Commits

Branch selector

 master

User selector

jianyuh

Datepicker

All time

Commit History

Commits on Feb 10, 2020

• Reuse the partition scheme for GEMM in Depthwise Convolution (pytorch#293)

  Show description for 07dbb5d
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  07dbb5d

Commits on Feb 1, 2020

• clang-format the code (pytorch#278)

  Show description for fdb82b2
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  fdb82b2

• Fix the root cause for _aligned_free issues on row_offset_ allocation (pytorch#277)

  Show description for e4122f7
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e4122f7

Commits on Jan 31, 2020

• Fix the _aligned_free() issue on Windows (pytorch#275)

  Show description for 03f4ec2
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  03f4ec2

Commits on Jan 28, 2020

• Remove extra paranthesis (pytorch#270)

  Show description for 9166d9f
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  9166d9f

Commits on Dec 11, 2019

• Revert D18138146: Add support for AVX512-256(YMM) in FBGEMM16

  Show description for 9b38c64
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  9b38c64

Commits on Dec 10, 2019

• Fix FBGEMM OSS CI (pytorch#211)

  Show description for b9e495c
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  b9e495c

Commits on Dec 9, 2019

• Add the the FP16 <-> FP32 conversion benchmark (pytorch#210)

  Show description for 4b50573
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4b50573

Commits on Dec 7, 2019

• Add the option to clip when we convert FP32 to FP16 (pytorch#208)

  Show description for 5760376
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  5760376

• Migrate the FP16/BF16 <-> FP32 conversion routines into OSS (pytorch#203)

  Show description for 40ac0e5
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  40ac0e5

Commits on Nov 18, 2019

• Re-apply "D18566687: [fbgemm] Fix the OSS build issue for FP16Benchmark when MKL/BLAS is not available" (pytorch#200)

  Show description for b2a3d89
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  b2a3d89

• Fix the Lint issue for the header files (pytorch#197)

  Show description for 62c3b48
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  62c3b48

• Fix the OSS build issue for FP16Benchmark when MKL/BLAS is not available (pytorch#196)

  Show description for 4624a94
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4624a94

• Update CMakeLists.txt to fix the OSS build issues (pytorch#195)

  Show description for 7d6599b
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7d6599b

Commits on Nov 16, 2019

• Re-apply "[fbgemm] Enable the n dimensional parallelization" (pytorch#189)

  Show description for 56fc7ed
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  56fc7ed

Commits on Nov 14, 2019

• Fix the benchmark performance number

  Show description for 15b4a70
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  15b4a70

• Back out "Enable the n dimensional parallelization"

  Show description for 7e35fa8
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7e35fa8

• Enable the n dimensional parallelization (pytorch#170)

  Show description for 4147a9c
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4147a9c

Commits on Nov 12, 2019

• Add fast path for transpose_simd when M=1 or N=1 (pytorch#180)

  Show description for e27d9b5
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e27d9b5

Commits on Nov 7, 2019

• Fix the GEMM benchmark number (pytorch#166)

  Show description for ec22ece
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  ec22ece

Commits on Oct 1, 2019

• Fix the FP16 OSS failure (pytorch#136)

  Show description for e82b986
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e82b986

Commits on Sep 11, 2019

• Add assert to ensure the divisor is not 0 (#25960)

  Show description for 685b185
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  685b185

Commits on Aug 12, 2019

• Update README.md with mentioning PyTorch (pytorch#116)

  Show description for aceefe3
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  aceefe3

Commits on Aug 9, 2019

• Integrate VNNI into FBGEMM master branch (pytorch#114)

  Show description for 7b15607
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7b15607

Commits on Aug 6, 2019

• Back out "[fbgemm] Integrate VNNI into FBGEMM master branch"

  Show description for cf34b9a
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  cf34b9a

• Integrate VNNI into FBGEMM master branch (pytorch#113)

  Show description for d8b3323
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  d8b3323

Commits on Jul 23, 2019

• Update OSS build instructions for submodule build (pytorch#110)

  Show description for f712cb2
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  f712cb2

Commits on Jul 19, 2019

• Fix fbgemm OSS failure

  Show description for 1419a6e
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  1419a6e

Commits on Jul 10, 2019

• Refactoring unpack weight function (pytorch#103)

  Show description for f080393
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  f080393

Commits on Jul 1, 2019

• Refactor the code and avoid the duplication (pytorch#102)

  Show description for b0cf97d
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  b0cf97d

• Clean up some code for JIT code generator (pytorch#101)

  Show description for 61928df
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  61928df

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

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

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

• Fixes for FBGEMM FP16 performance (pytorch#82)

  Show description for 844dacc
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  844dacc

Pagination

PreviousNext