Try rewriting bcrypt-opencl to use recent GPUs' shuffles
Nobody has claimed this yet.
- Dominant language
- C
- Stars
- 13.6k
- Forks
- 2.6k
- PR merge metrics
- No merged PRs in 30d
Description
It might be possible to make bcrypt-opencl up to 4x faster through having the 4 parallel S-box lookups actually done in parallel and then the results transferred to one work-item using recent GPUs' shuffle instructions.
For NVIDIA, see the last few paragraphs in:
https://lists.openwall.net/phc-discussions/2018/04/28/1
For AMD:
https://gpuopen.com/amd-gcn-assembly-cross-lane-operations/
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
Locate the bcrypt-opencl implementation in the repository and read the linked NVIDIA and AMD cross-lane-operation references first. Determine whether the four parallel S-box lookups can use recent GPU shuffle instructions, then benchmark supported GPUs; done means a working rewrite with measured performance improvement.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- c
- Domain
- performance
- Issue type
- Refactor
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100