ChipKerchner:addRVVVectorizedPacking - Branch - OpenMathLib/OpenBLAS

Add and use vectorized packing in ZVL128B and ZVL256B for RISCV

#5422Merged

Comparing

ChipKerchner:addRVVVectorizedPacking

(

64401b4

) with

develop

(

5e43ba9

)

Untouched: 62

Benchmarks

62 total

test_nrm2[100-dz]

benchmark/pybench/benchmarks/bench_blas.py

28.8 µs28.7 µs

test_gesv[100-d]

benchmark/pybench/benchmarks/bench_blas.py

395.5 µs394.7 µs

test_gesv[100-z]

benchmark/pybench/benchmarks/bench_blas.py

938.4 µs937.1 µs

test_nrm2[100-d]

benchmark/pybench/benchmarks/bench_blas.py

36.7 µs36.6 µs

test_syev[50-d]

benchmark/pybench/benchmarks/bench_blas.py

1.4 ms1.4 ms

test_daxpy[1000-s]

benchmark/pybench/benchmarks/bench_blas.py

27.3 µs27.3 µs

test_nrm2[1000-d]

benchmark/pybench/benchmarks/bench_blas.py

30.4 µs30.4 µs

test_dgemv[100-z]

benchmark/pybench/benchmarks/bench_blas.py

230.9 µs230.9 µs

test_gesv[1000-c]

benchmark/pybench/benchmarks/bench_blas.py

188.6 ms188.6 ms

test_syrk[1000-d]

benchmark/pybench/benchmarks/bench_blas.py

130.4 ms130.3 ms

test_syev[200-d]

benchmark/pybench/benchmarks/bench_blas.py

58.6 ms58.6 ms

test_gesv[1000-d]

benchmark/pybench/benchmarks/bench_blas.py

93.3 ms93.3 ms

test_gesdd[mn1-s]

benchmark/pybench/benchmarks/bench_blas.py

65.2 ms65.2 ms

test_daxpy[1000-z]

benchmark/pybench/benchmarks/bench_blas.py

40.4 µs40.4 µs

test_dgemv[1000-c]

benchmark/pybench/benchmarks/bench_blas.py

14.8 ms14.8 ms

test_syev[200-s]

benchmark/pybench/benchmarks/bench_blas.py

49.1 ms49.1 ms

test_gesv[1000-z]

benchmark/pybench/benchmarks/bench_blas.py

353.6 ms353.6 ms

test_syrk[1000-c]

benchmark/pybench/benchmarks/bench_blas.py

227.5 ms227.5 ms

test_gesdd[mn1-d]

benchmark/pybench/benchmarks/bench_blas.py

93.8 ms93.8 ms

test_dgemv[1000-s]

benchmark/pybench/benchmarks/bench_blas.py

7 ms7 ms

test_gemm[1000-s]

benchmark/pybench/benchmarks/bench_blas.py

117.4 ms117.4 ms

test_gemm[1000-z]

benchmark/pybench/benchmarks/bench_blas.py

875.6 ms875.6 ms

test_dgemv[1000-d]

benchmark/pybench/benchmarks/bench_blas.py

13.9 ms13.9 ms

test_daxpy[100-c]

benchmark/pybench/benchmarks/bench_blas.py

25 µs25 µs

test_daxpy[100-d]

benchmark/pybench/benchmarks/bench_blas.py

24 µs24 µs

Commits

Click on a commit to change the comparison range

Base

develop

5e43ba9

+0.04%

Add and use vectorized packing to ZVL128B and ZVL256B. Up to 3x+ faster than generic scalar functions.

c00afc8

6 months ago

by ChipKerchner

-0.11%

Disable vectorized packing for DGEMM - since it is slower than scalar.

64401b4

6 months ago

by ChipKerchner

Home Terms Privacy Docs