Enable CUPTI to measure kernel execution time instead of CUDA Events
Open
Nobody has claimed this yet.
- Dominant language
- Cuda
- Stars
- 927
- Forks
- 123
- Avg merge
- 2d 10h
- Merged PRs (30d)
- 2
Description
CUDA events suffer from low accuracy and include the kernel launch overhead. On the other hand, CUPTI provides a more reliable way to get consistent timing measurement.
This request asks to add an option to replace CUDA Events with CUPTI.
Details
CUDA events issues:
- Accuracy and Stability:
cudaEventcan fluctuate in the range of 10-30us, making measurements of small computations unreliablecudaEventtake into account the kernel launch overhead that depends on host/CPU execution and/or driver version
CUPTI:
- ~0.5us granularity vs. 10-30us
- Not affected by kernel launch overhead
- Consistency: measurements close to the profiler (nsys)
- Efficiency: avoid using waiting/delay kernels to hide CPU overhead
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
No files, tests, or entry points are named. Start by locating the existing CUDA Events timing path and determine how an optional CUPTI path would fit the benchmark configuration. Done means kernel execution timing can use CUPTI while the existing CUDA Events approach remains available.
Written by the indexing model from the issue text.
Assessment
- Domain
- performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100