Add Triton benchmarks for blog #509
Open
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
Signed-off-by: Rishi Chandra <[email protected]>
rishic3
force-pushed
the
triton-benchmarks
branch
from
7c26cc0 to
2614369
Compare
eordentlich
reviewed
Mar 12, 2025
| 1. [`spark_resnet.py`](spark_resnet.py): Uses predict_batch_udf to perform in-process prediction on the GPU. | ||
| 2. [`spark_resnet_triton.py`](spark_resnet_triton.py): Uses predict_batch_udf to send inference requests to Triton, which performs inference on the GPU. | ||
|
|
||
| Spark cannot change the task parallelism within a stage based on the resources required (i.e., multiple CPUs for preprocessing vs. single GPU for inference). Therefore, implementation (1) will limit to 1 task per GPU to enable one instance of the model on the GPU. In contrast, implementation (2) allows as many tasks to run in parallel as cores on the executor, since Triton handles inference on the GPU. |
There was a problem hiding this comment.
For resnet-50 could multiple model instances fit in the GPU? If so, might be good to benchmark that case, where multiple spark tasks run per GPU with each having its own model instance. Due to multiple processes, GPU compute will be time sliced so perf could be hit, but still interesting to compare.
There was a problem hiding this comment.
Would it make sense to consolidate this script with spark_resnet.py and select library or triton via cli argument?
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Outdated suggestions cannot be applied.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.
Scripts, configs, and instructions to reproduce the blog benchmarks, displaying the benefit of using Triton for CPU parallelism..