Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 5, 2023

• X32-PACKW add prefetch to AVX microkernels

  Show description for 1f4c825
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  1f4c825

  Copy full SHA for 1f4c825

• Add an 'oddalbert' model with K and N values testing remainder performance

  Show description for bca7588
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  bca7588

  Copy full SHA for bca7588

• Make amalgamate script work with open source builds

  Show description for 81212bb
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  81212bb

  Copy full SHA for 81212bb

• F16 Dynamic Fully Connected NC Operator

  Show description for b8e8397
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  b8e8397

  Copy full SHA for b8e8397

• X32-PACKW SSE add prefetch versions of microkernels

  Show description for 4463262
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  4463262

  Copy full SHA for 4463262

• allocator.h: add extern C around declarations for Visual C compatability

  Show description for e6a3904
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  e6a3904

  Copy full SHA for e6a3904

Commits on Apr 4, 2023

• Add a runtime flag to enable JIT

  Show description for 60e4f48
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  60e4f48

  Copy full SHA for 60e4f48

• X32-PACKW add X16 scalar microkernels for performance comparisons

  Show description for f9d626d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  f9d626d

  Copy full SHA for f9d626d

• X32-PACKW C4 SSE and WASMSIMD handle NC remainder by duplicating last row

  Show description for 3c18695
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  3c18695

  Copy full SHA for 3c18695

• Fix missing packw microkernels in GEMM E2E benchmarks

  Show description for 4f12a76
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  4f12a76

  Copy full SHA for 4f12a76

• Enable X32-PACKW AVX and AVX512 in production

  Show description for 45b1c51
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  45b1c51

  Copy full SHA for 45b1c51

• Remove use of designated initializers in test

  Show description for 054b4e4
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  054b4e4

  Copy full SHA for 054b4e4

• Scalar packb (pack bias) and zerob (zero bias) microkernels

  Show description for 8dafac3
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  8dafac3

  Copy full SHA for 8dafac3

• Detect if __builtin_wasm_laneselect_i32x4 works correctly with only the sign bit set.

  Show description for 19cf31a
  

  alankelly

  authored and

  xnnpack-bot

  committed
  19cf31a

  Copy full SHA for 19cf31a

• Test memory allocation failures during FP16 rewrite

  Show description for c676ab2
  

  fredrec

  authored and

  xnnpack-bot

  committed
  c676ab2

  Copy full SHA for c676ab2

Commits on Apr 3, 2023

• Use 3f3m3l to replace 9p DWCONV unipass for Wasm SIMD

  Show description for 6e1f60c
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  6e1f60c

  Copy full SHA for 6e1f60c

• Remove XNN_NO_*_OPERATORS macros

  Show description for f3a0fa8
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  f3a0fa8

  Copy full SHA for f3a0fa8

• Make FP16/FP32 models take a bool for enabling JIT

  Show description for 7b23502
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  7b23502

  Copy full SHA for 7b23502

• X32-PACKW SSE2 add K unrolled x8

  Show description for dbb8c33
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  dbb8c33

  Copy full SHA for dbb8c33

• X32-PACKW rename AVX2 to AVX and check AVX (haswell minimum) flags

  Show description for 042f4fc
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  042f4fc

  Copy full SHA for 042f4fc

• Fix check for FP16 native support

  Show description for 03e8ec1
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  03e8ec1

  Copy full SHA for 03e8ec1

• Add 3f3m3l DWCONV multipass microkernels for Wasm and Wasm SIMD

  Show description for 2e92932
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  2e92932

  Copy full SHA for 2e92932

• Move variable declaration to prevent unused variable error.

  Show description for 1f75332
  

  alankelly

  authored and

  xnnpack-bot

  committed
  1f75332

  Copy full SHA for 1f75332

• X32-PACKW S4 AVX2 microkernels

  Show description for 2822d64
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  2822d64

  Copy full SHA for 2822d64

• X32-PACKW S4 SSE2 NR remainder remove vzero for NR-1 row

  Show description for 1537e1d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  1537e1d

  Copy full SHA for 1537e1d

• Rerun amalgamation scripts for packw change to null bias

  Show description for ff6d5d6
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  ff6d5d6

  Copy full SHA for ff6d5d6

Commits on Apr 2, 2023

• F32 Dynamic Fully Connected NC operator

  Show description for 7aa4aba
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  7aa4aba

  Copy full SHA for 7aa4aba

• Zero-initialize packed weights then bias is null in PACKW microkernels

  Show description for fe74cb8
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  fe74cb8

  Copy full SHA for fe74cb8

Commits on Apr 1, 2023

• X32-PACKW AVX Microkernels

  Show description for 415697d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  415697d

  Copy full SHA for 415697d

• X32-PACKW AVX512 Microkernel

  Show description for 2448bfd
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  2448bfd

  Copy full SHA for 2448bfd

Commits on Mar 31, 2023

• X32-PACKW AVX Microkernels

  Show description for 95abf02
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  95abf02

  Copy full SHA for 95abf02

• Run test template scripts

  Show description for b03566e
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  b03566e

  Copy full SHA for b03566e

• Bug fix for sorting tests where last line requires sort but has no newline

  Show description for 3363653
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  3363653

  Copy full SHA for 3363653

• Remove JIT support in quantized MobileNet models

  Show description for 25f09e5
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  25f09e5

  Copy full SHA for 25f09e5

• Fix typo in DWCONV multipass name in config

  Show description for efd524d
  

  ngzhian

  authored and

  xnnpack-bot

  committed
  efd524d

  Copy full SHA for efd524d

Pagination

Previous

Next