Commits

Branch selector

 main

User selector

r-devulap

Datepicker

All time

Commit History

Commits on Sep 26, 2022

• BUG: Add memoverlap check

  r-devulap

  committed
  7dfcd39

  Copy full SHA for 7dfcd39

• MAINT: Move AVX512 fp16 universal intrinsic to dispatch file

  r-devulap

  committed
  a13006a

  Copy full SHA for a13006a

• MAINT: Fix linter error

  r-devulap

  committed
  80f0015

  Copy full SHA for 80f0015

• ENH: Vectorize FP16 math functions on Intel SkylakeX

  r-devulap

  committed
  8dd6761

  Copy full SHA for 8dd6761

• SIMD: Add universal intrinsic support for SIMD float16 using AVX-512

  r-devulap

  committed
  0e87c99

  Copy full SHA for 0e87c99

• BENCH: Add benchmarks for fp16 umath functions

  r-devulap

  committed
  6070491

  Copy full SHA for 6070491

Commits on Jul 27, 2022

• TST: Add special FP tests for FP16 umath

  r-devulap

  committed
  2fa2e2e

  Copy full SHA for 2fa2e2e

• TST: Add accuracy tests for FP16 umath functions

  r-devulap

  committed
  df168ac

  Copy full SHA for df168ac

Commits on Feb 28, 2022

• BUILD: Disable MMX registers when building with -mavx512f

  Show description for b3eed14
  r-devulap

  committed
  b3eed14

  Copy full SHA for b3eed14

Commits on Feb 23, 2022

• BUG: Call scalar loop when SIMD is path is rejected

  r-devulap

  committed
  650f1b1

  Copy full SHA for 650f1b1

• ENH: Use SVML for f64 exp and log for AVX512_SKX

  r-devulap

  committed
  e16f5b0

  Copy full SHA for e16f5b0

• BUG: Fix spurious underflow FP exception for exp(-inf)

  r-devulap

  committed
  8ff71c7

  Copy full SHA for 8ff71c7

• Revert "ENH: Use SVML for f64 exp and log"

  Show description for 9a8e1e8
  r-devulap

  committed
  9a8e1e8

  Copy full SHA for 9a8e1e8

Commits on Feb 4, 2022

• BUG: Use an zmm register explicitly for neg inf

  r-devulap

  committed
  919a078

  Copy full SHA for 919a078

Commits on Feb 3, 2022

• BUG: Fix spurious underflow FP exception for expf(-inf)

  r-devulap

  committed
  87fb01c

  Copy full SHA for 87fb01c

• TST: Update exp test to check for spurious over and underflow flag

  r-devulap

  committed
  e3cb0a7

  Copy full SHA for e3cb0a7

• TST: Increase ULP tol to 2 for f64 exp and log

  r-devulap

  committed
  0d1d76e

  Copy full SHA for 0d1d76e

• ENH: Use SVML for f64 exp and log

  r-devulap

  committed
  413ae90

  Copy full SHA for 413ae90

• MAINT: rename reduce_min/max functions to not clash with local def in loops_minmax.dispatch.c.src

  r-devulap

  committed
  4db0cc7

  Copy full SHA for 4db0cc7

Commits on Feb 2, 2022

• MAINT: Use cvt instruction instead of extract

  r-devulap

  committed
  b814437

  Copy full SHA for b814437

• SIMD: Add intrinsics emulation for reduce_min and reduce_max instructions

  r-devulap

  committed
  0f1397f

  Copy full SHA for 0f1397f

Commits on Feb 1, 2022

• MAINT: Remove unnecessary function optimize attribute

  r-devulap

  committed
  f312f00

  Copy full SHA for f312f00

Commits on Jan 20, 2022

• MAINT: Use new CPU dispatch mechanism

  r-devulap

  committed
  60e7c19

  Copy full SHA for 60e7c19

Commits on Jan 14, 2022

• MAINT: Use _mm512_cmp_ps_mask instead of _mm512_cmpunord_ps_mask

  Show description for cf60656
  r-devulap

  committed
  cf60656

  Copy full SHA for cf60656

• TST: Use default_rng to generate random numbers for the test

  r-devulap

  committed
  b96d2e4

  Copy full SHA for b96d2e4

• MAINT: Workaround for gcc missing intrinsic: use loadu_si512 instead of loadu_epi32

  r-devulap

  committed
  8c5e7e6

  Copy full SHA for 8c5e7e6

• MAINT: Fix linter errors

  r-devulap

  committed
  917cdda

  Copy full SHA for 917cdda

• MAINT: Adding comments

  r-devulap

  committed
  16a6d45

  Copy full SHA for 16a6d45

• ENH: Vectorize sort for int32 and uint32 using AVX-512

  r-devulap

  committed
  f467d6b

  Copy full SHA for f467d6b

• TST: Add unit tests for vectorized quicksort

  r-devulap

  committed
  12d02f5

  Copy full SHA for 12d02f5

• BENCH: Add benchmarks for 32-bit data type sorting

  r-devulap

  committed
  61dec07

  Copy full SHA for 61dec07

• ENH: Use AVX-512 for float quicksort

  r-devulap

  committed
  0eb1a92

  Copy full SHA for 0eb1a92

Commits on Nov 12, 2021

• MAINT: Update SVML sources to prevent an executable stack

  r-devulap

  committed
  a688ed6

  Copy full SHA for a688ed6

Commits on Aug 24, 2021

• CI/CD: get submodule before build

  r-devulap

  committed
  1cbf913

  Copy full SHA for 1cbf913

• CI/CD: Pull submodules for circleci

  r-devulap

  committed
  35c1dfe

  Copy full SHA for 35c1dfe

Pagination

Previous

Next