Code

Xccl all#7

Code

Open

Chao1Han wants to merge 109 commits intomainChao1Han/pytorch:mainfrom xccl-p2pChao1Han/pytorch:xccl-p2pCopy head branch name to clipboard

+4,978-21

Commits

Commits on Aug 29, 2024

• Happy Init

  Chao1Han

  committed
  c0e1dc1

  Copy full SHA for c0e1dc1

Commits on Aug 30, 2024

• update

  Chao1Han

  committed
  93a4bdb

  Copy full SHA for 93a4bdb

• update

  Chao1Han

  committed
  ba6c4b7

  Copy full SHA for ba6c4b7

• update

  Chao1Han

  committed
  68a6aee

  Copy full SHA for 68a6aee

Commits on Sep 2, 2024

• register

  Chao1Han

  committed
  31eeee9

  Copy full SHA for 31eeee9

• update

  Chao1Han

  committed
  b977abc

  Copy full SHA for b977abc

Commits on Sep 3, 2024

• update

  Chao1Han

  committed
  486b61a

  Copy full SHA for 486b61a

• fix typo and register frontend

  Chao1Han

  committed
  6844932

  Copy full SHA for 6844932

• update

  Chao1Han

  committed
  7f6f8b9

  Copy full SHA for 7f6f8b9

• update

  Chao1Han

  committed
  be68320

  Copy full SHA for be68320

• update

  Chao1Han

  committed
  2e21d4f

  Copy full SHA for 2e21d4f

Commits on Sep 4, 2024

• update

  Chao1Han

  committed
  c9ef78f

  Copy full SHA for c9ef78f

• update

  Chao1Han

  committed
  076db36

  Copy full SHA for 076db36

• update

  Chao1Han

  committed
  8d739ac

  Copy full SHA for 8d739ac

Commits on Sep 5, 2024

• register again

  Chao1Han

  committed
  fb9746b

  Copy full SHA for fb9746b

• update

  Chao1Han

  committed
  4f73180

  Copy full SHA for 4f73180

Commits on Sep 6, 2024

• update

  Chao1Han

  committed
  7c2f018

  Copy full SHA for 7c2f018

Commits on Sep 9, 2024

• refine cmake

  Chao1Han

  committed
  229a80a

  Copy full SHA for 229a80a

• register all dist op and enable getXcclReduceOp

  Chao1Han

  committed
  746007b

  Copy full SHA for 746007b

• update

  Chao1Han

  committed
  5195f52

  Copy full SHA for 5195f52

Commits on Sep 10, 2024

• update

  Chao1Han

  committed
  227e98d

  Copy full SHA for 227e98d

• update

  Chao1Han

  committed
  2eb0446

  Copy full SHA for 2eb0446

• update flag

  Chao1Han

  committed
  0f61762

  Copy full SHA for 0f61762

Commits on Sep 11, 2024

• update

  Chao1Han

  committed
  df81919

  Copy full SHA for df81919

• update

  Chao1Han

  committed
  b0c0592

  Copy full SHA for b0c0592

• rm redundance code

  Chao1Han

  committed
  366d208

  Copy full SHA for 366d208

• enable timeout

  Chao1Han

  committed
  3530e43

  Copy full SHA for 3530e43

Commits on Sep 12, 2024

• add oneccl env

  Chao1Han

  committed
  485ae8b

  Copy full SHA for 485ae8b

• update

  Chao1Han

  committed
  0cfd224

  Copy full SHA for 0cfd224

• Add simple test

  Chao1Han

  committed
  b99fd8c

  Copy full SHA for b99fd8c

• update

  Chao1Han

  committed
  dc41d6a

  Copy full SHA for dc41d6a

• enable coalese

  Chao1Han

  committed
  4c3f49f

  Copy full SHA for 4c3f49f

• Support broadcast

  Chao1Han

  committed
  afa2adc

  Copy full SHA for afa2adc

• update

  Chao1Han

  committed
  8efb5d0

  Copy full SHA for 8efb5d0

• update

  Chao1Han

  committed
  e85c268

  Copy full SHA for e85c268

• update

  Chao1Han

  committed
  7488dbd

  Copy full SHA for 7488dbd

• update

  Chao1Han

  committed
  e5d6f37

  Copy full SHA for e5d6f37

• add allgather

  Chao1Han

  committed
  0da5e77

  Copy full SHA for 0da5e77

Commits on Sep 13, 2024

• support allgather_into_tensor_coalesced

  Chao1Han

  committed
  0ad5677

  Copy full SHA for 0ad5677

• support reduce_scatter

  Chao1Han

  committed
  009e334

  Copy full SHA for 009e334

• refine test cases

  Chao1Han

  committed
  ecbd989

  Copy full SHA for ecbd989

• update ut

  Chao1Han

  committed
  a23ffb2

  Copy full SHA for a23ffb2

• add mpi check

  Chao1Han

  committed
  1d02dfe

  Copy full SHA for 1d02dfe

• update datatype map

  Chao1Han

  committed
  c485bd8

  Copy full SHA for c485bd8

• update

  Chao1Han

  committed
  2d1ae87

  Copy full SHA for 2d1ae87

• Merge branch 'xccl' into xccl-group

  Chao1Han

  committed
  04226de

  Copy full SHA for 04226de

• update

  Chao1Han

  committed
  6184261

  Copy full SHA for 6184261

• update

  Chao1Han

  committed
  91d26d9

  Copy full SHA for 91d26d9

Commits on Sep 14, 2024

• update

  Chao1Han

  committed
  2a83d68

  Copy full SHA for 2a83d68

• update

  Chao1Han

  committed
  7f62b86

  Copy full SHA for 7f62b86

• Support reduce_scatter_base

  Chao1Han

  committed
  c48f5eb

  Copy full SHA for c48f5eb

• Support reduce_scatter_tensor_coalesced

  Chao1Han

  committed
  9b17dc4

  Copy full SHA for 9b17dc4

• support barrier

  Chao1Han

  committed
  6cb3227

  Copy full SHA for 6cb3227

• Merge branch 'xccl' into xccl-group

  Chao1Han

  committed
  e59f051

  Copy full SHA for e59f051

• update

  Chao1Han

  committed
  d858c81

  Copy full SHA for d858c81

• update

  Chao1Han

  committed
  fea20f5

  Copy full SHA for fea20f5

• update

  Chao1Han

  committed
  e0e27f3

  Copy full SHA for e0e27f3

Commits on Sep 18, 2024

• add ut

  Chao1Han

  committed
  029026d

  Copy full SHA for 029026d

• Support all2all_base

  Chao1Han

  committed
  682f40f

  Copy full SHA for 682f40f

• update

  Chao1Han

  committed
  2694617

  Copy full SHA for 2694617

• support all2all

  Chao1Han

  committed
  612df42

  Copy full SHA for 612df42

Commits on Sep 19, 2024

• use lintrunner format code

  Chao1Han

  committed
  001dac2

  Copy full SHA for 001dac2

• rm allgatherv align with nccl

  Chao1Han

  committed
  f13b449

  Copy full SHA for f13b449

• Support reduce

  Chao1Han

  committed
  af29d96

  Copy full SHA for af29d96

• Support gather

  Chao1Han

  committed
  20b1188

  Copy full SHA for 20b1188

• Support scatter

  Chao1Han

  committed
  1463eca

  Copy full SHA for 1463eca

• update

  Chao1Han

  committed
  156c2ac

  Copy full SHA for 156c2ac

Commits on Sep 20, 2024

• Merge branch 'xccl' into xccl-group

  Chao1Han

  committed
  79fd17e

  Copy full SHA for 79fd17e

• Xccl process group for Pytorch

  Chao1Han

  committed
  652da01

  Copy full SHA for 652da01

• Merge remote-tracking branch 'upstream/main' into xccl-bak

  Chao1Han

  committed
  0cb0016

  Copy full SHA for 0cb0016

• Align latest

  Chao1Han

  committed
  a71d69a

  Copy full SHA for a71d69a

• Merge branch 'xccl-bak' into xccl-group

  Chao1Han

  committed
  a1c2d6b

  Copy full SHA for a1c2d6b

• Merge branch 'xccl-group' into xccl-p2p

  Chao1Han

  committed
  ea7a4c9

  Copy full SHA for ea7a4c9

• update

  Chao1Han

  committed
  4bf448d

  Copy full SHA for 4bf448d

• update

  Chao1Han

  committed
  1f83fbf

  Copy full SHA for 1f83fbf

• update

  Chao1Han

  committed
  4f4ecf4

  Copy full SHA for 4f4ecf4

• Merge branch 'xccl-group' into xccl-p2p

  Chao1Han

  committed
  53142c2

  Copy full SHA for 53142c2

• update

  Chao1Han

  committed
  b6bc4a8

  Copy full SHA for b6bc4a8

Commits on Sep 23, 2024

• support p2p

  Chao1Han

  committed
  1fbb7ed

  Copy full SHA for 1fbb7ed

Commits on Sep 29, 2024

• refine findccl code

  Chao1Han

  committed
  88bea25

  Copy full SHA for 88bea25

Commits on Sep 30, 2024

• Add comments for build xccl

  Chao1Han

  committed
  f6ea934

  Copy full SHA for f6ea934

Commits on Oct 9, 2024

• minor fix

  Chao1Han

  committed
  31d092d

  Copy full SHA for 31d092d

• rm duplicate code and refine cmake

  Chao1Han

  committed
  cbea299

  Copy full SHA for cbea299

Commits on Oct 10, 2024

• update cmake

  Chao1Han

  committed
  ef261c6

  Copy full SHA for ef261c6

• hidden xccl specific

  Chao1Han

  committed
  6c648cd

  Copy full SHA for 6c648cd

Commits on Oct 11, 2024

• fix ci fail

  Chao1Han

  committed
  e621fe6

  Copy full SHA for e621fe6

• Merge branch 'xccl-bak' into xccl-group

  Chao1Han

  committed
  4a45d29

  Copy full SHA for 4a45d29

• Merge branch 'xccl-group' into xccl-p2p

  Chao1Han

  committed
  7bee4d9

  Copy full SHA for 7bee4d9

Commits on Oct 12, 2024

• rm ccl attr

  Chao1Han

  committed
  f85a845

  Copy full SHA for f85a845

Commits on Oct 17, 2024

• Refine specific code

  Chao1Han

  committed
  56a5e7f

  Copy full SHA for 56a5e7f

• accept comments

  Chao1Han

  committed
  a062f9f

  Copy full SHA for a062f9f

Commits on Oct 21, 2024

• refine code

  Chao1Han

  committed
  86b66c3

  Copy full SHA for 86b66c3

• Merge branch 'xccl-bak' into xccl-p2p

  Chao1Han

  committed
  1e68c30

  Copy full SHA for 1e68c30

• align to latest

  Chao1Han

  committed
  d9ce636

  Copy full SHA for d9ce636

• refine code

  Chao1Han

  committed
  385c218

  Copy full SHA for 385c218

• Merge branch 'xccl-bak' into xccl-p2p

  Chao1Han

  committed
  ea4fe8c

  Copy full SHA for ea4fe8c

• update

  Chao1Han

  committed
  e36a99c

  Copy full SHA for e36a99c

• update

  Chao1Han

  committed
  5096354

  Copy full SHA for 5096354

Commits on Oct 22, 2024

• add RECORD_PARAM_COMMS_DATA

  Chao1Han

  committed
  9e6448b

  Copy full SHA for 9e6448b

Commits on Oct 25, 2024

• update

  Chao1Han

  committed
  8d9c24e

  Copy full SHA for 8d9c24e

• update

  Chao1Han

  committed
  e808b6c

  Copy full SHA for e808b6c

Commits on Oct 28, 2024

• update

  Chao1Han

  committed
  193d946

  Copy full SHA for 193d946

Commits on Oct 31, 2024

• fix all_gather_v bug

  Chao1Han

  committed
  eb447f2

  Copy full SHA for eb447f2

Commits on Nov 4, 2024

• Merge remote-tracking branch 'origin/main' into xccl-p2p

  Chao1Han

  committed
  55150c8

  Copy full SHA for 55150c8

Commits on Nov 11, 2024

• correct get kvs

  Chao1Han

  committed
  20b60b1

  Copy full SHA for 20b60b1

Commits on Nov 12, 2024

• update kvs key

  Chao1Han

  committed
  b442419

  Copy full SHA for b442419

Commits on Nov 14, 2024

• Merge remote-tracking branch 'origin/main' into xccl-p2p

  Chao1Han

  committed
  0aedc00

  Copy full SHA for 0aedc00

• WA AVG reduction

  Chao1Han

  committed
  65e0d9d

  Copy full SHA for 65e0d9d

Commits on Nov 15, 2024

• update test case

  Chao1Han

  committed
  3e97e67

  Copy full SHA for 3e97e67