Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 16, 2024

• Merge pull request #77 from ROCm/IFU-20041216

  Show description for fbf3cd0
  liligwu

  authored
  fbf3cd0

• Merge branch 'IFU-20041216' into HEAD

  liligwu

  committed
  aeeb3cb

Commits on Dec 15, 2024

• use RocksDB iterator to read key range from ssd embedding (pytorch#3495)

  Show description for b8796b3
  

  Yulu Jia

  authored and

  facebook-github-bot

  committed
  b8796b3

• Back out "support quantize_fp8_row for up to 4d non contiguous tensor" (pytorch#3505)

  Show description for 532861b
  

  jwfromm

  authored and

  facebook-github-bot

  committed
  532861b

Commits on Dec 14, 2024

• custom allgather support multiple dtypes (pytorch#3498)

  Show description for c932a35
  

  xw285cornell

  authored and

  facebook-github-bot

  committed
  c932a35

Commits on Dec 13, 2024

• Migrate multi_head_jagged_flash_attention SLL ops to OSS (pytorch#3485)

  Show description for 6e9b083
  

  q10

  authored and

  facebook-github-bot

  committed
  6e9b083

• Add type error suppressions for upcoming upgrade (pytorch#3504)

  Show description for d70a19e
  

  generatedunixname89002005307016

  authored and

  facebook-github-bot

  committed
  d70a19e

• Fixed FBGEMM fp8 rowwise for irregular shapes (pytorch#3491)

  Show description for e4650b4
  

  zjing14

  authored and

  facebook-github-bot

  committed
  e4650b4

• Install pytorch-triton for the ROCm case (pytorch#3501)

  Show description for 5fecfea
  

  q10

  authored and

  facebook-github-bot

  committed
  5fecfea

• Add support for Python 3.13 in OSS (pytorch#3502)

  Show description for 8d80e34
  

  q10

  authored and

  facebook-github-bot

  committed
  8d80e34

• Modularize CMake Build [7/N] (pytorch#3500)

  Show description for ee39c89
  

  q10

  authored and

  facebook-github-bot

  committed
  ee39c89

• Unit test for new move tbe from device/cache_load method (pytorch#3437)

  Show description for 5d361fc
  

  aporialiao

  authored and

  facebook-github-bot

  committed
  5d361fc

• Remove setting total_cache_hash_size as buffer (pytorch#3441)

  Show description for 5326971
  

  aporialiao

  authored and

  facebook-github-bot

  committed
  5326971

• Create move TBE to right device, and set Cache Load in TBE class (pytorch#3438)

  Show description for 3ce55ad
  

  aporialiao

  authored and

  facebook-github-bot

  committed
  3ce55ad

• fix to add f-string format to log (pytorch#3433)

  Show description for bdbac01
  

  varun2784

  authored and

  facebook-github-bot

  committed
  bdbac01

Commits on Dec 12, 2024

• Modularize CMake Build [6/N] (pytorch#3492)

  Show description for 5bbb93c
  

  q10

  authored and

  facebook-github-bot

  committed
  5bbb93c

• write_k_back for bf16 ROPE (pytorch#3463)

  Show description for faba0e9
  

  bottler

  authored and

  facebook-github-bot

  committed
  faba0e9

• Attempt to fix FBGEMM CPU build (pytorch#3499)

  Show description for 5c16f4b
  

  huydhn

  authored and

  facebook-github-bot

  committed
  5c16f4b

• Update triton version to 3.2.0+35c6c7c6284582b3f41c71c150e11b517acf074a (pytorch#3497)

  Show description for 4dac248
  

  brad-mengchi

  authored and

  facebook-github-bot

  committed
  4dac248

• skip broken inference test that uses ssd TBE (pytorch#3494)

  Show description for 7aaec2b
  

  Yulu Jia

  authored and

  facebook-github-bot

  committed
  7aaec2b

Commits on Dec 11, 2024

• support quantize_fp8_row for up to 4d non contiguous tensor (pytorch#3493)

  Show description for 9a74419
  

  Jingyuan Fan

  authored and

  facebook-github-bot

  committed
  9a74419

• Log total duration spent prefetching (pytorch#3487)

  Show description for 18af808
  

  keyan

  authored and

  facebook-github-bot

  committed
  18af808

Commits on Dec 10, 2024

• Add generate_vbe_metadata test (pytorch#3483)

  Show description for 6a447e9
  

  sryap

  authored and

  facebook-github-bot

  committed
  6a447e9

• Add backwards compatibility checks for v1.1.0 release (pytorch#3489)

  Show description for 5dbeebf
  

  q10

  authored and

  facebook-github-bot

  committed
  5dbeebf

• Add a benchmark for VBE (pytorch#3464)

  Show description for bdb7095
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  bdb7095

Commits on Dec 9, 2024

• Fix CK FP8 rowwise quantization for some GEMM shapes (pytorch#3486)

  Show description for 9210796
  

  zixi-qi

  authored and

  facebook-github-bot

  committed
  9210796

• migrate "jagged_dense_flash_attention" (pytorch#3482)

  Show description for 6bffbcc
  

  brad-mengchi

  authored and

  facebook-github-bot

  committed
  6bffbcc

• Fix an integer overflow in permute_multi_embedding() (pytorch#3465)

  Show description for 0feee94
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  0feee94

• Migrate jagged_dense_elementwise_add SLL ops to OSS (pytorch#3479)

  Show description for bcb1f8f
  

  q10

  authored and

  facebook-github-bot

  committed
  bcb1f8f

Commits on Dec 7, 2024

• Fix for T209749755 ("Your diff, D66861418, broke one test") (pytorch#3480)

  Show description for 3fb2862
  

  generatedunixname537391475639613

  authored and

  facebook-github-bot

  committed
  3fb2862

• Update .SO loader in OSS (pytorch#3477)

  Show description for 887e5bf
  

  q10

  authored and

  facebook-github-bot

  committed
  887e5bf

• Implement Vec2 load/store for ROCm devices (pytorch#3475)

  Show description for 3bf676c
  

  avbokovoy

  authored and

  facebook-github-bot

  committed
  3bf676c

• FP8 Rowwise compile fix followup for AMD (pytorch#3478)

  Show description for 72da187
  

  jwfromm

  authored and

  facebook-github-bot

  committed
  72da187

• Properly define preallocated output as mutable in fp8 rowwise gemm (pytorch#3476)

  Show description for 60bbdd7
  

  jwfromm

  authored and

  facebook-github-bot

  committed
  60bbdd7

Commits on Dec 6, 2024

• Release Triton jagged tensor ops to OSS (pytorch#3474)

  Show description for f1804aa
  

  q10

  authored and

  facebook-github-bot

  committed
  f1804aa

Pagination

PreviousNext