OpenCL auto-tune: Try getting a sane figure for min_keys_per_crypt
Open
Nobody has claimed this yet.
- Dominant language
- C
- Stars
- 13.6k
- Forks
- 2.6k
- PR merge metrics
- No merged PRs in 30d
Description
See also #3779
While tuning for best GWS, we should note the first size that starts "flying" and set min_kpc to that.
When is that though? In some cases it seems we can pick the smallest GWS that doesn't almost double the wall clock time (it can even be faster!) from previous size. Or we could just pick the next best size (which sometimes is half of max_kpc and sometimes way lower).
Contributor guide
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 file, test, or entry point is named. Start by tracing the existing OpenCL auto-tuning logic for min_kpc and GWS, then review the discussion in #3779 and compare wall-clock measurements across sizes. Done means there is an agreed rule for selecting a sane min_kpc value.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- c
- Domain
- performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100