ENH: Vectorize quicksort for 16-bit and 64-bit dtype using AVX512#22315

Merged

mattip merged 33 commits intonumpy:mainfrom

r-devulap:avxsort

Feb 15, 2023

Commits on Jan 30, 2023

ENH: Add x86-simd-sort source files
Raghuveer Devulapalli
committed
ENH: Add AVX-512 based 64-bit dtype sort
Raghuveer Devulapalli
committed
ENH: Add AVX-512 based 16-bit dtype sort
Raghuveer Devulapalli
committed
BUG: Use longlong when NPY_SIZEOF_LONG is 4
Raghuveer Devulapalli
committed
Revert "ENH: Add AVX-512 based 16-bit dtype sort"
Raghuveer Devulapalli
committed
BUG: Ensure long/longlong is 8 bytes for 64-bit qsort
Raghuveer Devulapalli
committed
MAINT: Force inline bitonic network functions

r-devulap
authored and
Raghuveer Devulapalli
committed
ENH: Use npyv_* for missing intrinsics in gcc-6
Raghuveer Devulapalli
committed
MAINT: Disable AVX-512 qsort on macOS and WIN32
Raghuveer Devulapalli
committed
BUG: Do not use a global static const __m512 variable
Raghuveer Devulapalli
committed
ENH: Add AVX-512 based 16-bit dtype sort
Raghuveer Devulapalli
committed
MAINT: Fix comment
Raghuveer Devulapalli
committed
MAINT: Use loadu intrinsic instead of set1_epi16
Raghuveer Devulapalli
committed
TST: Add quicksort test coverage for all 16, 32, 64 bit dtypes
Raghuveer Devulapalli
committed
MAINT: Fix linter errors
Raghuveer Devulapalli
committed
ENH: Use AVX-512 qsort for half precision float
Raghuveer Devulapalli
committed
BENCH: Add float16 to sort benchmarks

r-devulap
authored and
Raghuveer Devulapalli
committed
TST: Add test for float16 quicksort

r-devulap
authored and
Raghuveer Devulapalli
committed
Fix linter errors'

r-devulap
authored and
Raghuveer Devulapalli
committed
BUG: Use log2 instead a builtin
Raghuveer Devulapalli
committed
Adding x86-simd-sort as submodule
Raghuveer Devulapalli
committed

Commits on Jan 31, 2023

Add x86 simd sort dispatch files to meson.build
Raghuveer Devulapalli
committed
Fetch submodules in macOS and Windows build
Raghuveer Devulapalli
committed
Update to latest commit x86-simd-sort
Raghuveer Devulapalli
committed

Commits on Feb 1, 2023

Fix azure-pipelines.yml to checkout submodules
Raghuveer Devulapalli
committed