mirror of
https://github.com/aljazceru/turso.git
synced 2025-12-25 12:04:21 +01:00
60cc8c0347
This adds a benchmark for "SELECT 1" that essentially just evaluates the overhead of calling into the library. Looks like WebAssembly adds 2.5x overhead compared to native code... ``` penberg@vonneumann perf % node perf-limbo.js cpu: Apple M1 runtime: node v18.19.0 (arm64-darwin) benchmark time (avg) (min … max) p75 p99 p999 ------------------------------------------------- ----------------------------- • Statement ------------------------------------------------- ----------------------------- SELECT 1 771 ns/iter (624 ns … 2'183 µs) 750 ns 1'085 ns 1'750 ns summary for Statement SELECT 1 penberg@vonneumann perf % node perf-better-sqlite3.js cpu: Apple M1 runtime: node v18.19.0 (arm64-darwin) benchmark time (avg) (min … max) p75 p99 p999 ------------------------------------------------- ----------------------------- • Statement ------------------------------------------------- ----------------------------- SELECT 1 302 ns/iter (166 ns … 779 µs) 292 ns 375 ns 1'000 ns summary for Statement SELECT 1 ```
12 lines
195 B
JSON
12 lines
195 B
JSON
{
|
|
"name": "limbo-perf",
|
|
"type": "module",
|
|
"private": true,
|
|
"dependencies": {
|
|
"limbo-wasm": "../pkg",
|
|
"better-sqlite3": "^9.5.0",
|
|
"libsql": "..",
|
|
"mitata": "^0.1.11"
|
|
}
|
|
}
|