https://github.com/devidw/vector-db-benchs
https://github.com/devidw/vector-db-benchs
Last synced: 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/devidw/vector-db-benchs
- Owner: devidw
- Created: 2023-03-29T18:32:10.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2023-03-29T18:32:14.000Z (about 2 years ago)
- Last Synced: 2024-10-18T20:45:18.289Z (8 months ago)
- Language: TypeScript
- Size: 726 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: readme.md
Awesome Lists containing this project
README
# vector database benchmarks
comparing pinecone with supabase (+ pgvector extension) for embeddings records, running queries to find a given embedding
in these benchs operation region of both platforms is us-west and request region is eu-north
the used dataset can be found in `./dataset/embeddings.json` (100 one-words)
## Benchmarks
date: 2023-03-29
| Database | Hz | Mean | Variance | SD | P99 |
|------------|----------|----------|----------|----------|----------|
| Pinecone | 3.501599 | 285.5838 | 44023.23 | 209.8171 | 885.0765 |
| Supabase | 2.185156 | 457.6331 | 50348.20 | 224.3840 | 835.6537 |
full results can be found in `./results.json`
bases on these benchs the average search process using pinecone is roughly 1.6x faster than with supabase postgres + pgvector extension