@piman51277/fast-format

fast-format

A faster implementation of @danocmx's node-tf2-item-format/static. This repo provides an alternative ISchema that is optimized for speed.

Memory usage will be higher than the original implementation.

Why?

The author of node-tf2-item-format has stated that performance in static schema is not a priority. However, I have run into multiple situations where the performance of node-tf2-item-format/static becomes a major bottleneck. This package is an attempt to address that.

Usage

In general, this package is a drop-in replacement for node-tf2-item-format/static (v5) or node-tf2-item-format (v4).

TS

//v4 or v5/static style
import { parseEconItem } from "@piman51277/fast-format";

//v5 style
import { FastSchema } from "@piman51277/fast-format";
import { createFormat } from "tf2-item-format";

const format = createFormat(new FastSchema());

format.parseEconItem = format.parseEconItem.bind(format);
format.parseString = format.parseString.bind(format);
format.stringify = format.stringify.bind(format);
format.fixName = format.fixName.bind(format);
format.createBPListing = format.createBPListing.bind(format);
format.toSKU = format.toSKU.bind(format);
format.parseSKU = format.parseSKU.bind(format);

Note, due to the incompatibilities between CommonJS and ES Modules, you will need to use the default export when using CommonJS.

JS

//v4 or v5/static style
const {
  default: { parseEconItem },
} = require("@piman51277/fast-format");

For more information, refer to node-tf2-item-format's README

Performance

Across the board, fast-format performs around the same or better than node-tf2-item-format/static. Notably, parseEconItem (with defindexes=true) is around 2x faster than the original implementation.

Methodology

Benchmarking was done by parsing random EconItems from a pool of 1000 items in the author's inventory. Link to backpack

Precompute steps are not included, by means of running the functions multiple times before the benchmark.

Benchmark Machine Specs:

CPU:Intel i7-1360P @ 5.0GHz
RAM: 32GB of DDR5 6000Mhz
OS: Debian 12.0

tf2-item-format version: 5.9.0

tf2-static-schema version: 1.49.0

Results

(You can check these yourself using the npm run benchmark command)

parseEconItem defindex=true
fast-format x 32,351 ops/sec ±0.41% (93 runs sampled)
tf2-item-format/static x 16,913 ops/sec ±0.76% (94 runs sampled)
Fastest is fast-format

parseEconItem defindex=false
fast-format x 34,066 ops/sec ±0.35% (98 runs sampled)
tf2-item-format/static x 33,983 ops/sec ±0.44% (99 runs sampled)
Fastest is fast-format,tf2-item-format/static

toSKU
fast-format x 8,713,657 ops/sec ±0.23% (96 runs sampled)
tf2-item-format/static x 8,724,382 ops/sec ±0.24% (99 runs sampled)
Fastest is tf2-item-format/static,fast-format