torch.compile

torch._inductor The default compiler in torch.compile, builds loops schedules and codegens triton or C++ kernels. FX (Functional Transformations) FX is a toolkit for developers to use to transform nn.Module instances. FX consists of three main components: a symbolic tracer, an intermediate representation , and Python code generation . torch._dynamo PyTorch’s compiler frontend which can output subgraphs when programs aren’t traceable.

Topic	Replies	Views	Activity
About the torch.compile category torch.compile	0	1110	January 9, 2023
Does dynamo trigger real kernel execution? torch._dynamo	2	15	December 19, 2024
Compiling a model that uses sdp_kernel to enable the backends does not work torch._dynamo	4	26	December 18, 2024
Forward function not being compiled by default torch.compile	1	25	December 17, 2024
Onnx export with dynamo using torch.cond for dynamic models torch._dynamo	0	11	December 16, 2024
How to add custom operators to export_for_training? torch.compile	1	16	December 12, 2024
Sharing torch compile kernels between layers torch.compile	1	13	December 11, 2024
Is there a danger when doing DDP that different processes compiling kernels will overwrite each other? torch.compile	0	8	December 10, 2024
Saving modified fx graph torch.compile	0	21	December 9, 2024
Bug with isin assume_unique=True torch.compile	1	9	December 9, 2024
Scatter add much slower when compiled torch.compile	1	24	December 6, 2024
How to disable ___check_obj_id guards on dict? torch._dynamo	0	12	December 5, 2024
Not seeing any training time speedups when using torch.compile torch.compile	1	27	December 5, 2024
A node type in export IR graph FX (Functional Transformations)	1	15	November 28, 2024
Scaled_dot_product_attention higher head num cost much more memory torch.compile	1	12	November 28, 2024
CUDA memory allocation for result tensor torch.compile	0	13	November 26, 2024
Compile and vmap in custom op with quantile torch.compile	0	21	November 25, 2024
Compiling vmapped custom op torch.compile	5	38	November 25, 2024
Closures are being gc'd and causing failures to compile torch.compile	1	26	November 24, 2024
Why does the inductor reduction Triton Codegen use the Welford algorithm instead of the Naive? torch.compile	1	22	November 20, 2024
Image_process.postprocess slow after torch.compile torch.compile	0	25	November 18, 2024
Compiling a method other than forward torch.compile	2	31	November 19, 2024
Error module torchvision in CUDA 11.4 torch.compile	2	26	November 19, 2024
Increased memory footprint with custom kernel and all reduce torch.compile	2	39	November 18, 2024
The forward graphs captured by torch.export and aot_export_module are different torch._dynamo	2	41	November 17, 2024
Dynamic slicing torch.export torch._dynamo	2	42	November 16, 2024
Discrepancies Between Compiled and Non-Compiled Models with Convolutional Layers in PyTorch torch.compile	1	42	November 16, 2024
Any chance to preserve some ops while decomposing PT2E model? FX (Functional Transformations)	1	21	November 16, 2024
Multiple compiled versions of the same model torch.compile	2	44	November 16, 2024
Torch.compile - what is the best scope of compilation? torch.compile	7	2324	November 16, 2024