Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Jun 7, 2019

• Remove duplicated header and undo some changes in D15399811

  Show description for bf2f45f
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  bf2f45f

  Copy full SHA for bf2f45f

Commits on Jun 5, 2019

• Unified convolution interface

  Show description for 8197494
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  8197494

  Copy full SHA for 8197494

Commits on Jun 4, 2019

• Add quantized::fbgemm_linear_unpack operator for serialization (pytorch#97)

  Show description for 7786841
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7786841

  Copy full SHA for 7786841

Commits on May 30, 2019

• Adding -02 flag to the cmake build

  Show description for 85f4105
  

  Protonu Basu

  authored and

  facebook-github-bot

  committed
  85f4105

  Copy full SHA for 85f4105

Commits on May 24, 2019

• fix broken test

  Show description for 846785d
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  846785d

  Copy full SHA for 846785d

Commits on May 23, 2019

• Fix kernel logging

  Show description for d059448
  

  evhunter

  authored and

  facebook-github-bot

  committed
  d059448

  Copy full SHA for d059448

Commits on May 16, 2019

• update readme

  Show description for 9ae8912
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  9ae8912

  Copy full SHA for 9ae8912

• fixing compiler warnings for uninitialized MR, NCB, KCB

  Show description for d8f740d
  

  Protonu Basu

  authored and

  facebook-github-bot

  committed
  d8f740d

  Copy full SHA for d8f740d

Commits on May 14, 2019

• Fix CI indent error

  Show description for 1b14303
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  1b14303

  Copy full SHA for 1b14303

• fix circleci build with submodules

  Show description for 49be9f8
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  49be9f8

  Copy full SHA for 49be9f8

• Use submodules instead of cmake downloads

  Show description for 7de846a
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  7de846a

  Copy full SHA for 7de846a

Commits on May 13, 2019

• Back out "[FBGEMM][PR] switch from cmake downloads to git submodules"

  Show description for b14f582
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  b14f582

  Copy full SHA for b14f582

• switch from cmake downloads to git submodules (pytorch#95)

  Show description for c21a93d
  

  dhpollack

  authored and

  facebook-github-bot

  committed
  c21a93d

  Copy full SHA for c21a93d

Commits on Apr 19, 2019

• make sure cpuinfo_initialize called before fbgemmHasAvx2/512Support (pytorch#94)

  Show description for 6ec218e
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  6ec218e

  Copy full SHA for 6ec218e

Commits on Apr 3, 2019

• optimize dw conv for symmetric quant (pytorch#73)

  Show description for c6e8606
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  c6e8606

  Copy full SHA for c6e8606

Commits on Apr 2, 2019

• Exposing tuning parameters in FBGEMM (MCB, NCB, KCB, MR, NR, Row Interleave) (pytorch#90)

  Show description for f12ec12
  

  Protonu Basu

  authored and

  facebook-github-bot

  committed
  f12ec12

  Copy full SHA for f12ec12

Commits on Apr 1, 2019

• reduce the number of shapes tested in GConvTest (pytorch#91)

  Show description for d8e0d44
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  d8e0d44

  Copy full SHA for d8e0d44

Commits on Mar 25, 2019

• Packing B documentation

  Show description for 23aa5a3
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  23aa5a3

  Copy full SHA for 23aa5a3

Commits on Mar 21, 2019

• Improves small N cases back to what they were

  Show description for f65f0eb
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  f65f0eb

  Copy full SHA for f65f0eb

• Allocate some registers for B matrix loading and reuse loaded results

  Show description for 452627c
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  452627c

  Copy full SHA for 452627c

• Further optimize acc16 kernel and cache blocking dimension for B matrix is now free to be autotuned (pytorch#88)

  Show description for d53c022
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  d53c022

  Copy full SHA for d53c022

• Further optimize acc32 kernel and cache blocking dimension for B matrix is now free to be autotuned (pytorch#89)

  Show description for fe1c3d9
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  fe1c3d9

  Copy full SHA for fe1c3d9

Commits on Mar 19, 2019

• Dump generated kernels in files

  Show description for 0d8f88c
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  0d8f88c

  Copy full SHA for 0d8f88c

Commits on Mar 18, 2019

• Add the Naive bfloat16 implementation based on MKL

  Show description for 1351790
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  1351790

  Copy full SHA for 1351790

Commits on Mar 13, 2019

• optimize requantize for float out processing (pytorch#85)

  Show description for 6011ce3
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  6011ce3

  Copy full SHA for 6011ce3

Commits on Mar 8, 2019

• No need for PackA when m==1 (pytorch#83)

  Show description for 50b4316
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  50b4316

  Copy full SHA for 50b4316

• Fixes for FBGEMM FP16 performance (pytorch#82)

  Show description for 844dacc
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  844dacc

  Copy full SHA for 844dacc

Commits on Mar 6, 2019

• Add Avx512BW/VL/DQ check (pytorch#84)

  Show description for 66b4135
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  66b4135

  Copy full SHA for 66b4135

Commits on Mar 1, 2019

• Add documentations for the cache/register blocking parameters (pytorch#81)

  Show description for 2eb84b8
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  2eb84b8

  Copy full SHA for 2eb84b8

Commits on Feb 25, 2019

• barebone int8-acc16 and int8-acc32 benchmarks

  Show description for 426d7be
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  426d7be

  Copy full SHA for 426d7be

Commits on Feb 23, 2019

• specialization for first conv (pytorch#80)

  Show description for 7b99ce4
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  7b99ce4

  Copy full SHA for 7b99ce4

Commits on Feb 22, 2019

• Optimize PackB routine by removing addr function

  Show description for 5720ee7
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  5720ee7

  Copy full SHA for 5720ee7

Commits on Feb 20, 2019

• optimize PackAWithIm2Col for symmetric b quant

  Show description for 9ebc998
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  9ebc998

  Copy full SHA for 9ebc998

Commits on Feb 19, 2019

• increase test coverage (pytorch#78)

  Show description for a5f1f53
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a5f1f53

  Copy full SHA for a5f1f53

• remove unused member var kBlock_ (pytorch#77)

  Show description for 610378f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  610378f

  Copy full SHA for 610378f

Pagination

Previous

Next