WWDC22 · 25 min · Graphics & Games

Scale compute workloads across Apple GPUs

Discover how you can create compute workloads that scale efficiently across Apple GPUs. Learn how to saturate the GPU by improving your work distribution, minimize GPU timeline gaps with effective pipelining and concurrent dispatches, and use atomic operations effectively. We’ll also take you through the latest counters and tools in Xcode and Instruments that can help you optimize spatial and temporal memory access patterns.

Transcript ↓ .srt ↓ .txt all transcripts