ZacharyZhang-NY
e80883cfc9
Both reports were produced by a harness that had just changed, so their stamps named a commit without that code in it. Re-running them on an idle machine puts the stamp and the source back in sync. Recall did not move: all seven leg configurations reproduce to the third decimal, which is the part of the number anyone should act on. Latency moved within its usual load band, and the encoder probe came in slower this time, so the picture index's build cost is now quoted as a band rather than a single remembered figure.
1.6 KiB
1.6 KiB
DENSE: the vector leg at corpus size (commit 71733d86d)
- database:
bench/runs/dense_scale/data/ely_eye.sqlite - atoms: 165973, memories with words: 153857, vectors: 153857, pending: 0
- model:
sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2, width 384 - build: 1444.9s for 153857 memories (106 per second)
- writing one memory into a full index: p50 508ms · p95 535ms over 50 writes
- the repair scan (
ely-eye build-dense-index) over a full index: 1.3s, and it is not on the search path - search p50 73.0ms · p95 75.7ms · max 85.6ms over 300 queries
- of which encoding the question: p50 11.3ms · p95 24.0ms
- against a brute-force scan: 0.998 of the top 10 over 40 probes (a returned atom counts when its true cosine is at least the k-th best, so ties cannot fail the check)
- bytes per stored vector: 1729 (266059776 bytes for 153857 vectors, dbstat page sizes over the vec0 shadow tables)
- machine: Linux 6.18.40.1-microsoft-standard-WSL2, 32 cores, 4 threads per query, load [3.53, 4.63, 4.85]
One million vectors on this machine
- p50 475ms · p95 492ms
- vector table: 1.6 GiB
- first build: 156.5 minutes
- writing one vector: 3303ms — a vec0 delete by id scans the table, so this grows with the count too
- assumption: sqlite-vec 0.1.9 scans every vector, so latency and bytes grow linearly with the vector count; every row here is scaled by 1000000 / 153857 vectors, and the model, the machine and the thread budget stay the same
Latency is wall time per search on the machine named above with the thread budget the benchmark pins; a run on a loaded machine is not comparable.