Skip to content

Actions: NVIDIA/cuDecomp

Actions

Build

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
build.yml
80 workflow runs
80 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Remove inter-trial device syncs during autotuning. Build #81: Issue comment #116 (comment) created by romerojosh
10m 46s
10m 46s
Fix verbose NVSHMEM_CUMEM_GRANULARITY warning. Build #80: Issue comment #115 (comment) created by romerojosh
6m 52s
6m 52s
Add new NVSHMEM transpose communication backend with SM-based P2P copies. Build #79: Issue comment #114 (comment) created by romerojosh
7m 25s
7m 25s
Adjust workspace sizing to allow 256-byte alignment of workspace offset pointers. Build #78: Issue comment #109 (comment) created by romerojosh
5m 56s
5m 56s
Add workaround for cuTENSOR 2.3.x - 2.5.x OOB host write bug impacting large problem sizes Build #77: Issue comment #113 (comment) created by romerojosh
6m 28s
6m 28s
Use native NVSHMEM synchronization APIs in NVSHMEM backends Build #76: Issue comment #111 (comment) created by romerojosh
5m 44s
5m 44s
Use native NVSHMEM synchronization APIs in NVSHMEM backends Build #75: Issue comment #111 (comment) created by romerojosh
5m 38s
5m 38s
Force workspace usage with MPI backends on MNNVL communicators when fabric allocated workspace is available. Build #74: Issue comment #108 (comment) created by romerojosh
8m 1s
8m 1s
Force workspace usage with MPI backends on MNNVL communicators when fabric allocated workspace is available. Build #73: Issue comment #108 (comment) created by romerojosh
6m 30s
6m 30s
Use native NVSHMEM synchronization APIs in NVSHMEM backends Build #72: Issue comment #107 (comment) created by romerojosh
10m 54s
10m 54s
Use native NVSHMEM synchronization APIs in NVSHMEM backends Build #71: Issue comment #107 (comment) created by romerojosh
5m 29s
5m 29s
Use native NVSHMEM synchronization APIs in NVSHMEM backends Build #70: Issue comment #107 (comment) created by romerojosh
4m 58s
4m 58s
Enforce NVSHMEM minimum version of 2.6.0 to remove old workarounds. Update NVSHMEM usage guidance. Build #69: Issue comment #106 (comment) created by romerojosh
5m 39s
5m 39s
Enforce NVSHMEM minimum version of 2.6.0 to remove old workarounds. Update NVSHMEM usage guidance. Build #68: Issue comment #106 (comment) created by romerojosh
4m 25s
4m 25s
Improve support for very large inputs. Build #67: Issue comment #105 (comment) created by romerojosh
6m 10s
6m 10s
Mininum changes to compile HIP version Build #66: Issue comment #104 (comment) created by loostrum
1s
1s
Relax empty pencil check during grid descriptor creation to allow more flexibility Build #65: Issue comment #103 (comment) created by romerojosh
6m 30s
6m 30s
autotuneHaloBackend aggressively rejects valid decompositions for flat 3D grids (e.g. NxNx1) Build #64: Issue comment #101 (comment) created by romerojosh
1s
1s
Enabling support for process decompositions with empty pencils. Build #63: Issue comment #102 (comment) created by romerojosh
1s
1s
autotuneHaloBackend aggressively rejects valid decompositions for flat 3D grids (e.g. NxNx1) Build #62: Issue comment #101 (comment) created by romerojosh
1s
1s
autotuneHaloBackend aggressively rejects valid decompositions for flat 3D grids (e.g. NxNx1) Build #61: Issue comment #101 (comment) created by romerojosh
2s
2s
autotuneHaloBackend aggressively rejects valid decompositions for flat 3D grids (e.g. NxNx1) Build #60: Issue comment #101 (comment) created by romerojosh
1s
1s
0.6.1 release Build #59: Issue comment #100 (comment) created by romerojosh
8m 13s
8m 13s
Add CUDECOMP_DISABLE_MNNVL debug option. Build #58: Issue comment #99 (comment) created by romerojosh
7m 32s
7m 32s
Add CUDECOMP_DISABLE_MNNVL debug option. Build #57: Issue comment #99 (comment) created by romerojosh
8m 5s
8m 5s