WASM speed test

A place to compare JS vs WASM vs native speeds.

Method
How code is executed
Run the benchmark
Results

Method

The benchmark is calculating PI with a Leibniz formula

The algorithm is implemented in

Plain JavaScript
Plain C
AssemblyScript

How code is executed

You can find the algorithm implementations in the implementations directory
You can find executing scripts in then runners directory

Plain JavaScript

Not compiled AOT (obviously)
Executed by node v16.13.2
Loaded by a require function
Measured with performance.now()

Plain C

Compiled with the clang (cc) compiler

Apple clang version 13.1.6 (clang-1316.0.21.2.5)

cc implementations/plainC.c -O2 -o build/plainC

Executed natively

./runners/plainC

Measured with clock() function from time.h

AssemblyScript

Compiled with asc v0.20.13

node_modules/.bin/asc implementations/assemblyScript.ts --target release

Node

Executed by node v16.13.2
Loaded with WebAssembly.instantiate
Measured with performance.now()

WASM3 CLI

Executed by wasm3 v0.5.0
Loaded with --func argument
Measured with the time command

PyWasm3

Executed by python 3.10.4
Loaded with pywasm3 v0.5.0
Measured with time.time() function

Run the benchmark

Install dependencies

assemblyscript for all AssemblyScript benchmarks
```
npm install
```
wasm3 for AssemblyScript -> WASM -> WASM3 CLI benchmark
```
# macOS: 
brew install wabt
```
pywasm3 for AssemblyScript -> WASM -> PyWASM3 benchmark
```
pip3 install pywasm3
```

Build C and AssemblyScript

Build all

npm run build

AssemblyScript only

npm run build:assembly-script

C only

npm run build:c

Actually run benchmarks

Run all

npm run start

Plain JavaScript only

npm run start:js

Plain C only

npm run start:c

WASM on node only

npm run start:wasm-node

WASM3 CLI only

npm run start:wasm3-cli

PyWASM3 only

npm run start:pywasm3

Results

Measured on a MacBook Air M1 16Gb RAM

Method	Time, ms
Plain JS	4861
Plain C	4119
WASM on node	4035
wasm3 CLI	39327
pywasm3	39652

Gibitop/wasm-speed-test