perf(vamana): port search and prefetch optimizations - #727
Open
luoxiaojian wants to merge 12 commits into
Open
Conversation
luoxiaojian
requested review from
Cuiyus,
JalinWang,
chinaux,
iaojnh,
richyreachy and
zhourrr
as code owners
September 7, 2026 07:29
Reuse the unsigned query norm in VNNI and portable preprocessing without changing the encoded layout or distance formula. Recompute oversized norms to preserve fallback output, and cover canonical and overflow boundaries in tests.
Keep raw PO/PL requests separate from resolved values so partial updates retain automatic defaults and schema-clamped overrides. Resolve untouched streaming contexts when preparing graph queries, refresh resolved settings after entity changes, and leave construction initialization unchanged. Cover parameter lifecycle and first contiguous graph queries with regression tests.
Cover quantized batch boundaries, fast-search transitions and pool API equivalence. Add a single-thread two-pass contiguous benchmark, smoke checks, and a deferred Ice Lake validation guide with precise prefetch and timing semantics.
Move pop and pop_with_next into the header with force-inline annotations and share cursor advancement. Keep AVX2 block insertion out of line and retain ByteMap. Add mixed-interface rewind coverage and update the Ice Lake validation notes.
luoxiaojian
force-pushed
the
codex/vamana-search-prefetch-optimizations
branch
from
September 7, 2026 07:31
d491658 to
26d6b39
Compare
Preserve the aligned graph-row range while replacing the local address loop with ailego_prefetch_lines. Record the Ice Lake code-generation difference for later same-graph performance validation.
Drop the benchmark executable, utilities, smoke script and dedicated tests from the branch. Keep production optimizations and targeted regressions, and retain an Ice Lake validation plan without a new benchmark dependency.
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.
No description provided.