events
GPU Detective Stories: PyTorch Recompiles & Hidden Bottlenecks
An online session from the AI Performance Engineering Meetup (UK) with two technical talks on hidden costs in model serving and training. Chaim Rand compares six ways to handle PyTorch recompilation…

Auf Englisch
- Starts
- 2026-11-16 17:00 UTC
- Organizer
- AI Performance Engineering Meetup (UK)
- Location
- Online event
An online session from the AI Performance Engineering Meetup (UK) with two technical talks on hidden costs in model serving and training. Chaim Rand compares six ways to handle PyTorch recompilation when input shapes or layouts change during inference, from cache preloading and bounded recompilation with eager fallback to ahead-of-time compilation via torch.export. Israel Ogbole of Zymtrace walks through a case study on profiling a model called Jev, describing how continuous CPU/GPU profiling exposed compute, memory, communication and synchronisation bottlenecks.
Event website