Repository navigation
Plan: contiguous fast path for the 16-byte key match (D-KEY16) - #1475
Conversation
|
Warning Review limit reachedYour organization has reached its usage spending cap. Adjust your spending cap in the billing tab. Next included review available in 14 minutes. View limit details
Warning Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption. Comment |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6aa37c1790
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
ndarray's ternary_match_strided16_to_mask gathers each 16-byte key with two scalar loads even at stride 16, where one U64x8 load holds four keys. The plan adds a facade-only stride-16 path and measures an AVX-512 gather for wider strides, both gated on a probe with pre-registered thresholds. Parity coverage for the existing kernel comes first: it is missing from ndarray's simd-masking-parity. The Rust 1.99 u128-in-vector-register asm! path is recorded as rejected, with its reason. Consumers (mask-risc, quack, dir-sim) change no code. Board: INTEGRATION_PLANS and STATUS_BOARD rows prepended. The supersession index was regenerated last and is unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015JH7YkXs5us6g3tG8NiFfP
The gate's pattern needs at least two letters after D-, so D-K16 was invisible to it (measured: plan_dids.py exit 1 before, 0 after). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015JH7YkXs5us6g3tG8NiFfP
6aa37c1 to
a3aa459
Compare
The kernel's input is &[u8] and the first offset may be 1..15, so building a &[u64] for U64x8::from_slice is undefined behaviour. W3 now adds a byte load, U64x8::from_le_bytes, to the facade and every backend, with its own parity case at offsets 0..63; the probe's arm B uses the same load. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015JH7YkXs5us6g3tG8NiFfP
Plan only; no code. Adds
.claude/plans/key16-contiguous-match-v1.mdand its board rows.Why.
ndarray::simd::ternary_match_strided16_to_mask(behind mask-risc'sPred::MatchFacet16Strided, quack'sCmp::MatchFacet16Stridedand dir-sim'ssubtree) gathers every 16-byte key with two scalar 8-byte loads, even at stride 16, where oneU64x8load already holds four whole keys. The compare after the gather is already a fusedXOR_ANDternlog; the gather is the cost.What the plan does.
simd-masking-parity, so no backend besides the in-crate tests checks it.Rejected, with reason: Rust 1.99's
u128-in-vector-registerasm!path. One key perxmminstruction is slower than eight 64-bit halves perU64x8; ndarray's 69asm!sites pass no 128-bit values.Board:
INTEGRATION_PLANS.mdandSTATUS_BOARD.mdrows prepended (line counts grew 3770→3779 and 2449→2462).plan_dids.pypasses on the new plan; the supersession index was regenerated last and is unchanged. The second commit renames the series fromD-K16toD-KEY16, because the gate's pattern needs two letters afterD-.The ndarray work (D-KEY16-1..4) needs a session with push access to
AdaWorldAPI/ndarray.🤖 Generated with Claude Code
https://claude.ai/code/session_015JH7YkXs5us6g3tG8NiFfP
Generated by Claude Code