jackylee-ch opened a new pull request, #92: URL: https://github.com/apache/paimon-vector-index/pull/92
`scan_4bit_simd` builds the u8 quantization factor from `qmax`, the largest *summed* distance over the first 200 rows, then applies it to individual table entries. Entries are ~1/m of a sum, so the table collapses into the low `255/m` of the u8 range and rows past 200 rank worse. `fastscan.rs` already scales by the table's own max; this reuses its `quantize_distance_table`. Tests: - new `test_scan_4bit_simd_large` fails before this change (row 200 off by 4.04, tolerance 2.80), passes after - `cargo test --release --workspace` (503 passed) - `cargo fmt --all -- --check` - `cargo +1.98.0 clippy --release --all-targets --workspace -- -D warnings` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
