bot]
commented
2 weeks ago
F32-GEMM AVX512 generate up to 16x64
add MR 9-16 and NR 32 and 64
F32-GEMM AVX512 generate up to 16x64