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

Performance issue: slurm commission is too slow #1829

Closed
shchen-idmod opened this issue Sep 7, 2022 · 6 comments
Closed

Performance issue: slurm commission is too slow #1829

shchen-idmod opened this issue Sep 7, 2022 · 6 comments
Assignees
Labels
Architecture COMPS python Pull requests that update Python code SLURM

Comments

@shchen-idmod
Copy link
Collaborator

shchen-idmod commented Sep 7, 2022

Running this in our slurm cluster: 10.24.15.109
with example: https://github.com/InstituteforDiseaseModeling/idmtools/blob/dev/examples/native_slurm/python_sims.py

Change sweep a with 30, b with 500:
builder.add_sweep_definition(partial(param_update, param="a"), range(30))
builder.add_sweep_definition(partial(param_update, param="b"), range(500))
...
dry_run=True:
suite.run(platform=platform, wait_until_done=True, max_running_jobs=10,
retries=5, dry_run=True)

run example:
Commissioning Simulations: 5%|███▌ | 768/15000 [11:04<19:53, 11.92simulation/s]
It about commission 500 sims/10mins. I think this is way too slow. To finish commission15000 sims, may need at least 5 hours.

@issuelabeler issuelabeler bot added the python Pull requests that update Python code label Sep 7, 2022
@issuelabeler issuelabeler bot added the COMPS label Jan 18, 2023
@shchen-idmod
Copy link
Collaborator Author


Meanwhile comps only took 5 mins to commssion for same amount of simulations
C:\venv\idmtools_dev_173\Scripts\python.exe C:/github/idmtools/examples/native_slurm/python_sims_comps.py
INI File Used: C:\github\idmtools\examples\idmtools.ini

[SlurmStage]
{
"endpoint": "https://comps2.idmod.org",
"environment": "SlurmStage"
}
Uploading 1 files/613 bytes
Uploading files: 100%|██████████| 1/1 [00:01<00:00, 1.35s/file]
Initializing objects for creation: 0simulation [00:00, ?simulation/s]
The created experiment can be viewed at https://comps2.idmod.org/#explore/Simulations?filters=ExperimentId=5992db17-6697-ed11-92f2-f0921c167864
Simulations are still being created

Creating Simulations on Comps: 100%|██████████| 15000/15000 [05:38<00:00, 44.31simulation/s]

@shchen-idmod
Copy link
Collaborator Author

This may relate to perf issue Anne from NU seeing:
NU-image

@shchen-idmod
Copy link
Collaborator Author

her example:
perf_example.txt

@ZDu-IDM
Copy link
Collaborator

ZDu-IDM commented Jan 19, 2023

Have discussion with Clinton and it seems like there are several possible reasons that may cause the commission performance issue. We are investigating the issue now but not sure when to get it fixed (as well-known, performance issue is always hard to debug and fix).

@shchen-idmod
Copy link
Collaborator Author

See update perf number in ticket #1996. Fix seems very fast now.

@ZDu-IDM
Copy link
Collaborator

ZDu-IDM commented Feb 9, 2023

Done with idmtools release 1.7.4

@ZDu-IDM ZDu-IDM closed this as completed Feb 9, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Architecture COMPS python Pull requests that update Python code SLURM
Projects
None yet
Development

No branches or pull requests

2 participants