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

VID experiments with many CPU cores #484

Open
Tracked by #358
akonring opened this issue Feb 8, 2024 · 2 comments
Open
Tracked by #358

VID experiments with many CPU cores #484

akonring opened this issue Feb 8, 2024 · 2 comments
Assignees
Labels
cappuccino-sprint2 optimize-vid https://www.notion.so/espressosys/9d835f79d4504926b8b3bb3d015abf06?v=b7028cdaea804b7aa918af95c0cd651 vid

Comments

@akonring
Copy link
Contributor

akonring commented Feb 8, 2024

Do VID benchmarks with many CPU cores:

  1. Try (locally) to identify a trend when measuring the duration of disperse while increasing the number of available CPU cores.
  2. Spin up AWS machine with 32-64 cores to confirm the trend more powerful instances.
@akonring akonring changed the title VID experiments with many CPU VID experiments with many CPU cores Feb 8, 2024
@akonring akonring added vid optimize-vid https://www.notion.so/espressosys/9d835f79d4504926b8b3bb3d015abf06?v=b7028cdaea804b7aa918af95c0cd651 labels Feb 8, 2024
@akonring akonring self-assigned this Feb 8, 2024
@akonring
Copy link
Contributor Author

Using RAYON_NUM_THREADS as a proxy for locally confirming the effect of increasing the number of accessible CPUs.
Justification: Since Rayon is the primary (parallelization) workhorse in the VID disperse computation (FK20, MSMs), it makes sense to limit its access to threads to identify the trend. Furthermore, Rayon defaults to allocating one thread per core (logical if hyper-threading is supported).

@akonring
Copy link
Contributor Author

All findings documented in this Notion document:
https://www.notion.so/espressosys/VID-Optimization-2-CPU-cores-666dfb5c4ae04e9c9b306c210af803a9

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
cappuccino-sprint2 optimize-vid https://www.notion.so/espressosys/9d835f79d4504926b8b3bb3d015abf06?v=b7028cdaea804b7aa918af95c0cd651 vid
Projects
None yet
Development

No branches or pull requests

2 participants