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.

Sign up for GitHub

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

Jump to bottom

Parallel Tuning #241

Open
wants to merge 105 commits into
base: master
Choose a base branch
from
Open

Parallel Tuning #241

wants to merge 105 commits into from

Conversation

MiloLurati
Copy link
Contributor

@MiloLurati MiloLurati commented Jan 30, 2024 •
edited
Loading

This pull request introduces a new ParallelRunner designed to compile and benchmark kernel configurations in parallel across multiple GPUs. Additionally, it implements an ensemble strategy to execute various strategies concurrently on multiple GPUs.

@sonarqubecloud SonarQubeCloud
Copy link

Quality Gate Failed Quality Gate failed

Failed conditions

25.6% Duplication on New Code (required ≤ 3%)

See analysis details on SonarCloud

Milo Lurati and others added 21 commits April 2, 2024 16:07
changed naming from remote to parallel
fe5da39
added get_num_devices function
a43dc84
added ensemble and parallel runner related stuff
ab3aa24
switched to new naming of parallel remote and some clean up
e8a7228
added class instances needed down the line in the execution of the en…
3dd748c 
…semble strategy
changed naming due to ensemble implementation, this was the original …
e743bec 
…remote_actor
started ensemble implementation, very basic functionality works
df949d0
updated tests
45a1747
clean up in parallel runner
5fb5927
moved to sub directory ray
a96ef43
added subdirectory ray with all 3 actor classes
c831f5f
itegrated calls to cache manager functions when running in ensemble
0cc2a6e
added cache manager logic
b816f3d
added instances needed for the ensemble down the line of execution
781839a
added strategy option to get_options function
9f8d212
added ignore_reinit_error to ray init
d08b5d4
added ignore_reinit_error to ray init
903c981
added cache manager to parallel tuning
1a2219a
re-assign tuning options to final version from the cache manager at t…
a476585 
…he end of the parallel runner execution
@MiloLurati
small bug fix in execute
6233e09
@MiloLurati
Merge branch 'KernelTuner:master' into parallelTuning
cde62ae
@sonarqubecloud SonarQubeCloud
Copy link

Quality Gate Failed Quality Gate failed

Failed conditions
E Reliability Rating on New Code (required ≥ A)

See analysis details on SonarCloud

Catch issues before they fail your Quality Gate with our IDE extension SonarLint

MiloLurati and others added 16 commits June 18, 2024 10:52
@MiloLurati
moved reinitialization of actor observers to execute method, before w…
e604510 
…as in init
@MiloLurati
changes related to re-initialization of observers in actor init and d…
5933a69 
…evice interface
@MiloLurati
removed unnecesary blocking ray.get
4e4c47b
@MiloLurati
removed debug prints
104205d
@MiloLurati
added greedy ils esemble instead of default
123fba5
@MiloLurati
added check on strategy_options
d381011
@MiloLurati
removed all memetic algo related stuff
7e832e3
@MiloLurati
Merge branch 'KernelTuner:master' into parallelTuning
e976bf8
@MiloLurati
added ray to pyproject.toml
65d32c1
@MiloLurati
Merge branch 'parallelTuning' of https://github.com/MiloLurati/kernel…
a841f2a 
…_tuner into parallelTuning
@MiloLurati
updated toml file with ray dashboard
503df1b
@MiloLurati
fix small bug in _evaluate_configs
c126a01
@MiloLurati
adapted test for ensemble
4df1b0d
@MiloLurati
cleaned up not used imports
29a507c
@MiloLurati
added comments
7c49a29
@MiloLurati
added documentation and related fixes
eb5db41
@MiloLurati MiloLurati marked this pull request as ready for review September 30, 2024 13:50
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@MiloLurati