Commits

Breadcrumbs

History for

onmain

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 10, 2022

• Back out "Add support for Embedding_dim is not multiple of 4 on CPU" (pytorch#1388)

  Show description for 4d1738b
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4d1738b

Commits on Sep 29, 2022

• Add support for Embedding_dim is not multiple of 4 on CPU (pytorch#1357)

  Show description for 9cd90e9
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  9cd90e9

Commits on Sep 19, 2022

• optimize 2x2 Avx2 kernel (pytorch#1280)

  Show description for 728cdb8
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  728cdb8

• reland the 1x2 unrolling optimization (pytorch#1275)

  Show description for 62820ec
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  62820ec

Commits on Sep 16, 2022

• Add fbgemmHasArmNeon support (pytorch#1328)

  Show description for 54594d0
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  54594d0

Commits on Sep 15, 2022

• adapting to fbgemm tests (pytorch#1324)

  Show description for 927db42
  

  Pallab Bhattacharya

  authored and

  facebook-github-bot

  committed
  927db42

Commits on Sep 6, 2022

• fbgemm building aarch64 using simde (pytorch#1271)

  Show description for 0f2ff7a
  

  psaab

  authored and

  facebook-github-bot

  committed
  0f2ff7a

Commits on Aug 30, 2022

• Add Mx2, Mx4, 2xN, and 4xN avx512 transpose (pytorch#1177)

  Show description for 4449436
  

  CaoE

  authored and

  facebook-github-bot

  committed
  4449436

• fix index overflow for FindMinMax and QuantizeAvx2 (pytorch#1261)

  Show description for fdc8939
  

  XiaobingSuper

  authored and

  facebook-github-bot

  committed
  fdc8939

Commits on Aug 24, 2022

• Add x86 instrinsics guard (pytorch#1257)

  Show description for de510f4
  

  psaab

  authored and

  facebook-github-bot

  committed
  de510f4

• Back out "proper unrolling for 1x2 avx256 kernel" (pytorch#1256)

  Show description for c4b0f7e
  

  Yasharth Bhartiya

  authored and

  facebook-github-bot

  committed
  c4b0f7e

• Fix building under Xcode 14 (pytorch#1254)

  Show description for 8d833b0
  

  mattjgalloway

  authored and

  facebook-github-bot

  committed
  8d833b0

Commits on Aug 23, 2022

• proper unrolling for 1x2 avx256 kernel (pytorch#1253)

  Show description for 6d950a2
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  6d950a2

Commits on Aug 5, 2022

• add sequence embedding reference implementation (pytorch#1232)

  Show description for 5a03b7c
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  5a03b7c

Commits on Aug 2, 2022

• Update Utils.h (pytorch#1233)

  Show description for 499cd22
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  499cd22

Commits on Jul 28, 2022

• Revert D37670913 "[fbgemm] Use std::fma in tail of Fused8BitRowwiseQuantizedSBFloatToFloatOrHalfAvx2" (pytorch#1222)

  Show description for 4812758
  

  swolchok

  authored and

  facebook-github-bot

  committed
  4812758

Commits on Jul 8, 2022

• Use std::fma in tail of Fused8BitRowwiseQuantizedSBFloatToFloatOrHalfAvx2 (pytorch#1191)

  Show description for 56cd603
  

  swolchok

  authored and

  facebook-github-bot

  committed
  56cd603

• Use the new asmjit to fix //deeplearning/fbgemm (pytorch#1077)

  Show description for 64a5c4a
  

  r-barnes

  authored and

  facebook-github-bot

  committed
  64a5c4a

Commits on Jun 21, 2022

• Don't store on each iteration in EmbeddingSpMDMBlockSize1 (pytorch#1164)

  Show description for a1046b4
  

  swolchok

  authored and

  facebook-github-bot

  committed
  a1046b4

Commits on May 20, 2022

• skip weights when the row is pruned (pytorch#1129)

  Show description for 0d6a4e8
  

  Ruichao Xiao

  authored and

  facebook-github-bot

  committed
  0d6a4e8

• QuantizeAvx2: Manually hoist zero_point loads (pytorch#1124)

  Show description for 41d598c
  

  swolchok

  authored and

  facebook-github-bot

  committed
  41d598c

Commits on May 15, 2022

• ROCm changes (pytorch#1102)

  Show description for ac5db35
  

  liligwu

  authored and

  facebook-github-bot

  committed
  ac5db35

Commits on May 6, 2022

• Add FP8 embedding weight support for inference TBE (pytorch#1091)

  Show description for ec21412
  

  Sarunya Pumma

  authored and

  facebook-github-bot

  committed
  ec21412

Commits on May 5, 2022

• Add missing FMA in Fused8BitRowwiseQuantizedSBFloatToFloatOrHalfAvx2 (pytorch#1098)

  Show description for 06c6369
  

  swolchok

  authored and

  facebook-github-bot

  committed
  06c6369

Commits on Apr 19, 2022

• Add i16 support for transpose (pytorch#1054)

  Show description for 86c6571
  

  CaoE

  authored and

  facebook-github-bot

  committed
  86c6571

Commits on Apr 15, 2022

• add 7x7 depthwise (pytorch#1049)

  Show description for 40101e8
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  40101e8

Commits on Apr 14, 2022

• fix local_thread (#75290)

  Show description for 2ab113d
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  2ab113d

Commits on Mar 8, 2022

• remove output needs to requantized assertion

  Show description for 814ec8c
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  814ec8c

Commits on Mar 1, 2022

• fix handling nullptr rowOffsetBuf (pytorch#956)

  Show description for 389f89d
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  389f89d

• handling empty matrix in transpose (pytorch#955)

  Show description for 85ff0ff
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  85ff0ff

Commits on Feb 23, 2022

• disable direct conv path

  Show description for c2d46af
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  c2d46af

Commits on Feb 15, 2022

• use int64_t in fbgemmPartition1D (pytorch#933)

  Show description for f7050c9
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  f7050c9

Commits on Feb 8, 2022

• add direct conv to fbgemm (pytorch#901)

  Show description for f0f6ca7
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  f0f6ca7

Commits on Feb 5, 2022

• use thread local code cache in cpu quantized emb fwd (pytorch#903)

  Show description for e779169
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  e779169

Commits on Jan 21, 2022

• Transposed Direct Convolution for kh=2 kw=6

  Show description for 81719d6
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  81719d6

Pagination

PreviousNext