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	1286	January 9, 2023
InductorError in autogenerated code in torchinductor_cache - triton, rocm torch._inductor	0	3	November 22, 2025
Torch.compile - getting `nans` with the latest version of torch torch.compile	1	14	November 21, 2025
Choice of torch.compile vs. triton torch.compile	8	8511	November 18, 2025
Any specific reason for chosing the default batch sizes for the torch dynamo benchmark suite torch.compile	3	187	November 18, 2025
Torch compile cache sharing across different GPUs torch.compile	0	24	November 17, 2025
Why torch==2.9 compile qwen3 model with block ptr will crash? torch._inductor	0	16	November 14, 2025
Performance bottleneck with autocast in AOT Inductor export: repeated weight casting due to disabled cache torch._inductor	0	20	November 11, 2025
Is it possible to combine compiled vectorized function without recompiling already compiled part? torch.compile	0	15	November 7, 2025
How do I know which placeholder op is an input? torch.compile	0	23	November 6, 2025
What requirements do I need to use Torch.compile? torch.compile	0	58	November 3, 2025
Torch.export + fx.GraphModule: Unable to lift constants (parameters/buffers) to placeholders for memory sharing across modules torch.compile	0	24	October 31, 2025
OSError: [Errno 36] File name too long torch.compile	3	51	October 31, 2025
Backend Compiler Graph Break torch._dynamo	0	22	October 30, 2025
About dynamo.disable torch._dynamo	0	33	October 29, 2025
PyTorch with CUDA support is not installed, although everything worked before that torch.compile	1	311	October 22, 2025
Compiling a loop on MPS - exceeding limit of buffers torch.compile	0	18	October 17, 2025
Why does custom passes before joint graph (joint_custom_pre_pass) executed twice? torch._inductor	0	17	October 16, 2025
About some issue about torch.median() torch.compile	1	28	October 15, 2025
Torch.compile [WARNING] not enough SMs to use max_autotune_gemm mode torch.compile	4	8543	October 14, 2025
Is it normal for compiled model outputs to have different tensor strides? torch.compile	3	55	October 14, 2025
TS2EPConverter and `dynamic_shapes` FX (Functional Transformations)	0	25	October 5, 2025
CUDA Graphs with FSDP 2 and flex attention? torch.compile	0	75	October 3, 2025
Module 'torch.library' has no attribute 'custom_op' torch.compile	6	1197	September 24, 2025
How many patterns does torch compile match and replace for now? torch._inductor	0	86	September 24, 2025
PyTorch Error: Compiler: cl is not found torch._inductor	4	1956	September 23, 2025
Torch.compile + flex_attention with custom score_mod fails (FlexibleLayout / NoValidChoicesError) torch.compile	0	58	September 19, 2025
Are there any related works on "megakernel"? torch.compile	0	45	September 17, 2025
Coordinate_descent_tuning errors out with torch.AcceleratorError: CUDA error: invalid argument torch.compile	10	174	September 15, 2025
In Dynamo+AOTAutograd, why run Faketensor through the code multiple times? torch._dynamo	0	23	September 15, 2025