Refactor(PrefixCache): New load API, per-layer Tries, async ops & stats #269
+736
−239
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
vipannalla
reviewed
May 7, 2025
mailvijayasingh
approved these changes
May 7, 2025
github-actions
bot
added
the
pull ready
yuyanpeng-google
force-pushed
the
yuyan-prefix-cache
branch
2 times, most recently
from
fc0d025 to
8e22444
Compare
Add async to prevent device_get blocking on the critical paths waiting prefill result. Use per-layer tries to prevent load cache from DRAM when common length tie. Add statistic for debug and benchmark.
yuyanpeng-google
force-pushed
the
yuyan-prefix-cache
branch
from
8e22444 to
f40d0da
Compare
There is no formally result before this PR. There is just some causal small experiments and found that the device_get would block at the critical path. After this PR there is a first version benchmark result in b/397854862 |
jyj0w0
approved these changes
May 9, 2025
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Outdated suggestions cannot be applied.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.
Add async to prevent device_get blocking on the critical paths waiting prefill result. Use per-layer tries to prevent load cache from DRAM when common length tie. Add statistic for debug and benchmark.