search
GPU architectures
Trends
- 1Developer uses iPhone as second GPU to speed up MacBook AI workloads●I made my iPhone a second GPU for my MacBook-Qwen 3.8 27B prefills 29–44% faster
A developer reports hooking an iPhone up to a MacBook as a secondary GPU, claiming that Qwen 3.8 27B, a large language model, prefills 29 to 44 percent faster with the phone attached. The experiment relies on Apple's unified memory architecture across devices, and tech enthusiasts are debating whether phones can realistically serve as practical compute companions for local AI.
Repos
- gatewai-dev/gitframes Code-first video, rendered natively on WebGPU.