pygfx-benchmarks

Introducion

Benchmarks for wgpu and pygfx

This repository contains a collection of scripts that perform benchmarks on wgpu and pygfx. Their purpose include:

• To understand best practices for using wgpu.
• To make informed decisions in pygfx and other libraries.
• To track performance of pygfx logic, so that the effect of changes on performance can be measured.

Scripts go into the benchmarks folder. Matching results go into the results folder.

Writing a benchmark

Benchmarks are generator functions (functions that use yield). The function should accept a canvas object, but does not have to use it. This is so that visual results can be shown to see that the benchmark actually does what it is meant to do.

The code leading up to the first yield is for setting things up. After that, it should enter an infinite loop to do repeated measurements. The number of measurements can be set in the @benchmark decorator.

Each yield may contain a dict with additional (time) measurements (e.g. GPU times).

@benchmark(20)
def my_benchmark(canvas):

    foo = "set things up"
    yield  # done setting things up

    while True:
        time.sleep(0.01)  # do something that takes time
        yield  # measurement point

See bm_example.py for a functioning example.