Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 19, 2020

• implement a 8x32 transpose kernel for 8-bit integer (pytorch#459)

  Show description for bb111f3
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  bb111f3

  Copy full SHA for bb111f3

Commits on Nov 13, 2020

• fixed an issue with the TransposeTest (pytorch#458)

  Show description for 9b01311
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  9b01311

  Copy full SHA for 9b01311

Commits on Nov 12, 2020

• PackedMatrixB: templetize data conversion (pytorch#448)

  Show description for 56f7e4b
  

  efiks

  authored and

  facebook-github-bot

  committed
  56f7e4b

  Copy full SHA for 56f7e4b

Commits on Nov 10, 2020

• Remove dead includes in deeplearning/fbgemm/bench (pytorch#456)

  Show description for 92c5f37
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  92c5f37

  Copy full SHA for 92c5f37

Commits on Nov 9, 2020

• Remove dead includes in deeplearning/fbgemm/test

  Show description for 398deb4
  

  generatedunixname89002005287564

  authored and

  facebook-github-bot

  committed
  398deb4

  Copy full SHA for 398deb4

• Generalize GEMM Benchmark (pytorch#447)

  Show description for eb55572
  

  efiks

  authored and

  facebook-github-bot

  committed
  eb55572

  Copy full SHA for eb55572

Commits on Nov 7, 2020

• Clamping before conversion (pytorch#454)

  Show description for c438e6e
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  c438e6e

  Copy full SHA for c438e6e

Commits on Nov 3, 2020

• Make matrix dimentions to transpose_simd unsigned (pytorch#449)

  Show description for 8eb6dcb
  

  malfet

  authored and

  facebook-github-bot

  committed
  8eb6dcb

  Copy full SHA for 8eb6dcb

Commits on Nov 2, 2020

• Avoid potential integer overflow in transpose utils

  Show description for 5b7566f
  

  malfet

  authored and

  facebook-github-bot

  committed
  5b7566f

  Copy full SHA for 5b7566f

Commits on Oct 29, 2020

• Generalize functional FP test (pytorch#440)

  Show description for 51245fa
  

  efiks

  authored and

  facebook-github-bot

  committed
  51245fa

  Copy full SHA for 51245fa

• Change autogenerated headers for FBGEMM (pytorch#439)

  Show description for 39d5add
  

  efiks

  authored and

  facebook-github-bot

  committed
  39d5add

  Copy full SHA for 39d5add

Commits on Oct 20, 2020

• add im2col for transposed conv path (pytorch#445)

  Show description for 23cb1db
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  23cb1db

  Copy full SHA for 23cb1db

• add ref tranposed convolution for 1,2,3D cases (pytorch#446)

  Show description for 4096024
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  4096024

  Copy full SHA for 4096024

Commits on Oct 16, 2020

• Add missing copyright header (pytorch#443)

  Show description for e1db75c
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  e1db75c

  Copy full SHA for e1db75c

• Fixes compilation on gcc 10 (pytorch#441)

  Show description for f20d61e
  

  zasdfgbnm

  authored and

  facebook-github-bot

  committed
  f20d61e

  Copy full SHA for f20d61e

Commits on Oct 15, 2020

• Move FP16 function to common location (pytorch#438)

  Show description for abc56f6
  

  efiks

  authored and

  facebook-github-bot

  committed
  abc56f6

  Copy full SHA for abc56f6

• Move PackedMatrixB into separate file (pytorch#436)

  Show description for 48bc1b2
  

  efiks

  authored and

  facebook-github-bot

  committed
  48bc1b2

  Copy full SHA for 48bc1b2

Commits on Oct 14, 2020

• Fix variable initialization (pytorch#433)

  Show description for be57354
  

  efiks

  authored and

  facebook-github-bot

  committed
  be57354

  Copy full SHA for be57354

Commits on Oct 13, 2020

• remove NewType in SparseAdaGradSignature (pytorch#437)

  Show description for a570f94
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a570f94

  Copy full SHA for a570f94

Commits on Oct 11, 2020

• cleanup temp interface (pytorch#435)

  Show description for da05c8d
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  da05c8d

  Copy full SHA for da05c8d

Commits on Oct 9, 2020

• match old sparse adagrad interface with the new one (pytorch#434)

  Show description for 75ea7ce
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  75ea7ce

  Copy full SHA for 75ea7ce

• add new sparse adagrad interface takes weight_decay at call time not at codegen time (pytorch#432)

  Show description for 974d2b4
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  974d2b4

  Copy full SHA for 974d2b4

Commits on Oct 2, 2020

• Matrix transpose kernels for int8/uint8 datatypes (pytorch#431)

  Show description for fe91640
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  fe91640

  Copy full SHA for fe91640

Commits on Sep 14, 2020

• part3 Move embedding dequantization kernels to fbgemm for better sharing between C2/PT (pytorch#426)

  Show description for 1d71039
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  1d71039

  Copy full SHA for 1d71039

Commits on Sep 12, 2020

• Improve SLS performance on BDW by reducing port 5 pressure (pytorch#429)

  Show description for 0725301
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  0725301

  Copy full SHA for 0725301

Commits on Sep 4, 2020

• Fix gcc-9 compilation (pytorch#428)

  Show description for 74a0846
  

  malfet

  authored and

  facebook-github-bot

  committed
  74a0846

  Copy full SHA for 74a0846

• part2: Move embedding quantization kernels to fbgemm for better sharing between C2/PT (pytorch#425)

  Show description for 7e401c0
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  7e401c0

  Copy full SHA for 7e401c0

• add avx512vnni support for GroupConv (pytorch#423)

  Show description for 1289a1f
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  1289a1f

  Copy full SHA for 1289a1f

• add avx512 support for groupConv (pytorch#422)

  Show description for d5ace7c
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  d5ace7c

  Copy full SHA for d5ace7c

Commits on Sep 2, 2020

• To fix a windows failure (pytorch#424)

  Show description for a591925
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  a591925

  Copy full SHA for a591925

Commits on Aug 31, 2020

• Move embedding quantization kernels to fbgemm for better sharing between C2/PT (pytorch#419)

  Show description for a9397bf
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  a9397bf

  Copy full SHA for a9397bf

• Back out "add avx512 support for groupConv"

  Show description for 0048c0d
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  0048c0d

  Copy full SHA for 0048c0d

• Back out "add avx512vnni support for GroupConv"

  Show description for 36e0509
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  36e0509

  Copy full SHA for 36e0509

Commits on Aug 30, 2020

• add avx512vnni support for GroupConv (pytorch#411)

  Show description for a288e17
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  a288e17

  Copy full SHA for a288e17

• add avx512 support for groupConv (pytorch#420)

  Show description for c2d2c2c
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  c2d2c2c

  Copy full SHA for c2d2c2c

Pagination

Previous

Next