Explore slow performance of `parallel=True` flag in the numpyro `Predictive` class.
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 13
- Forks
- 3
- Avg merge
- 19h 43m
- Merged PRs (30d)
- 1
Description
When using numpyro.Predictive there is a flag to parallelize the generation of the posterior predictive, at least on the VAP, this is substantially slower than the non-parallel mode. There is some discussion about it here which mentions parallel being slower on CPU, but can we test the difference with vs without GPU? May be worth it to get a single GPU azure node for all of our projections if it is dramatically faster.
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 numpyro.Predictive and its parallel flag, then review the linked forum discussion about CPU performance. Benchmark posterior predictive generation with parallel enabled and disabled, comparing CPU and GPU runs where available. Done means documenting the timing difference and whether a single GPU Azure node is justified for projections.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100