Evaluation & Benchmarks

HyperspaceDB is optimized for two critical metrics: Throughput (Ingestion speed) and Latency (Search speed).

Test Environment

  • Hardware: Apple M4 Pro (Emulated Environment) / Linux AVX2
  • Dataset: 1,000,000 vectors, 1024 Dimensions, Random Distribution in Unit Ball.
  • Config: ef_construction=400, ef_search=400

Results

🚀 Ingestion Speed

Thanks to the Async Write Buffer (WAL) and background indexing, ingestion does not block user requests.

CountTimeThroughputStorage Sements
10,0000.6s15,624 vec/s1
100,0006.5s15,300 vec/s2
1,000,00064.8s15,420 vec/s15

🔍 Search Latency (1M Scale)

At 1 million vectors, search performance degrades linearly with graph depth ($\log N$), proving effective HNSW implementation.

MetricValue
QPS14,668 queries/sec
Avg Latency0.07 ms
P99 Latency< 1.0 ms

Why is it so fast?

  1. ScalarI8 Quantization: Fits 8x more vectors in CPU cache.
  2. No acosh: Inner loop uses a monotonic proxy function ($\delta$).
  3. SIMD: Vector operations use platform-specific intrinsics.