microsoft / microsoft/TileFusion
[Bug]: GEMM Example 1 crashing
Open
Nobody has claimed this yet.
- Dominant language
- Cuda
- Stars
- 118
- Forks
- 7
- PR merge metrics
- No merged PRs in 30d
Description
Example 1 for GEMM 01_gemm_global_reg crashes during kernel for default problem size:
- M = 1024
- N = 1024
- K = 2048
- TileM = 256
- TileN = 128
- TileK = 64
System:
OS: Ubuntu 24.04.2
Device: RTX 3070 Mobile (SM86)
Driver: 560.35.03
CUDA Version: 12.6
Output:
kThreads: 128
RegA: RowMajor(16, 16)
RegB: ColMajor(16, 8)
RegC: RowMajor(16, 16)
IteratorA: numel = 524288, ChunkShape = (256, 64), stripe count = (1, 32)
IteratorB: numel = 262144, ChunkShape = (64, 128), stripe count = (32, 1)
blocks: [4, 8]
terminate called after throwing an instance of 'thrust::THRUST_200400_860_NS::system::system_error'
what(): trivial_device_copy D->H failed: cudaErrorIllegalAddress: an illegal memory access was encountered
Aborted (core dumped)
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the 01_gemm_global_reg GEMM example and reproduce the crash using its default problem size on the reported CUDA setup. Trace the kernel execution and the failing trivial_device_copy path; done means the example completes without the reported cudaErrorIllegalAddress crash.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- hpc
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100