Make TRITON_INTERPRET=1 work with inductor generated kernels #123956
Description
🚀 The feature, motivation and pitch
If you try to run TRITON_INTERPRET=1 with inductor generated kernels you'll get an exception:
File "/data/users/eellison/pytorch/torch/_inductor/triton_heuristics.py", line 365, in _precompile_config
binary = triton.compile(*compile_args, **compile_kwargs)
File "/home/eellison/.conda/envs/pytorch-3.10/lib/python3.10/site-packages/triton/compiler/compiler.py", line 231, in compile
key = f"{triton_key()}-{src.hash()}-{backend.hash()}-{options.hash()}-{str(sorted(get_env_vars().items()))}"
File "/home/eellison/.conda/envs/pytorch-3.10/lib/python3.10/site-packages/triton/compiler/compiler.py", line 106, in hash
key = f"{self.fn.cache_key}-{self.attrs.hash()}-{sorted_sig}-{sorted_constants}"
It would be nice for debugging to add compatibility.
Alternatives
No response
Additional context
No response
cc @ezyang @msaroufim @bdhirsh @anijain2305 @chauhang @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @aakhundov @ColinPeppler @amjames @desertfire