Skip to content

feat(qy): enable QY CUDA runtime device backend (2026.11.12) - #54

Open
Ceng23333 wants to merge 2 commits into
InfiniTensor:masterfrom
Ceng23333:feat/qy-device-20261112
Open

Ceng23333 wants to merge 2 commits into
InfiniTensor:masterfrom
Ceng23333:feat/qy-device-20261112

Conversation

@Ceng23333

@Ceng23333 Ceng23333 commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor

Summary

  • Enable Denglin QY CUDA runtime device backend (Device::Type::kQy, runtime/graph APIs).
  • Define kEventNotReady for QY so test_qy_runtime compiles (parity with other CUDA backends).

Motivation / related

Type of change

  • feat
  • fix (kEventNotReady)

Platforms affected

  • QY

Validation

  • Host Denglin QY, DENGLIN_DEVICES=3.
  • Configure/build with WITH_QY=ON + testing; ctest → 13/13 PASS (incl. test_qy_runtime / test_qy_graph).
  • See PR smoke comments. Log: nov_milestone/logs/prhead-verify-20261113/rt-ctest.log

Checklist

  • format/build/ruff green; head includes kEventNotReady fix
  • Conventional Commits title/branch

Stack PR index (merge order)

  1. InfiniOps: feat(qy): QY ops backend with FA and longctx MoE/RoPE kernels (2026.11.12) InfiniOps#1038 — 71e85162ab96
  2. InfiniRT: feat(qy): enable QY CUDA runtime device backend (2026.11.12) #54 — 047340f45efa
  3. InfiniCCL: feat(qy): enable InfiniCCL NCCL backend for Denglin QY (2026.11.12) InfiniCCL#90 — 691c3092c4c9
  4. InfiniCore: chore(qy): pin InfiniRT/Ops/CCL for QY MoE bf16-warpnk (2026.11.12) InfiniCore#1583 — 0b69ee9002bd
  5. InfiniLM: feat(qy): MiniCPM5 FA/longctx/cudagraph + MoE bf16/warpnk surpass (2026.11.12) InfiniLM#618 — 36beae5376f4

ITW pin: https://github.com/Ceng23333/InfiniTensorWorktree/tree/v2026.11.12-qy-moe-bf16-warpnk

Add Denglin/QY native CUDA runtime stubs, dlcc wrapper, and CMake/test
wiring so InfiniRT builds and dispatches on DEVICE_TYPE_QY.
Match other CUDA backends so test_qy_runtime compiles against Denglin.

Co-authored-by: Cursor <cursoragent@cursor.com>
@Ceng23333

Copy link
Copy Markdown
Contributor Author

QY required smoke (PR head)

  • Head SHA: 047340f45efaa305b61515900c83ec37ad210437
  • Host: Denglin QY, DENGLIN_DEVICES=3
  • Command: ctest --test-dir …/build/infinirt --output-on-failure (WITH_QY=ON, testing ON)
  • Result: PASS (13/13, incl. test_qy_runtime / test_qy_graph)
  • Log: nov_milestone/logs/prhead-verify-20261113/rt-ctest.log
100% tests passed, 0 tests failed out of 13

@Ceng23333

Copy link
Copy Markdown
Contributor Author

QY required smoke (PR head) — detailed

  • Head SHA: 047340f45efaa305b61515900c83ec37ad210437
  • Host: Denglin QY, DENGLIN_DEVICES=3
  • Command: ctest --test-dir …/build/infinirt --output-on-failure
  • Overall: PASS — 13/13
  • Log: nov_milestone/logs/prhead-verify-20261113/rt-ctest.log
# Test Result
1 test_smoke PASS
2 test_core PASS
3 test_device_guard PASS
4 test_small_vector PASS
5 test_metadata_view PASS
6 test_shape_strides_storage PASS
7 test_tensor_view_allocations PASS
8 test_cpu_runtime PASS
9 test_qy_runtime PASS
10 test_qy_graph PASS
11 test_runtime_dispatch PASS
12 test_install PASS
13 test_install_consumer PASS

QY-critical rows: test_qy_runtime, test_qy_graph. Tip includes kEventNotReady fix.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant