v2 vs v1 — Performance (v2, Beta)

The measured v2-vs-v1 comparison — prediction accuracy and latency, cold-start warm-up, scaling, memory/disk footprint, write throughput, and a direct Elasticsearch head-to-head — now lives in one place so every number has a single home and can't drift between copies:

→ Benchmarks → Write & query performance

The short version from that page:

  • Accuracy (10M rows). On acceptor, v2 already matches the mature v1 engine. glCode and processor — the two targets that lean on the cross-table re-expression (whitening) path v1 has tuned and v2 hasn't yet — still trail on accuracy; that's the understood, concentrated gap, not an engine-wide one. See the performance benchmarks for the per-target numbers.
  • Latency. Interactive — roughly a tenth to a half a second per predict at 10M rows — with v2 cold-start markedly faster than v1 on the lower-cardinality targets (memory-mapped columnar reads).
  • Footprint & scale. Predict latency grows sub-linearly with data on the tested targets (interactive — sub-second per predict — through 10M rows), a leaner memory-mapped columnar footprint, and incremental writes (no reindex step).

v2 is in beta — a parallel codepath under active development, measured, not yet the default — so read these as "where rep2 is now," not a release claim.