All pull requests

Pull requests

• Bump the github-actions group across 1 directory with 9 updates

  dependenciesgithub_actions
  #1302·dependabot[bot] opened Jun 29, 2026Bot·

  ·

• Add dynamic MXFP4 quantization kernel and tests

  #1299·suryajasper opened Apr 28, 2026Contributor·

  ·

• Bump IREE requirement pins to their latest versions.

  #1298·iree-pr-automator[bot] opened Apr 28, 2026Bot·

  ·

• Add gfx90a WaveASM target support

  #1297·feimaomomo opened Apr 27, 2026·

  ·
  1

• Add stream-K MXFP4 GEMM kernel and tests

  #1296·willghatch opened Apr 23, 2026Contributor·

  ·
  1

• Add MultipleBuffer, MBSK, LocalSplitU, and TreeStreamK split-K MXFP4 …

  #1295·willghatch opened Apr 23, 2026Contributor·

  ·
  1

• GEMM support in waveasm LLVM path

  #1288·Hardcode84 opened Apr 9, 2026Contributor·

  ·
  36

• [asm]Add wide_stores support for MXFP4 (4wave) GEMM

  #1286·xintin opened Apr 9, 2026Contributor·

  ·

• Hoist loop-variant affine index ops out of scf.for during codegen

  #1283·willghatch opened Apr 8, 2026Contributor·

  ·
  3

• Add CLAUDE.md files for Claude Code guidance

  #1273·tgymnich opened Apr 7, 2026Contributor·

  ·
  23

• Cherry pick salu promotion

  #1265·panditsa opened Apr 6, 2026Contributor·3 tasks done·

  ·

• Add reads_merged flag to track scale read merge status in E2E tests

  #1227·Hardcode84 opened Apr 1, 2026Contributor·

  ·
  2

• Emit magic-number division for dynamic kernels

  #1222·adedespirlet opened Apr 1, 2026Contributor·

  ·
  10

• Add lizard cyclomatic complexity check to pre-commit

  #1221·Hardcode84 opened Apr 1, 2026Contributor·

  ·
  7

• [asm] Add cross-class register spilling (VGPR -> AGPR) to linear scan allocator

  #1205·Hardcode84 opened Mar 28, 2026Contributor·

  ·

• [wip][asm] coalesce post loop store

  #1199·xintin opened Mar 27, 2026Contributor·

  ·

• [wip] Coalesce epilogue store

  #1194·xintin opened Mar 25, 2026Contributor·

  ·

• Harden hyperparameter and vector_shapes verification to only admit i64 entries

  #1180·ftynse with CopilotAI opened Mar 24, 2026·

  ·
  1

  16

  

• [wip][wavasm] Enable 64x64x256 dynamic MXFP4 GEMM with ee=True on 3-stage pipeline on waveasm

  #1163·xintin opened Mar 21, 2026Contributor·

  ·

• Add dynamic MXFP4 double-buffer schedule for preshuffle-B kernels (MT0 == 64)

  #1146·adityakankariya opened Mar 18, 2026Contributor·

  ·

• Ixsimpl integration

  #1130·Hardcode84 opened Mar 15, 2026Contributor·

  ·
  7

• Add schedule for 256x224x256 macro tile

  #1129·willghatch opened Mar 13, 2026Contributor·

  ·
  4

• 8Wave PP race-free kernel w/o double barrier

  #1123·adedespirlet opened Mar 12, 2026Contributor·

  ·

• [asm] Add --mlir-timing support to waveasm-translate

  #1122·Hardcode84 opened Mar 12, 2026Contributor·

  ·

• Add MLIR verifier to check HardwareConstraint.vector_shapes consistency with tile sizes

  #1094·tgymnich opened Mar 10, 2026Contributor·

  ·
  1

  44