ONNX Runtime Windows GPU Doc Gen CI

Workflow runs for ONNX Runtime Windows GPU Doc Gen CI

• [CPU] Speed up sub-byte DequantizeLinear (int4/uint4/int2/uint2)

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12103:Pull request#33784synchronizebyTheNetherWatcher
  TheNetherWatcher:subbyte-dq-cpu

  Oct 31, 2026Action required
  More actions for workflow run 12103

• [WebGPU] Support float16 GRU and LSTM (#33678)

  currently running:
  ONNX Runtime Windows GPU Doc Gen CI#12102:Commit121e34dpushedbyadrastogi
  main

  Oct 31, 2026In progress
  More actions for workflow run 12102

• Auto-generated baselines by 1ES Pipeline Templates

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12101:Pull request#33368synchronizebymicrosoft-github-policy-service[bot]Bot
  users/merlinbot/1es-pt-auto-baselining-pr

  Oct 31, 20262h 10m 34s
  More actions for workflow run 12101

• CPU: fuse transposed INT4 Gemm while preserving activation and output QDQ

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12100:Pull request#33458synchronizebyes0m
  es0m:user/ersomme/cpu-qdq-gemm-preserve-upstream

  Oct 31, 2026Action required
  More actions for workflow run 12100

• [CPU] Speed up sub-byte DequantizeLinear (int4/uint4/int2/uint2)

  currently running:
  ONNX Runtime Windows GPU Doc Gen CI#12099:Pull request#33784openedbyTheNetherWatcher
  TheNetherWatcher:subbyte-dq-cpu

  Oct 31, 2026In progress
  More actions for workflow run 12099

• [CUDA] MatMulNBits 8-bit: wide fp16 M=1 kernel for decode GEMVs

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12098:Pull request#33728synchronizebyy-chan
  y-chan:perf/matmulnbits-8bit-m1-wide

  Oct 31, 2026Action required
  More actions for workflow run 12098

• [CUDA] Add portable QMoE CPU/GPU offload backend with doorbell decode, adaptive promotion, and DMA prefill

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12097:Pull request#33452synchronizebytianleiwu
  tlwu/20261019/qmoe-offload-p3-cuda-operator

  Oct 31, 20262h 11m 1s
  More actions for workflow run 12097

• [CUDA] Pass cached SM version to cuDNN frontend Conv graphs (#33770)

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12096:Commit4edd78apushedbytianleiwu
  main

  Oct 31, 20262h 11m 19s
  More actions for workflow run 12096

• [CUDA] Add portable QMoE CPU/GPU offload backend with doorbell decode, adaptive promotion, and DMA prefill

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12095:Pull request#33452synchronizebytianleiwu
  tlwu/20261019/qmoe-offload-p3-cuda-operator

  Oct 31, 20262h 11m 43s
  More actions for workflow run 12095

• [CUDA] Add portable QMoE CPU/GPU offload backend with doorbell decode, adaptive promotion, and DMA prefill

  cancelled:
  ONNX Runtime Windows GPU Doc Gen CI#12094:Pull request#33452synchronizebytianleiwu
  tlwu/20261019/qmoe-offload-p3-cuda-operator

  Oct 31, 20263m 23s
  More actions for workflow run 12094

• [CUDA] Extend NVFP4 QMoE decode GEMV envelope and add 256-expert fused routing

  failed:
  ONNX Runtime Windows GPU Doc Gen CI#12093:Pull request#33545synchronizebytianleiwu
  tlwu/20261022/qmoe-nvfp4-256e-8row-decode

  Oct 31, 20261h 43m 19s
  More actions for workflow run 12093

• [WebGPU] Fuse rank-32 quantized LoRA with MatMulNBits

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12092:Pull request#33370synchronizebyqjia7
  qjia7:feat/webgpu-matmul-nbits-qlora

  Oct 31, 20262h 10m 22s
  More actions for workflow run 12092

• [CUDA] Add H256 BF16-query INT8-cache group-6/16 paged speculative XQ…

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12091:Commit5a83ed9pushedbytianleiwu
  main

  Oct 31, 20262h 11m 14s
  More actions for workflow run 12091

• Update Kusto dashboard SDK dependencies to 6.0.0

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12090:Pull request#33783openedbymustjab
  users/mustjab/kusto-dashboard-sdk-6

  Oct 31, 20262h 10m 59s
  More actions for workflow run 12090

• [CUDA] Add portable QMoE CPU/GPU offload backend with doorbell decode, adaptive promotion, and DMA prefill

  cancelled:
  ONNX Runtime Windows GPU Doc Gen CI#12089:Pull request#33452synchronizebytianleiwu
  tlwu/20261019/qmoe-offload-p3-cuda-operator

  Oct 31, 20261h 44m 30s
  More actions for workflow run 12089

• [CUDA] Optimize INT4 QMoE offload prefill/decode and S=8 doorbell execution

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12088:Pull request#33622synchronizebytianleiwu
  tlwu/20261026/qmoe-offload-s8-int4-opt

  Oct 31, 20262h 10m 55s
  More actions for workflow run 12088

• [CUDA] Add INT4 and INT3 QMoE CPU/GPU offload support

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12087:Pull request#33575synchronizebytianleiwu
  tlwu/20261024/qmoe-offload-int4-int3

  Oct 31, 20262h 9m 30s
  More actions for workflow run 12087

• [CUDA] Add 8-route grouped INT3 offload prefill, zero-cold doorbell short-circuit, and S=8 streaming optimizations

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12086:Pull request#33629synchronizebytianleiwu
  tlwu/20261026/qmoe-offload-avx2-int3-s8-opt

  Oct 31, 20262h 13m 9s
  More actions for workflow run 12086

• [CUDA] Add session-scoped MoE/QMoE offload planner, lifecycle control…

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12085:Commit649f516pushedbytianleiwu
  main

  Oct 31, 20262h 9m 55s
  More actions for workflow run 12085

• Align Windows WebGPU profiling and visualize CPU/GPU timelines

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12084:Pull request#33781synchronizebygyagp
  gyagp:pr-improve-profile-a82348b4-e08a-4244-9eb6-5ffa9998ddec

  Oct 31, 2026Action required
  More actions for workflow run 12084

• fix(web): handle opaque origins for bundled file URLs

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12083:Pull request#33782openedbyleninworld
  leninworld:fix-web-file-origin-33735

  Oct 31, 2026Action required
  More actions for workflow run 12083

• Align Windows WebGPU profiling and visualize CPU/GPU timelines

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12082:Pull request#33781openedbygyagp
  gyagp:pr-improve-profile-a82348b4-e08a-4244-9eb6-5ffa9998ddec

  Oct 31, 2026Action required
  More actions for workflow run 12082

• Fix severe CPU STFT performance for batched short transforms

  requires action with the application:
  ONNX Runtime Windows GPU Doc Gen CI#12081:Pull request#32457synchronizebythe0cp
  the0cp:stft-cpu-decomposition

  Oct 31, 2026Action required
  More actions for workflow run 12081

• [CUDA] Add session-scoped MoE/QMoE offload planner, lifecycle controller, and memory budget ledger

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12080:Pull request#33451synchronizebytianleiwu
  tlwu/20261019/qmoe-offload-p2-cuda-memory-lifecycle

  Oct 31, 20262h 19m 0s
  More actions for workflow run 12080

• Fix spelling typos flagged by misspell linter

  completed successfully:
  ONNX Runtime Windows GPU Doc Gen CI#12079:Pull request#33780openedbyadrastogi
  adityar-microsoft-fix-misspell-typos

  Oct 31, 20262h 17m 10s
  More actions for workflow run 12079