Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[FEA] GFQL Graph500 BFS benchmark & writeup #534

Open
9 tasks
lmeyerov opened this issue Dec 25, 2023 · 0 comments
Open
9 tasks

[FEA] GFQL Graph500 BFS benchmark & writeup #534

lmeyerov opened this issue Dec 25, 2023 · 0 comments

Comments

@lmeyerov
Copy link
Contributor

lmeyerov commented Dec 25, 2023

Is your feature request related to a problem? Please describe.

GFQL can likely do decently on Graph500. Worth having code + writeup

It may be worth taking a quick look at https://ldbcouncil.org/ as an alternative as well

Describe the solution you'd like

Benchmark:

  • BFS benchmark applicability: hop() seems to correspond to Graph500 BFS, afaict . The benchmark may frown on the extreme parallelism GFQL unlocks, but that seems OK for now? We should investigate
  • We should benchmark T4 (cheap gpu) + some sort of modern GPU with bigger memory + more cores
  • I'd be curious about the impact of some sort of UVM for bigger-than-memory operation before resorting to dask_cudf etc . E.g., how do the GTEPS/MTEPS look once we're bigger than memory?
  • We should check if any other benchmarks seem applicable as well

Writeup:

  • executable script
  • that generates a valid benchmark submission
  • that generates charts
  • ipynb for lightweight use
  • and a blogpost :)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant