# Backend Compiler Graph Break

**URL:** <https://discuss.pytorch.org/t/backend-compiler-graph-break/223754>\
**Category:** torch.\_dynamo\
**Created:** [October 30, 2025, 12:53am UTC](https://discuss.pytorch.org/t/backend-compiler-graph-break/223754 "2025-10-30T00:53:36Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![walrus\_23](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/walrus_23/32/78552_2.png) [@walrus\_23](https://discuss.pytorch.org/u/walrus_23)\
**Post date:** [October 30, 2025, 12:53am UTC](https://discuss.pytorch.org/t/backend-compiler-graph-break/223754/1 "2025-10-30T00:53:36Z")

</div>

See GitHub issue #166582 on the main PyTorch repo.

TLDR: I have an int8 matmul on a custom ASIC (in simulation) that I am compiling to replace all linear layers of any PyTorch model. However, the model is trained using QAT from TorchAO. When running the compiled model, it runs into.

W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Backend compiler exception  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Explanation: Backend compiler `_backend` failed with aten.\_local\_scalar\_dense.default  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1]  
W1029 18:26:40.182000 156731 torch/_dynamo/exc.py:593] [24/0\_1] While executing %dut\_matmul\_sync : [num\_users=1] = call\_function[target=torch\_backend.dut\_matmul\_sync](args = (tb, %dequantize\_per\_tensor\_default\_2, %dequantize\_per\_tensor\_default, %l\_self\_modules\_fc1\_parameters\_bias_), kwargs = {})  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Original traceback:  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] None  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Use tlparse to see full graph.. Adding a graph break.  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Hint: Report an issue to the backend compiler repo.  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1]  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Developer debug context: Backend: \_backend  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Exception:aten.\_local\_scalar\_dense.default  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1]  
W1029 18:26:40.182000 156731 torch/_dynamo/exc.py:593] [24/0\_1] While executing %dut\_matmul\_sync : [num\_users=1] = call\_function[target=torch\_backend.dut\_matmul\_sync](args = (tb, %dequantize\_per\_tensor\_default\_2, %dequantize\_per\_tensor\_default, %l\_self\_modules\_fc1\_parameters\_bias_), kwargs = {})  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Original traceback:  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] None  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Use tlparse to see full graph. ([https://github.com/pytorch/tlparse?tab=readme-ov-file#tlparse-parse-structured-pt2-logs](https://github.com/pytorch/tlparse?tab=readme-ov-file#tlparse-parse-structured-pt2-logs))  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] Traceback:  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] File “\<eval\_with\_key\>.464”, line 23, in forward  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] return pytree.tree\_unflatten((dequantize\_per\_tensor\_default\_4,), self.\_out\_spec)  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1]  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1]  
W1029 18:26:40.182000 156731 torch/\_dynamo/exc.py:593] [24/0\_1] For more details about this graph break, please visit: [https://meta-pytorch.github.io/compile-graph-break-site/gb/gb0219.html](https://meta-pytorch.github.io/compile-graph-break-site/gb/gb0219.html)