Add integration test harness that bypasses the FastAPI server to directly test the batcher #959
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
…art_generate_service contextmanager
renxida
force-pushed
the
test-generation-request
branch
from
2094338 to
fa10955
Compare
|
Note that this introduces some code duplication in server_management.py. Will put in a follow up PR to dedupe the code there, and between model_management.py and sharktank's hf_datasets.py |
|
Temporarily dropping this in favor of a micro-server like SDXL's. |
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Outdated suggestions cannot be applied.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.
Previously, to test concurrent generation, we quickly submit multiple requests.
This does not consistently result in the same batching, and timing differences across shortfin can cause e.g. 4 requests to go from being all batches together to boarding a batch of 3 and then a batch of 1. This caused some very difficult-to-replicate errors and accounted for some of the regressions we observed leading up to the v3.2.0 release.
This harness and it's associated tests will eliminate inconsistent batching as an untested point of failure, allowing us to detect and replicate future errors that depend on certain batching conditions to appear.