Repository navigation
[GG] perf(dcp): split prefill queries and gather selected CKV #111
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Closed
voipmonitor
wants to merge
7
commits into
dev/gilded-gnosis-old-base
from
codex/gg-dcp-prefill-ckv-query-split-20260717
Closed
Changes from all commits
Commits
Show all changes
7 commits
Select commit
Hold shift + click to select a range
8f2bcf9
Add DCP prefill query split (Fix A) and CKV gather with layer prefetc…
koush 0f063fa
Remove dead _env_flag, add cache geometry stability comment
koush 0e4a576
Fix B12X CKV-gather prefetch: enable it and fix correctness
koush 964a904
fix(dcp): keep CKV communicators opt-in and format paths
voipmonitor dcb41c9
Fix CKV-gather small-context truncation of selected KV
koush 6d50564
fix(dcp): keep CKV metadata allocations opt-in
voipmonitor fdc6599
test(dcp): cover CKV gather global causal lengths
voipmonitor File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
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
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
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
Oops, something went wrong.
Oops, something went wrong.
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🩺 Stability & Availability | 🔴 Critical | ⚡ Quick win
Keep query-split groups inside each TP cohort.
group_rankscontains DCP groups from every DP/PP/PCP cohort. Collecting each DCP position across the entire list merges unrelated replicas. For DP2 with TP8/DCP8, this even creates two-rank query groups instead of the intended singleton no-op, risking incorrect outputs or collective hangs.Proposed fix
📝 Committable suggestion
🤖 Prompt for AI Agents