All benchmark types
Vector / ANN
Approximate-nearest-neighbour recall deficits (HNSW / Vamana / PQ) vs exact kNN.
Latest commit a40cf2a1 · every metric smaller-is-better · each group header shows a competitive summary computed live from real same-box competitor numbers (or an honest placeholder where none exist yet).
No same-box competitor baseline has been gathered for this suite yet — sparq's absolute numbers are shown below.
| Benchmark | sparq | Unit |
|---|---|---|
Vector / ANN diskann — query latencyvectors_diskann_query_us | 347.4 | us |
Vector / ANN diskann — recall@10 deficit, gatevectors_diskann_recall_at10 | 34 | milli |
Vector / ANN hnsw — query latencyvectors_hnsw_query_us | 450.8 | us |
Vector / ANN hnsw — recall@10 deficit, gatevectors_hnsw_recall_at10 | 2 | milli |
Vector / ANN pq — query latencyvectors_pq_query_us | 430.5 | us |
Vector / ANN pq — recall@10 deficit, gatevectors_pq_recall_at10 | 22 | milli |
Vector / ANN — HNSW index build timevectors_build_s | 40.99 | s |