Zhihao Jia·Jun 19, 2025Compiling LLMs into a MegaKernel: A Path to Low-Latency InferenceTL;DR: We developed a compiler that automatically transforms LLM inference into a single megakernel — a fused GPU kernel that performs…A response icon3A response icon3
Zhihao Jia·Sep 29, 2024Generating Fast GPU Kernels without Programming in CUDA/TritonAs GPUs grow more sophisticated, achieving optimal performance for modern AI applications such as LLMs and various GenAI tasks depends…A response icon1A response icon1