[llvm] [AArch64] Optimize when storing symmetry constants (PR #93717)
via llvm-commits
llvm-commits at lists.llvm.org
Tue Jun 18 07:05:26 PDT 2024
ParkHanbum wrote:
I got the pass in test-suite again.
in changed environment from MacOSX to Linux with docker.
```
/work/llvm/build-linux/test-suite-build/../../test-suite/utils/compare.py:206: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
name0 = names[0]
Program exec_time
results
LoopVector...meChecks4PointersDBeforeA/1000 1747554.95
LoopVector...imeChecks4PointersDAfterA/1000 394555.76
LoopVector...meChecks4PointersDEqualsA/1000 391361.31
LoopVector...tersAllDisjointIncreasing/1000 389818.10
LoopVector...tersAllDisjointDecreasing/1000 387404.53
LCALS/Subs...test:BM_MAT_X_MAT_LAMBDA/44217 200970.88
LCALS/Subs...aw.test:BM_MAT_X_MAT_RAW/44217 185668.65
harris/har...est:BENCHMARK_HARRIS/2048/2048 72338.17
ImageProce...t:BENCHMARK_GAUSSIAN_BLUR/1024 37052.50
LoopVector...timeChecks4PointersDBeforeA/32 33376.06
LoopVector...LoopFrom_uint64_t_To_uint32_t_ 22920.46
LoopVector...VW16From_uint64_t_To_uint32_t_ 20571.81
LoopVector...hVW8From_uint64_t_To_uint32_t_ 19181.25
LoopVector...hVW16From_uint64_t_To_uint8_t_ 17468.18
LoopVector...LoopFrom_uint64_t_To_uint16_t_ 17304.87
exec_time
run results
count 4.260000e+03
mean 1.104730e+03
std 2.967942e+04
min 0.000000e+00
25% 4.000000e-04
50% 1.360000e-02
75% 9.477289e-02
max 1.747555e+06
root at 6edef4f7cbda:/work/llvm/build-linux/test-suite-build# uname -a
Linux 6edef4f7cbda 6.4.16-linuxkit #1 SMP PREEMPT Thu Nov 16 10:49:20 UTC 2023 aarch64 aarch64 aarch64 GNU/Linux
```
@davemgreen please let me know still problem occur.
https://github.com/llvm/llvm-project/pull/93717
More information about the llvm-commits
mailing list