TorchInductor Update 4: CPU backend started to show promising performance boost

jgong5 · November 25, 2022, 7:53am

Test configuration for table 1-3: TorchDynamo commit: 2e6737a46d6c4eea5b888345c1ffc7a282eb8153, HW: Intel(R) Xeon(R) Platinum 8375C CPU @ 2.90GHz
Test configuration for table 4-6: Tested with PyTorch commit: b843f4db0a26aae6536e6b971f73bcc5af21c90a + #89109, #89209, HW: Intel(R) Xeon(R) Platinum 8375C CPU @ 2.90GHz

Topic		Replies	Views
TorchInductor Update 5: CPU backend backend performance update and deep dive on key optimizations compiler	0	3334	March 9, 2023
TorchInductor Update 6: CPU backend performance update and new features in PyTorch 2.1 compiler	0	2031	September 22, 2023
TorchInductor: a PyTorch-native Compiler with Define-by-Run IR and Symbolic Shapes compiler	46	69033	July 29, 2024
TorchInductor Update 9: Harden Vectorization Support and Enhance Loop Optimizations in TorchInductor CPP Backend compiler	0	542	September 4, 2024
TorchInductor Update 3: E2E model training with TorchDynamo + Inductor gets 1.67x/2.1x speedup compiler	3	2925	October 11, 2022