scalar_ignore_nan_bug - Branch - jvdd/argminmax

fix: allow double update in scalar ignore nan implementation

#40Merged

Comparing

scalar_ignore_nan_bug

(

6ba2249

) with

main

(

5034986

)

-6%

Regression: 2

Untouched: 50

Benchmarks

52 total

scalar_f32_argminmax_in

benches/bench_f32_ignore_nan.rs::benches::argminmax_in_f32_random_array_long

-6%

442.9 µs471.3 µs

scalar_f64_argminmax_in

benches/bench_f64_ignore_nan.rs::benches::argminmax_in_f64_random_array_long

-4%

629.6 µs658 µs

avx2_u8_argminmax

benches/bench_u8.rs::benches::argminmax_u8_random_array_long

73.9 µs73.9 µs

avx2_f16_argminmax_rn

benches/bench_f16_return_nan.rs::benches::argminmax_rn_f16_random_array_long

119.3 µs119.3 µs

impl_f16_argminmax_rn

benches/bench_f16_return_nan.rs::benches::argminmax_rn_f16_random_array_long

119.5 µs119.5 µs

impl_f32_argminmax_in

benches/bench_f32_ignore_nan.rs::benches::argminmax_in_f32_random_array_long

225.1 µs225.2 µs

sse_f16_argminmax_rn

benches/bench_f16_return_nan.rs::benches::argminmax_rn_f16_random_array_long

159.1 µs159.2 µs

impl_u32_argminmax

benches/bench_u32.rs::benches::argminmax_u32_random_array_long

225.1 µs225.1 µs

avx_f32_argminmax_in

benches/bench_f32_ignore_nan.rs::benches::argminmax_in_f32_random_array_long

225 µs225 µs

avx_f64_argminmax_in

benches/bench_f64_ignore_nan.rs::benches::argminmax_in_f64_random_array_long

448.9 µs449 µs

impl_i32_argminmax

benches/bench_i32.rs::benches::argminmax_i32_random_array_long

225 µs225 µs

impl_i64_argminmax

benches/bench_i64.rs::benches::argminmax_i64_random_array_long

459.5 µs459.5 µs

sse_f64_argminmax_in

benches/bench_f64_ignore_nan.rs::benches::argminmax_in_f64_random_array_long

551.9 µs551.9 µs

sse_i64_argminmax

benches/bench_i64.rs::benches::argminmax_i64_random_array_long

608.6 µs608.5 µs

sse_u64_argminmax

benches/bench_u64.rs::benches::argminmax_u64_random_array_long

622.8 µs622.8 µs

scalar_f16_argminmax_rn

benches/bench_f16_return_nan.rs::benches::argminmax_rn_f16_random_array_long

634 µs634 µs

impl_u16_argminmax

benches/bench_u16.rs::benches::argminmax_u16_random_array_long

113.4 µs113.4 µs

sse_f32_argminmax_in

benches/bench_f32_ignore_nan.rs::benches::argminmax_in_f32_random_array_long

276.4 µs276.4 µs

scalar_f32_argminmax_rn

benches/bench_f32_return_nan.rs::benches::argminmax_rn_f32_random_array_long

528.1 µs528.1 µs

scalar_u8_argminmax

benches/bench_u8.rs::benches::argminmax_u8_random_array_long

359.7 µs359.7 µs

sse_i8_argminmax

benches/bench_i8.rs::benches::argminmax_i8_random_array_long

85.2 µs85.2 µs

impl_i8_argminmax

benches/bench_i8.rs::benches::argminmax_i8_random_array_long

85.3 µs85.3 µs

sse_u8_argminmax

benches/bench_u8.rs::benches::argminmax_u8_random_array_long

85.9 µs85.9 µs

impl_u8_argminmax

benches/bench_u8.rs::benches::argminmax_u8_random_array_long

86.1 µs86.1 µs

scalar_f64_argminmax_rn

benches/bench_f64_return_nan.rs::benches::argminmax_rn_f64_random_array_long

714.8 µs714.8 µs

Commits

Click on a commit to change the comparison range

Base

main

5034986

-21.22%

fix: allow double update in scalar ignore nan implementation

4929c9b

3 years ago

by jvdd

+15.18%

perf: optimize ignore nan scalar impl

56e06f1

3 years ago

by jvdd

test: add test for scalar ignore nan bug

6ba2249

3 years ago

by jvdd

Home Terms Privacy Docs