torch.compile

torch._dynamo PyTorch’s compiler frontend which can output subgraphs when programs aren’t traceable. 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 .

Topic	Replies	Views	Activity
About the torch.compile category torch.compile	0	476	January 9, 2023
Inductor CPP codegen for WebAssembly target torch._inductor	1	9	September 27, 2023
"Boxed" functions torch._dynamo	1	22	September 26, 2023
How to define an in-place custom op in dynamo? torch.compile	2	42	September 25, 2023
‘len’ is not supported in symbolic tracing by default FX (Functional Transformations)	6	1367	September 25, 2023
Does torch.compile support inter-op parallelism? torch.compile	0	23	September 24, 2023
Emit Triton IR/ LLVM IR from TorchInductor torch._inductor	0	24	September 21, 2023
Shape inference in FX Graphs torch._dynamo	0	39	September 19, 2023
NvFuser with torch.compile torch._dynamo	6	107	September 20, 2023
Recreating Issue for self-referential list in torch._dynamo torch._dynamo	1	36	September 18, 2023
Any / Oneof patterns for SubgraphMatcher FX (Functional Transformations)	0	37	September 15, 2023
Graph execution in Pytorch 2.x torch.compile	0	37	September 15, 2023
What's the difference between symbolic_trace and Tracer FX (Functional Transformations)	1	607	September 14, 2023
CUDA 11.8 package version issues and CUDA 11.7 version issue torch.compile	1	54	September 13, 2023
The difference between torch.jit.script and torch.compile? torch.compile	1	42	September 12, 2023
Elementwise_kernel low efficiency when input size is small? torch.compile	0	39	September 12, 2023
`torch.compile` a vmapped ensemble torch.compile	0	34	September 11, 2023
Bug (?): aten ir nodes have wrong overload op torch.compile	0	25	September 8, 2023
Attempting to draw dynamo diagrams as a noobee torch.compile	0	35	September 7, 2023
Model only converges when wrapped in `torch.compile` torch.compile	9	70	September 6, 2023
Cannot call sizes() on tensor with symbolic sizes/strides torch.compile	2	76	September 6, 2023
Issues related to PyTorch Installation from Scratch torch.compile	6	66	September 5, 2023
Torch.compile does not work with importlib torch.compile	0	55	September 4, 2023
How to get intermediate results of ops in torchdynamo? torch.compile	0	48	September 2, 2023
Need help building Pytorch from source torch.compile	27	153	September 1, 2023
What's recommend to serialize a compiled torch model torch.compile	2	88	September 1, 2023
I build a decompiler to convert bytecode generated by dynamo into readable source code! torch.compile	3	90	August 29, 2023
Creating a GraphModule from FX Graph FX (Functional Transformations)	0	120	August 29, 2023
Error: static assertion failed: We only support non-string scalars in RepeatedField torch.compile	1	101	August 29, 2023
Inductor CPU C++ backend torch._inductor	0	63	August 28, 2023