Commits

Branch selector

 a4d3ca7

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 16, 2021

• use 2D parallelism in CPU emb fwd (pytorch#549)

  Show description for a4d3ca7
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a4d3ca7

• Update asmjit for fbcode and oss (pytorch#538)

  Show description for 5d36b7d
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  5d36b7d

• have bf16 quantize kernels use 2d blocks with loop to cover large tensors (pytorch#541)

  Show description for 2b5c02d
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  2b5c02d

• Register cpu path table_batched parameters as buffers to enable running on Mast (pytorch#547)

  Show description for bd470d9
  

  Chris Cai

  authored and

  facebook-github-bot

  committed
  bd470d9

Commits on Mar 15, 2021

• use at::acc_type (pytorch#546)

  Show description for 4e7dbd0
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  4e7dbd0

• parallelize across T rather than B to avoid race condition in exact sgd (pytorch#544)

  Show description for 316ef4f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  316ef4f

Commits on Mar 13, 2021

• Optimizing cache memory (pytorch#543)

  Show description for a7fd8fb
  

  Junjie Yang

  authored and

  facebook-github-bot

  committed
  a7fd8fb

Commits on Mar 11, 2021

• Enable enforce_hbm option to allow placing both emb and cache in HBM (pytorch#539)

  Show description for 81738bb
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  81738bb

Commits on Mar 10, 2021

• print total execution time of mkl (pytorch#537)

  Show description for d12fc48
  

  YangWang92

  authored and

  facebook-github-bot

  committed
  d12fc48

Commits on Mar 9, 2021

• Multi-threading support for table batched embedding ops (cpu mode) (pytorch#535)

  Show description for 06f5f62
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  06f5f62

• Add the cmake file for searching sphinx (pytorch#536)

  Show description for 4b88f40
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4b88f40

Commits on Mar 8, 2021

• Clang Format (pytorch#533)

  Show description for 25d1620
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  25d1620

• Performance improvements for embedding index weight gradient calculation (cpu mode)

  Show description for da1e687
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  da1e687

Commits on Mar 7, 2021

• Enabling codegen for approximate sparse optimizers for embedding bag

  Show description for becaf15
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  becaf15

Commits on Mar 5, 2021

• hfp8 conversion kernels (pytorch#531)

  Show description for 53c7846
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  53c7846

• Remember cuda device for managed allocations and use it to release the managed memory. (pytorch#530)

  Show description for 6ee23c2
  

  Stephan Uphoff

  authored and

  facebook-github-bot

  committed
  6ee23c2

• Improving performance of table batched embedding backward for cpu mode

  Show description for 186377a
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  186377a

Commits on Mar 4, 2021

• support 8bit rowwise quantize op in OSS tests (pytorch#522)

  Show description for 7dd5547
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  7dd5547

Commits on Mar 3, 2021

• Add Copyright Headers to comply the OSS requirements (pytorch#529)

  Show description for e0a2e71
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e0a2e71

• fix split_table_batched_embeddings_test on cpu (pytorch#524)

  Show description for 2b53662
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  2b53662

• Fix trunk breakage by extern fbgemm emb op (pytorch#528)

  Show description for 5ed1e8c
  

  xw285cornell

  authored and

  facebook-github-bot

  committed
  5ed1e8c

Commits on Mar 2, 2021

• improve perf for lru_cache_insert

  Show description for 5b6eaf5
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  5b6eaf5

• Only keep one copy of benchmark and unit test (pytorch#527)

  Show description for c3612e6
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  c3612e6

Commits on Mar 1, 2021

• Fix build with older gcc versions (pytorch#525)

  Show description for a431ee3
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  a431ee3

Commits on Feb 27, 2021

• open source sparse-dense matrix-matrix multiplication (int8 and fp32) (pytorch#523)

  Show description for 953e671
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  953e671

Commits on Feb 26, 2021

• fix dense split table batch kernel for shared embedding tables

  Show description for 77d96bb
  

  xing-liu

  authored and

  facebook-github-bot

  committed
  77d96bb

• Improved performance for embedding bag operator for CPU execution

  Show description for 0826262
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  0826262

Commits on Feb 24, 2021

• Update the env scripts for CentOS8 (pytorch#521)

  Show description for d123130
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  d123130

Commits on Feb 22, 2021

• Fix the mutation input issues for JIT op registration scheme (pytorch#518)

  Show description for 1718b3a
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  1718b3a

Commits on Feb 18, 2021

• Update SparseAdagradBenchmark with adjustable weight decay (pytorch#517)

  Show description for 975e41d
  

  venkatacrc

  authored and

  facebook-github-bot

  committed
  975e41d

Commits on Feb 17, 2021

• Add unit test for UVM APIs in FBGEMM_GPU (pytorch#516)

  Show description for 887359c
  

  Joe Hu

  authored and

  facebook-github-bot

  committed
  887359c

• Cooperlake performance debug: fix partition issues (pytorch#514)

  Show description for c520088
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  c520088

Commits on Feb 16, 2021

• Replaced APPROX_ADAGRAD and APPROX_ROWWISE_ADAGRAD with EXACT_ADAGRAD and EXACT_ROWWISE_ADAGRAD

  Show description for aa1b4e6
  

  Andrew Wang

  authored and

  facebook-github-bot

  committed
  aa1b4e6

• Fix 10x model

  Show description for b51f3e6
  

  Dmytro Ivchenko

  authored and

  facebook-github-bot

  committed
  b51f3e6

• Fix/Optimize the thread block allocation in sparse_ops (pytorch#494)

  Show description for 24e92f0
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  24e92f0

Pagination

PreviousNext