Use new TPU segment option #99
Merged
Conversation
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.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.
The new TPURunner architecture allows for any number of TPUs to share the work on any model segment. This way, with multi-TPU systems, we aren't stuck with a single segment being a bottleneck. Additional TPUs are dynamically allocated from fast segments to help with slow segments. This opens up additional strategies with how the segments are compiled and allocated for systems with many TPUs. Many-TPU systems also scale up much more elegantly, as do systems with mixed TPU architectures or interface types.
New segment archives can be downloaded from this archive. Also contains ipcam-general for YOLOv8:
https://drive.google.com/file/d/1EZMvBW9ykcj3k-SZPuUl5gIDDSiRR7lU/view?usp=sharing