#Simran Arora

LLMs Fail to Write Fast Multi-GPU Kernels

Simran Arora from Together AI discusses the challenges of multi-GPU kernel development and why current LLMs struggle to optimize them, despite ongoing research.