AMD Tests

Workflow runs for AMD Tests

• perf(amd): optimize muse spark fused gemv + iris lamport all gather for 2-8 decode rows

  completed successfully:
  AMD Tests#3329:Pull request#2637synchronizebyefric
  efric/muse-stack/18-post-expert-tail-rows

  Oct 23, 202623s
  More actions for workflow run 3329

• perf(amd): implement muse spark fused gemv + iris lamport all gather + post-expert norm

  completed successfully:
  AMD Tests#3328:Pull request#2633synchronizebyefric
  efric/muse-stack/14-post-expert-lamport-gather

  Oct 23, 202617s
  More actions for workflow run 3328

• perf(amd): optimize muse spark fused gemv + iris lamport all gather for 2-8 decode rows

  cancelled:
  AMD Tests#3327:Pull request#2637synchronizebyefric
  efric/muse-stack/18-post-expert-tail-rows

  Oct 23, 20267s
  More actions for workflow run 3327

• perf(amd): run muse spark decode router in the main stream

  completed successfully:
  AMD Tests#3326:Pull request#2635synchronizebyefric
  efric/muse-stack/16-router-main-stream

  Oct 23, 202626s
  More actions for workflow run 3326

• perf(amd): implement muse spark fused gated rmsnorm + o_proj gemv

  completed successfully:
  AMD Tests#3325:Pull request#2634synchronizebyefric
  efric/muse-stack/15-gated-attention-gemv

  Oct 23, 202630s
  More actions for workflow run 3325

• perf(amd): implement muse spark fused gated rmsnorm + o_proj gemv

  cancelled:
  AMD Tests#3324:Pull request#2634synchronizebyefric
  efric/muse-stack/15-gated-attention-gemv

  Oct 23, 20269s
  More actions for workflow run 3324

• perf(amd): optimize muse spark pre-expert gemv for 2-4 decode rows

  cancelled:
  AMD Tests#3323:Pull request#2636synchronizebyefric
  efric/muse-stack/17-pre-expert-rowcta-rows

  Oct 23, 20268s
  More actions for workflow run 3323

• perf(amd): optimize muse spark pre-expert gemv for 2-4 decode rows

  completed successfully:
  AMD Tests#3322:Pull request#2636synchronizebyefric
  efric/muse-stack/17-pre-expert-rowcta-rows

  Oct 23, 202627s
  More actions for workflow run 3322

• perf(amd): run muse spark decode router in the main stream

  cancelled:
  AMD Tests#3321:Pull request#2635synchronizebyefric
  efric/muse-stack/16-router-main-stream

  Oct 23, 202610s
  More actions for workflow run 3321

• perf(amd): optimize single-row bf16 gemv tiling

  completed successfully:
  AMD Tests#3320:Pull request#2632synchronizebyefric
  efric/muse-stack/13-rowcta-bf16-gemv

  Oct 23, 202616s
  More actions for workflow run 3320

• feat(amd): serve Qwen3.8-Flash-Next W4A8 MX checkpoints on gfx950

  completed successfully:
  AMD Tests#3319:Pull request#2643openedbyraikonenfnu
  raikonen/qwen38-w4a8-mx-runtime

  Oct 22, 202627s
  More actions for workflow run 3319

• chore(models): remove Qwen2 dense LM support

  failed:
  AMD Tests#3318:Pull request#2640ready_for_reviewbyborontion
  borontion/chore/remove-qwen2

  Oct 22, 20263h 31m 42s
  More actions for workflow run 3318

• perf(amd): optimize gfx1250 Gluon MHA prefill

  completed successfully:
  AMD Tests#3317:Pull request#2641openedbynithinsubbiah
  perf/gfx1250-fav3-host

  Oct 22, 202617s
  More actions for workflow run 3317

• chore(models): remove Qwen2 dense LM support

  completed successfully:
  AMD Tests#3316:Pull request#2640openedbyborontion
  borontion/chore/remove-qwen2

  Oct 22, 202621s
  More actions for workflow run 3316

• perf(inkling): commit every conv stream's verified window in one launch

  completed successfully:
  AMD Tests#3315:Pull request#2638ready_for_reviewbynperrin-fr
  perf/inkling-commit-one-launch

  Oct 22, 20261h 7m 0s
  More actions for workflow run 3315

• perf(inkling): commit every conv stream's verified window in one launch

  completed successfully:
  AMD Tests#3314:Pull request#2638synchronizebynperrin-fr
  perf/inkling-commit-one-launch

  Oct 22, 202616s
  More actions for workflow run 3314

• feat(models): add mixed-precision quantization and DSpark draft support

  failed:
  AMD Tests#3313:Pull request#2639synchronizebylightseek-bot
  bot/mxfp8-nvfp4-dspark

  Oct 22, 20261h 19m 56s
  More actions for workflow run 3313

• feat(models): add mixed-precision quantization and DSpark draft support

  cancelled:
  AMD Tests#3312:Pull request#2639labeledbylightseek-bot
  bot/mxfp8-nvfp4-dspark

  Oct 22, 20263m 12s
  More actions for workflow run 3312

• feat(models): add mixed-precision quantization and DSpark draft support

  cancelled:
  AMD Tests#3311:Pull request#2639openedbylightseek-bot
  bot/mxfp8-nvfp4-dspark

  Oct 22, 20262m 0s
  More actions for workflow run 3311

• feat(kda): Add optional BF16 KDA state and FlashInfer decode/verify support

  failed:
  AMD Tests#3310:Pull request#1666synchronizebylightseek-bot
  hongtaoc/kda-bf16-flashinfer

  Oct 22, 20261h 11m 29s
  More actions for workflow run 3310

• perf(amd): optimize muse spark fused gemv + iris lamport all gather for 2-8 decode rows

  completed successfully:
  AMD Tests#3309:Pull request#2637synchronizebyefric
  efric/muse-stack/18-post-expert-tail-rows

  Oct 22, 202626s
  More actions for workflow run 3309

• perf(amd): implement muse spark fused gemv + iris lamport all gather + post-expert norm

  completed successfully:
  AMD Tests#3308:Pull request#2633synchronizebyefric
  efric/muse-stack/14-post-expert-lamport-gather

  Oct 22, 202619s
  More actions for workflow run 3308

• perf(amd): run muse spark decode router in the main stream

  completed successfully:
  AMD Tests#3307:Pull request#2635synchronizebyefric
  efric/muse-stack/16-router-main-stream

  Oct 22, 202626s
  More actions for workflow run 3307

• perf(amd): optimize muse spark pre-expert gemv for 2-4 decode rows

  completed successfully:
  AMD Tests#3306:Pull request#2636synchronizebyefric
  efric/muse-stack/17-pre-expert-rowcta-rows

  Oct 22, 202626s
  More actions for workflow run 3306

• perf(amd): run muse spark decode router in the main stream

  cancelled:
  AMD Tests#3305:Pull request#2635synchronizebyefric
  efric/muse-stack/16-router-main-stream

  Oct 22, 20269s
  More actions for workflow run 3305