Get the latest tech news

Metal-benchmarks: Apple GPU microarchitecture


Apple GPU microarchitecture. Contribute to philipturner/metal-benchmarks development by creating an account on GitHub.

This document enables evidence-based reasoning about performance on the Apple GPU, helping people diagnose bottlenecks in real-world software. To find accurate latencies, benchmarks issue 250x the amount of work needed to fully occupy a core's register file. Even if a background task is taking several frames, a high-priority UI command can quickly seize half the GPU cores.

Get the Android app

Or read this on Hacker News

Read more on:

Photo of benchmarks

benchmarks

Photo of Apple GPU

Apple GPU

Related news:

News photo

Redox OS Unlocks Faster VM Performance, "Slightly Faster" Than Linux In Some Benchmarks

News photo

AmpereOne Performance Scaling From 32 To 192 Cores, Core-For-Core Benchmarks Against Ampere Altra Max

News photo

AmpereOne A192-32X Benchmarks: 192 Core ARM Server Performance & Power Efficiency