Add kv cache hit source breakdown (device/host/storage) to step profiling#32
Open
Weili-0234 wants to merge 1 commit into
Open
Add kv cache hit source breakdown (device/host/storage) to step profiling#32Weili-0234 wants to merge 1 commit into
Weili-0234 wants to merge 1 commit into
Conversation
When SGLang has HiCache or LMCache enabled, responses include sglext.cached_tokens_details with per-tier hit counts: - device: GPU KV cache - host: CPU memory (HiCache L2 / LMCache host buffers) - storage: L3 backend (when configured) Changes: - vllm_request_processor.py: extract cached_tokens_details from sglext, add maybe_enable_cached_token_details() to request return_cached_tokens_details when profiling is enabled, extend on_usage callback signature - profile/state.py: add cached_tokens_device/host/storage and cached_storage_backend fields to StepMetrics and CSV output - app.py: pass cached_tokens_details through on_usage callback
ergt10
pushed a commit
to ergt10/ThunderAgent
that referenced
this pull request
Apr 4, 2026
…g#32) # What does this PR do? Improves installation instructions in readme and docs. The quick start on the README is meant for single node with 4 GPUs. (ray need not even be installed, because `ray.init` will start a cluster if not running) The installation guide is also updated. I have verified that the quickstart works on a fresh instance. --------- Signed-off-by: SumanthRH <sumanthrh99@gmail.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When SGLang has HiCache or LMCache enabled, responses include sglext.cached_tokens_details with per-tier hit counts:
Changes: