adding benchmarking #3
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 3 commits
July 30, 2024 13:11
Summary: Test Plan: Reviewers: Subscribers: Tasks: Tags:
bertmaher
requested changes
Aug 15, 2024
models/llama/llama/mathOps.py
Outdated
| @@ -0,0 +1,143 @@ | |||
| import torch.nn.functional as F | |||
There was a problem hiding this comment.
This file should be math_ops.py to keep with standard conventions
@bertmaher fixed! was only in the test plan the params are properly named in the main.py file :) |
bertmaher
approved these changes
Aug 15, 2024
benchmarking/profiler.py
Outdated
Comment on lines
34
to
39
| start_time = time.perf_counter() | ||
|
|
||
| with torch.profiler.record_function(name): | ||
| result = func(*args, **kwargs) | ||
|
|
||
| end_time = time.perf_counter() |
There was a problem hiding this comment.
perf_counter isn't the right way to measure GPU time. Use torch.cuda.Event instead:
https://discuss.pytorch.org/t/how-to-measure-time-in-pytorch/26964
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
adding benchmarking by pulling out all the parts of torch we want to replace with triton and adding them to a single operators file. I also had to make some changes to the existing kernels since they had relative paths to triton itself
Test Plan:
made sure I didnt break anything so far
python3.9 -m main llama_chat_completion --profile=False --benchmark=True --ckpt_dir="models/llama/meta-llama/Meta-Llama-3-8B-Instruct/original" --tokenizer_path="models/llama/meta-llama/Meta-Llama-3-8B-Instruct/original/tokenizer.model" --use_triton=FalseOutput from benchmarks: