yuanjia111:develop - Branch - OpenMathLib/OpenBLAS - CodSpeed

Blog Docs Changelog

[ARM64] Add optimized fp16 shgemm kernels for Neoverse N2

#5716Merged

Comparing

yuanjia111:develop

(

e6eba9f

) with

develop

(

2671786

)

0%

Untouched: 62

Benchmarks

62 total

test_nrm2[100-d]

benchmark/pybench/benchmarks/bench_blas.py

-1%

38.6 µs39 µs

test_daxpy[100-c]

benchmark/pybench/benchmarks/bench_blas.py

-1%

25.3 µs25.5 µs

test_nrm2[100-dz]

benchmark/pybench/benchmarks/bench_blas.py

-1%

28.9 µs29.2 µs

test_daxpy[1000-c]

benchmark/pybench/benchmarks/bench_blas.py

-1%

32.8 µs33 µs

test_dot[100]

benchmark/pybench/benchmarks/bench_blas.py

-1%

22.4 µs22.6 µs

test_daxpy[100-d]

benchmark/pybench/benchmarks/bench_blas.py

-1%

24.3 µs24.5 µs

test_daxpy[100-z]

benchmark/pybench/benchmarks/bench_blas.py

-1%

26.1 µs26.2 µs

test_daxpy[100-s]

benchmark/pybench/benchmarks/bench_blas.py

-1%

24.2 µs24.3 µs

test_dgemv[100-d]

benchmark/pybench/benchmarks/bench_blas.py

-1%

140.9 µs141.7 µs

test_dgemv[100-c]

benchmark/pybench/benchmarks/bench_blas.py

-1%

149.5 µs150.3 µs

test_syrk[100-d]

benchmark/pybench/benchmarks/bench_blas.py

0%

341.4 µs339.9 µs

test_daxpy[1000-d]

benchmark/pybench/benchmarks/bench_blas.py

0%

32.6 µs32.7 µs

test_nrm2[1000-dz]

benchmark/pybench/benchmarks/bench_blas.py

0%

35.6 µs35.8 µs

test_daxpy[1000-s]

benchmark/pybench/benchmarks/bench_blas.py

0%

27.7 µs27.9 µs

test_dgbmv[1-100-c]

benchmark/pybench/benchmarks/bench_blas.py

0%

40.4 µs40.6 µs

test_nrm2[1000-d]

benchmark/pybench/benchmarks/bench_blas.py

0%

30.6 µs30.7 µs

test_daxpy[1000-z]

benchmark/pybench/benchmarks/bench_blas.py

0%

40.8 µs41 µs

test_dgbmv[1-100-d]

benchmark/pybench/benchmarks/bench_blas.py

0%

38.2 µs38.1 µs

test_syrk[100-c]

benchmark/pybench/benchmarks/bench_blas.py

0%

473.9 µs472.6 µs

test_gesdd[mn0-s]

benchmark/pybench/benchmarks/bench_blas.py

0%

109.6 µs109.4 µs

test_dgbmv[1-100-z]

benchmark/pybench/benchmarks/bench_blas.py

0%

42.4 µs42.4 µs

test_dgbmv[1-1000-s]

benchmark/pybench/benchmarks/bench_blas.py

0%

75.2 µs75.3 µs

test_dot[1000]

benchmark/pybench/benchmarks/bench_blas.py

0%

28.5 µs28.5 µs

test_gesv[100-d]

benchmark/pybench/benchmarks/bench_blas.py

0%

394.9 µs394.3 µs

test_gesdd[mn0-d]

benchmark/pybench/benchmarks/bench_blas.py

0%

120 µs120.1 µs

Commits

Click on a commit to change the comparison range

Base

develop

2671786

-0.15%

Add optimized FP16 shgemm for for NEOVERSEN2 target

e6eba9f

4 months ago

by yuanjia111

© 2026 CodSpeed Technology

Home Terms Privacy Docs