Work around miscompilation of the ARM64 non-SVE DDOT kernel
#5710Comparing
martin-frbg:issue5708
(e3ce462
) withdevelop
(b8697b3
)0%
- Untouched
- 62
Benchmarks
62 total
test_dgbmv[1-100-z]
benchmark/pybench/benchmarks/bench_blas.py+1%
42.6 µs42.3 µs
test_dgbmv[1-100-c]
benchmark/pybench/benchmarks/bench_blas.py+1%
40.7 µs40.5 µs
test_dot[1000]
benchmark/pybench/benchmarks/bench_blas.py+1%
28.7 µs28.5 µs
test_daxpy[100-d]
benchmark/pybench/benchmarks/bench_blas.py0%
24.5 µs24.4 µs
test_syrk[100-d]
benchmark/pybench/benchmarks/bench_blas.py0%
341.4 µs339.9 µs
test_gesdd[mn0-s]
benchmark/pybench/benchmarks/bench_blas.py0%
109.8 µs109.4 µs
test_gesv[100-s]
benchmark/pybench/benchmarks/bench_blas.py0%
258 µs257.2 µs
test_dgbmv[1-1000-d]
benchmark/pybench/benchmarks/bench_blas.py0%
83.9 µs83.6 µs
test_syrk[100-c]
benchmark/pybench/benchmarks/bench_blas.py0%
474.1 µs472.7 µs
test_daxpy[1000-c]
benchmark/pybench/benchmarks/bench_blas.py0%
33 µs32.9 µs
test_dgbmv[1-100-s]
benchmark/pybench/benchmarks/bench_blas.py0%
37.6 µs37.5 µs
test_dgbmv[1-100-d]
benchmark/pybench/benchmarks/bench_blas.py0%
38.1 µs38 µs
test_daxpy[1000-s]
benchmark/pybench/benchmarks/bench_blas.py0%
27.8 µs27.7 µs
test_nrm2[100-dz]
benchmark/pybench/benchmarks/bench_blas.py0%
29.1 µs29.1 µs
test_daxpy[1000-d]
benchmark/pybench/benchmarks/bench_blas.py0%
32.8 µs32.7 µs
test_nrm2[1000-dz]
benchmark/pybench/benchmarks/bench_blas.py0%
35.7 µs35.7 µs
test_dgbmv[1-1000-z]
benchmark/pybench/benchmarks/bench_blas.py0%
119.3 µs119.1 µs
test_nrm2[1000-d]
benchmark/pybench/benchmarks/bench_blas.py0%
30.7 µs30.7 µs
test_gesv[100-d]
benchmark/pybench/benchmarks/bench_blas.py0%
394.2 µs393.8 µs
test_dgemv[100-d]
benchmark/pybench/benchmarks/bench_blas.py0%
141.7 µs141.5 µs
test_gemm[100-s]
benchmark/pybench/benchmarks/bench_blas.py0%
273.6 µs273.4 µs
test_dgemv[100-z]
benchmark/pybench/benchmarks/bench_blas.py0%
231.3 µs231.2 µs
test_dgbmv[1-1000-c]
benchmark/pybench/benchmarks/bench_blas.py0%
99.7 µs99.7 µs
test_gesv[100-z]
benchmark/pybench/benchmarks/bench_blas.py0%
938.5 µs938.2 µs
test_dgemv[100-s]
benchmark/pybench/benchmarks/bench_blas.py0%
104.3 µs104.3 µs
Commits
Click on a commit to change the comparison rangeBase
develop
b8697b3
+0.09%
Use volatile attribute for SDOT only, to avoid creating new miscompilations
e3ce462
12 hours ago
by martin-frbg