Code

Merge OpenAI Triton commit 4f6f768#2634

Code

Merged

whitneywhtsang merged 14 commits intomainintel/intel-xpu-backend-for-triton:mainfrom whitneywhtsang/mergeintel/intel-xpu-backend-for-triton:whitneywhtsang/mergeCopy head branch name to clipboardNov 6, 2024

+1,116-722

Commits

Commits on Oct 29, 2024

• [Proton] Adding Sorting of Kernels (#4987)

  CRobeck

  authored
  30e7a2e

  Copy full SHA for 30e7a2e

• [FRONTEND] Fix transpose with tuple dims (#5006)

  Show description for ef61488
  peterbell10

  authored
  ef61488

  Copy full SHA for ef61488

• [AMD] remove redundant LDS bypass checks (#5002)

  Show description for 69f656c
  binarman

  authored
  69f656c

  Copy full SHA for 69f656c

• [AMD] Skip scaled_dot tests for gfx11 and gfx12 (#5008)

  Show description for bf6bd0b
  AlexAUT

  authored
  bf6bd0b

  Copy full SHA for bf6bd0b

• Add string representation for AttrsDescriptor (#4888)

  Show description for ebce7f3
  alexbaden

  authored
  ebce7f3

  Copy full SHA for ebce7f3

Commits on Oct 30, 2024

• [BACKEND][NVIDIA] Add Lowering for Shared-to-MMAv3-DotOp Copy (#5009)

  Show description for cfddb09
  ggengnv

  authored
  cfddb09

  Copy full SHA for cfddb09

• Don't specify -A x64 option and reuse cmake build type config for Windows (#5014)

  Show description for 0591b37
  anmyachev

  authored
  0591b37

  Copy full SHA for 0591b37

• [Frontend][Backend] Implement support for scale_dot(-, bf16) (#4996)

  Show description for 23c9ec1
  lezcano

  authored
  23c9ec1

  Copy full SHA for 23c9ec1

• [INTERPRETER] Make sure interpreter works with float16 by reusing NumPy HALF-related code (#5010)

  Show description for 61eb94e
  anmyachev

  authored
  61eb94e

  Copy full SHA for 61eb94e

• Allow windows cuda files to be used in setup.py (#5015)

  Show description for 018c139
  anmyachev

  authored
  018c139

  Copy full SHA for 018c139

• Fix formatting in docs for triton.language.dot (#5020)

  saagarjha

  authored
  ef319c8

  Copy full SHA for ef319c8

Commits on Oct 31, 2024

• Ignore autotune runs failed with PTXAS error (#5017)

  htyu

  authored
  6693ddd

  Copy full SHA for 6693ddd

• [AMD] Reland sinking the 2nd tt.load after local_load's (#4935)

  Show description for 4f6f768
  zhanglx13

  authored
  4f6f768

  Copy full SHA for 4f6f768

Commits on Nov 5, 2024

• Merge commit '4f6f76874ff623562903d5452d499cae3d40d448'

  whitneywhtsang

  committed
  49a52a2

  Copy full SHA for 49a52a2