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

WIP: Fix GPU checkpointing #19

Draft
wants to merge 11 commits into
base: master
Choose a base branch
from
Draft

WIP: Fix GPU checkpointing #19

wants to merge 11 commits into from

Conversation

fstracke
Copy link
Collaborator

@fstracke fstracke commented May 17, 2023

In-Kernel checkpoint/restart has been broken for a while. The current code also only supports Pascal / Compute Capability 6 Kernels. This PR addresses these issues.

  • fix current code to work with latest CUDA version (still only Pascal GPUs)
  • remove cuda-gdb dependency and use libelf instead
  • add support for Ampere / Compute Capability 8 kernels

The nbody sample might not always compile.
Added simple check to see if nbody executabel is present.
The mpfr library needed to be linked
- Removed two absolute paths in start command.
- Fixes issue in said command.
- The location of cricket can now be specified using CRICKET_PATH variable.
Make would stop compilation of program if a link existed.
This is now fixed with a simple conditional.
Signed-off-by: Fritz Stracke <fritz.stracke@rwth-aachen.de>
make clean would encounter an error in cpu. fixed.
make clean now removes bin directory.

Signed-off-by: Fritz Stracke <fritz.stracke@rwth-aachen.de>
Signed-off-by: Fritz Stracke <fritz.stracke@rwth-aachen.de>
@fstracke fstracke marked this pull request as draft May 25, 2023 09:21
Signed-off-by: Fritz Stracke <fritz.stracke@rwth-aachen.de>
Signed-off-by: Fritz Stracke <fritz.stracke@rwth-aachen.de>
@n-eiling n-eiling added the doing label Jun 6, 2023
@n-eiling n-eiling changed the title WIP: Improve GPU checkpointing WIP: Fix GPU checkpointing Jun 6, 2023
Signed-off-by: Fritz Stracke <fritz.stracke@rwth-aachen.de>
@n-eiling n-eiling removed the doing label Dec 30, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants