CPU Tests

Workflow runs for CPU Tests

• feat(oga-hipep): run Qwen3.5-9B text on the AMD GPU through hip-ep

  completed successfully:
  CPU Tests#96:Pull request#929synchronizebyandrew180917
  oga-hipep-upstream-20261103

  Nov 3, 20261m 9s
  More actions for workflow run 96

• feat(oga-hipep): run Qwen3.5-9B text on the AMD GPU through hip-ep

  completed successfully:
  CPU Tests#95:Pull request#929openedbyandrew180917
  oga-hipep-upstream-20261103

  Nov 3, 20261m 21s
  More actions for workflow run 95

• server: answer an unknown model with 404 instead of 200

  completed successfully:
  CPU Tests#94:Pull request#928openedbyzaneni6
  fix/unknown-model-404

  Nov 3, 20261m 24s
  More actions for workflow run 94

• feat(core): ryzenai-corelib 0.60.0; every core model AMD ships a Q4 of pulls only that

  completed successfully:
  CPU Tests#93:Pull request#927openedbychizamd
  chizamd:feat/corelib-0.60

  Nov 2, 20261m 23s
  More actions for workflow run 93

• perf: map the token tables of qwen3, gemma3:1b and embed-gemma from model.q4nx instead of copying them to RAM

  completed successfully:
  CPU Tests#92:Pull request#926openedbytawei-amd
  feat/mmap-embeddings-batch1

  Nov 2, 20261m 39s
  More actions for workflow run 92

• Merge pull request #911 from ROCm/fix/connection-slot-leak

  completed successfully:
  CPU Tests#91:Commitd4a4553pushedbyzaneni6
  main

  Nov 2, 20261m 17s
  More actions for workflow run 91

• Merge pull request #913 from ROCm/fix/qwen35-ctx-cap-overrun

  completed successfully:
  CPU Tests#90:Commit94ef35cpushedbyzaneni6
  main

  Nov 2, 20261m 23s
  More actions for workflow run 90

• Merge pull request #923 from ROCm/chore/peano-xclbins-followup

  completed successfully:
  CPU Tests#89:Commit9efc614pushedbyzaneni6
  main

  Nov 2, 20261m 16s
  More actions for workflow run 89

• chore(xclbins): switch 4 more kernels to their Peano builds

  completed successfully:
  CPU Tests#88:Pull request#923openedbyzaneni6
  chore/peano-xclbins-followup

  Nov 2, 20261m 24s
  More actions for workflow run 88

• Merge pull request #922 from ROCm/chore/peano-xclbins-from-iron

  completed successfully:
  CPU Tests#87:Commit164d606pushedbyzaneni6
  main

  Nov 2, 20261m 19s
  More actions for workflow run 87

• Merge pull request #921 from ROCm/feat/gemma4-mmap-embeddings

  completed successfully:
  CPU Tests#86:Commitbcc562fpushedbytawei-amd
  main

  Nov 2, 20261m 38s
  More actions for workflow run 86

• chore(xclbins): vendor the current Peano kernel builds

  completed successfully:
  CPU Tests#85:Pull request#922openedbyzaneni6
  chore/peano-xclbins-from-iron

  Nov 2, 20261m 19s
  More actions for workflow run 85

• chore: vendor the Gemma4 engine libs from IRON#140: map token / per-layer embeddings instead of copying (#807)

  completed successfully:
  CPU Tests#84:Pull request#921synchronizebytawei-amd
  feat/gemma4-mmap-embeddings

  Nov 2, 20261m 25s
  More actions for workflow run 84

• chore: vendor the Gemma4 engine libs from IRON#140: map token / per-layer embeddings instead of copying (#807)

  completed successfully:
  CPU Tests#83:Pull request#921synchronizebytawei-amd
  feat/gemma4-mmap-embeddings

  Nov 2, 20261m 20s
  More actions for workflow run 83

• chore: vendor the Gemma4 engine libs from IRON#140: map token / per-layer embeddings instead of copying (#807)

  completed successfully:
  CPU Tests#82:Pull request#921openedbytawei-amd
  feat/gemma4-mmap-embeddings

  Nov 2, 20261m 29s
  More actions for workflow run 82

• chore: add security scanning workflows

  completed successfully:
  CPU Tests#81:Pull request#754synchronizebyharibabug
  add-security-scanning

  Nov 2, 20261m 19s
  More actions for workflow run 81

• perf(gemma4e-flash, gemma4-it, nemotron3.5-bolt[-dflash]): vendor engine libs with single-context vision attention (IRON#138)

  completed successfully:
  CPU Tests#80:Pull request#920openedbytawei-amd
  perf/vision-single-ctx-libs-siblings

  Nov 2, 20261m 8s
  More actions for workflow run 80

• perf(gemma4e-flash, gemma4-it, nemotron3.5-bolt[-dflash[-flash]]): vendor Gemma4e engine libs with single-context vision attention (IRON#137)

  completed successfully:
  CPU Tests#79:Pull request#919openedbytawei-amd
  perf/vision-single-ctx-libs

  Nov 2, 20261m 16s
  More actions for workflow run 79

• fix(ctx limit): do not forward the EOS token past the engine's KV window

  completed successfully:
  CPU Tests#78:Pull request#915synchronizebytawei-amd
  fix/eos-forward-ctx-guard

  Nov 2, 20261m 11s
  More actions for workflow run 78

• Merge pull request #918 from ROCm/docs/qwen38-dflash2-followup

  completed successfully:
  CPU Tests#77:Commiteb8aef5pushedbytawei-amd
  main

  Nov 2, 20261m 6s
  More actions for workflow run 77

• docs: DFlash 2 in the catalog page, decode speed after long prompts

  completed successfully:
  CPU Tests#76:Pull request#918openedbytawei-amd
  docs/qwen38-dflash2-followup

  Nov 2, 20261m 8s
  More actions for workflow run 76

• Merge pull request #917 from ROCm/docs/third-party-amd-files

  completed successfully:
  CPU Tests#75:Commit86ae732pushedbytawei-amd
  main

  Nov 2, 20261m 14s
  More actions for workflow run 75

• docs: add AMD-owned vendored files to THIRD_PARTY_NOTICES

  completed successfully:
  CPU Tests#74:Pull request#917openedbytawei-amd
  docs/third-party-amd-files

  Nov 2, 20261m 7s
  More actions for workflow run 74

• Merge pull request #916 from ROCm/feat/qwen38-dflash2

  completed successfully:
  CPU Tests#73:Commit0d6c78fpushedbytawei-amd
  main

  Nov 2, 20261m 14s
  More actions for workflow run 73

• feat(qwen3.8): qwen3.8-dflash2:27b, Qwen3.8-27B with the DFlash 2 draft

  completed successfully:
  CPU Tests#72:Pull request#916synchronizebytawei-amd
  feat/qwen38-dflash2

  Nov 2, 20261m 4s
  More actions for workflow run 72