Skip to content

Actions: ahrefs/ocannl

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
2,500+ workflow runs
2,500+ workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
A Workgroup_reduce level over a body the peel refuses still takes the hardware binding, which races the shared cell Claude Code #1806: Issue comment #950 (comment) created by lukstafi
1s
1s
Workgroup_reduce race criterion: judge Tile_mma through its fallback; static-slot disjointness in the recurrence census Claude Code #1805: Issue #960 opened by lukstafi
11s
11s
Hardware binding does not prove lane-indexed cells disjoint across bound axes (acc[i + j] under Workgroup j, Workgroup_reduce i) Claude Code #1804: Issue #959 opened by lukstafi
1s
1s
sweep: an environment-red GPU unit should rerun its failing stanzas serially so a test-logic failure cannot hide under runtime noise Claude Code #1803: Issue comment #945 (comment) created by lukstafi
8s
8s
Windows CI setup broken: cygwin mirror rotated pkgconf 2.5.1 out, the pinned downgrade 404s Claude Code #1802: Issue comment #935 (comment) created by lukstafi
1s
1s
Serialize or retry concurrent HIPRTC module loads on WSL gfx1151 Claude Code #1801: Issue comment #927 (comment) created by lukstafi
1s
1s
Metal SharedEvent all-work signals do not serialize short untracked command buffers Claude Code #1800: Issue comment #909 (comment) created by lukstafi
1s
1s
Serialize or retry concurrent HIPRTC module loads on WSL gfx1151 Claude Code #1799: Issue comment #927 (comment) created by lukstafi
10s
10s
Windows: test_cell_group's killed-child partial-stdout case captures nothing (benchmarks/test_orchestrate.py) Claude Code #1798: Issue #958 opened by lukstafi
1s
1s
Windows: sweep_harness's driver invokes WSL's bash.exe, not Git Bash, and fails before the harness runs Claude Code #1797: Issue #957 opened by lukstafi
1s
1s
Windows CI setup broken: cygwin mirror rotated pkgconf 2.5.1 out, the pinned downgrade 404s Claude Code #1796: Issue comment #935 (comment) created by lukstafi
1s
1s
A dune invocation dune itself refuses runs no tests, but test-run.sh digests it as FAIL (exit 1) like a failing suite Claude Code #1795: Issue comment #944 (comment) created by lukstafi
10s
10s
Serialize or retry concurrent HIPRTC module loads on WSL gfx1151 Claude Code #1794: Issue comment #927 (comment) created by lukstafi
2s
2s
sweep: an environment-red GPU unit should rerun its failing stanzas serially so a test-logic failure cannot hide under runtime noise Claude Code #1793: Issue comment #945 (comment) created by lukstafi
1s
1s
sweep_harness intermittently fails on the Ubuntu CI leg: sweep rejects 'cc' as an unknown backend while listing it Claude Code #1792: Issue comment #949 (comment) created by lukstafi
1s
1s
A Workgroup_reduce level over a body the peel refuses still takes the hardware binding, which races the shared cell Claude Code #1791: Issue comment #950 (comment) created by lukstafi
1s
1s
cc's scalar half uniform1 consumes 16 bits of the key where CUDA and Metal consume 32, so one key draws different half values per backend (bf16 and the lane forms agree) Claude Code #1790: Issue comment #951 (comment) created by lukstafi
1s
1s
sweep_harness intermittently fails on the Ubuntu CI leg: sweep rejects 'cc' as an unknown backend while listing it Claude Code #1789: Issue comment #949 (comment) created by lukstafi
1s
1s
Metal SharedEvent all-work signals do not serialize short untracked command buffers Claude Code #1788: Issue comment #909 (comment) created by lukstafi
1s
1s
cc's scalar half uniform1 consumes 16 bits of the key where CUDA and Metal consume 32, so one key draws different half values per backend (bf16 and the lane forms agree) Claude Code #1787: Issue comment #951 (comment) created by lukstafi
1s
1s
A dune invocation dune itself refuses runs no tests, but test-run.sh digests it as FAIL (exit 1) like a failing suite Claude Code #1786: Issue comment #944 (comment) created by lukstafi
2s
2s
bin-smoke on the macOS runner: projection_shape_bench fails the job when every tuner timing is refused for host contention Claude Code #1785: Issue comment #905 (comment) created by lukstafi
1s
1s
tools/test-run.sh's per-worktree lock has no cooperative entry point, so format-sweep.sh reverse-engineers its internals (fd flock, owner files) and will drift with them Claude Code #1784: Issue comment #671 (comment) created by lukstafi
1s
1s
dune build @fmt prints 22 odoc doc-comment warnings across 9 files on every run, burying the fmt gate's one real diff in CI logs Claude Code #1783: Issue #956 opened by lukstafi
1s
1s
Goldens embed source line:col via ~here, so any line-shifting edit owes a promote round: the formatting sweep grew a convergence loop and a location-masking validator around pure noise Claude Code #1782: Issue comment #672 (comment) created by lukstafi
7s
7s