TorchInductor: a PyTorch-native Compiler with Define-by-Run IR and Symbolic Shapes

void-main December 27, 2022, 3:54am 21

Hi @jansel , I wonder why inductor chooses Triton to generate CUDA kernels instead of other solutions like TVM / XLA?

Topic		Replies	Views
TorchDynamo Update 4: LazyTensor & nvFuser Experiments compiler	4	4786	February 9, 2024
TorchDynamo Update 3: GPU Inference Edition compiler	12	6631	February 2, 2023
TorchDynamo: An Experiment in Dynamic Python Bytecode Transformation compiler	7	17060	March 9, 2023
TorchInductor Update 6: CPU backend performance update and new features in PyTorch 2.1 compiler	0	1966	September 22, 2023
TorchInductor Update 7: key optimizations with CPU backend in PyTorch 2.2 release compiler	4	927	March 8, 2024