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	893	January 9, 2023
Getting Triton to generate all kernels torch._inductor	3	682	May 1, 2024
Questions about compiling and exporting a model to TensorRT torch.compile	0	15	April 30, 2024
Unable to Utilize GPUs with PyTorch on p2.xlarge EC2 Instances torch.compile	5	100	April 30, 2024
Does torch.compile has garbage collection mechanism for intermediate tensors inside a compiled subgraph? torch.compile	0	24	April 29, 2024
Torch._dynamo.run vs torch.compile torch.compile	3	50	April 27, 2024
Compile PyTorch code to LLVM IR with Torch 2.0 torch.compile	0	31	April 27, 2024
Tracing PyTorch compiled code torch.compile	0	33	April 27, 2024
Torch.compile generated output files understanding torch.compile	1	48	April 26, 2024
Generate Triton kernels for CPU torch._inductor	4	110	April 26, 2024
Serving Torch_TersorRT converted model with Triton torch._dynamo	0	34	April 25, 2024
Slow convolutions in triton torch._dynamo	0	35	April 24, 2024
Le: "b8[64, 2048, 7, 7]" = torch.ops.aten.le.Scalar(relu_48, 0); relu_48 = None torch._dynamo	1	80	April 24, 2024
Export Multiple Functions of a Pytorch Module torch.compile	8	303	April 24, 2024
Any specific reason for chosing the default batch sizes for the torch dynamo benchmark suite torch.compile	1	46	April 22, 2024
Can toch.cond be used recursively? torch.compile	2	59	April 22, 2024
Error when backpropagating through a compiled PyTorch module multiple times torch.compile	1	48	April 21, 2024
Is possible export the model on cpu and run it gpu? torch.compile	4	83	April 17, 2024
Compile and deepcopy in the context of early stopping torch.compile	0	53	April 16, 2024
Obscure error message when trying to export a compiled function torch.compile	0	46	April 15, 2024
Torch.compile when /home is a read only filesystem torch.compile	11	259	April 14, 2024
Torch.compile cache_size_limit best practice torch._dynamo	1	73	April 12, 2024
Export and save problem（torch-tensorrt） torch._dynamo	0	55	April 11, 2024
Compile function that uses constant tensor torch.compile	1	97	April 10, 2024
How to achieve Torch-MLIR compatibility with Dynamo backend? torch._dynamo	1	85	April 9, 2024
Is it possible to define a custom back-end wrapped with aot_autograd which does not use fake tensors? torch.compile	0	47	April 9, 2024
The difference between `torch.ops.aten.addmm.default` and `torch.ops.aten.linear.default` torch._dynamo	0	60	April 8, 2024
[dynamo] [onnx] Models exported with torch.onnx.dynamo_export show worse performance during inference torch._dynamo	0	80	April 8, 2024
Return handle from compiled function from C++ to be used by backend torch.compile	1	261	April 4, 2024
Unique node name across graph re-compilations and graph breaks torch._dynamo	0	75	April 3, 2024