torch.compile

torch._dynamo PyTorch’s compiler frontend which can output subgraphs when programs aren’t traceable. 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._inductor The default compiler in torch.compile, builds loops schedules and codegens triton or C++ kernels.

Topic	Replies	Views	Activity
About the torch.compile category torch.compile	0	1170	January 9, 2023
How does pytorch triton work with dynamic shapes? torch._inductor	0	4	March 31, 2025
Torch.compile() causes RuntimeError: element 0 of tensors does not require torch.compile	1	10	March 31, 2025
Error: accessing tensor output of CUDAGraphs that has been overwritten by a subsequent run torch.compile	0	10	March 30, 2025
Need help debugging torch.compile issue torch._inductor	0	20	March 26, 2025
Dynamic slicing torch.compile	5	46	March 24, 2025
Best way to detect and fix graphs breaks torch.compile	5	59	March 24, 2025
Is it possible for torch SDPA to be slower than manual attention? torch.compile	5	17	March 24, 2025
Torch.compile dynamic shapes in non-functional style torch.compile	1	15	March 23, 2025
SYCL: feature test compile failed! torch.compile	0	19	March 23, 2025
torch.cuda.is_available()=False;torch._C._cuda_getDeviceCount() > 0 torch.compile	0	25	March 20, 2025
Shapes of FakeTensors after graph transformations on ATen IR torch.compile	6	53	March 19, 2025
Torch.compile(mode="max-autotune") produces different inference result from eager mode — is this expected? torch.compile	6	51	March 19, 2025
Triton CodeGen for torch linear operator torch.compile	0	4	March 19, 2025
How to solve the graph break happen in torch.compile torch.compile	11	257	March 18, 2025
The selectable configs for irregular shape when using triton.autotune torch.compile	0	9	March 18, 2025
Torch.compile CudaGraph creation & downstream systems torch.compile	2	23	March 18, 2025
Debugging slow TorchDynamo Cache Lookup torch.compile	1	29	March 18, 2025
How to ensure that auto-generated Triton Kernel is executed torch.compile	0	12	March 13, 2025
How to let torch.compile only target GPU torch.compile	0	24	March 13, 2025
Flex attention raises unnecessary assertion error torch._inductor	1	40	March 12, 2025
`cuda.h' missing during torch.compile in new environment torch.compile	3	176	March 7, 2025
Why is inductor re-recording my function with every call torch.compile	3	43	March 6, 2025
PyTorch v.2.3.1 unable to configure torch.compile	1	209	March 4, 2025
Torch._check failing with torch compile torch.compile	1	63	March 1, 2025
How to make TorchDynamoc decompose torch.nn.functional.scaled_dot_product_attention? torch._dynamo	1	38	February 27, 2025
Mapping between TorchDynamo compiled functions and FX GraphModules torch._dynamo	2	39	February 27, 2025
PyTorch 2.5.1 -fstack-protector-strong does not result in stack-protected libpytorch.so torch.compile	1	41	February 25, 2025
Ways to pass compilation target device to dynamo backend torch.compile	2	383	February 20, 2025
Converting model that has dynamic control flow to load in C++ torch.compile	0	19	February 19, 2025