Error when using torch.compile on Jetson Orin NX

too_ym86 · June 24, 2024, 7:55am

The following code

import torch

def test(a, b):
    return a * b

a = torch.randn(3,3,3).to("cuda")
b = torch.tensor([[[1,-1,-1]]], device=a.device, dtype=a.dtype)
test_opt = torch.compile(test)
test_opt(a, b)

throws an error.

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 454, in _fn
    return fn(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 904, in catch_errors
    return callback(frame, cache_entry, hooks, frame_state, skip=1)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 769, in _convert_frame
    result = inner_convert(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 398, in _convert_frame_assert
    return _compile(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 669, in _compile
    guarded_code = compile_inner(code, one_graph, hooks, transform)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 542, in compile_inner
    out_code = transform_code_object(code, transform)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/bytecode_transformation.py", line 1033, in transform_code_object
    transformations(instructions, code_options)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 163, in _fn
    return fn(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 507, in transform
    tracer.run()
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 2122, in run
    super().run()
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 785, in run
    and self.step()
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 748, in step
    getattr(self, inst.opname)(inst)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 2241, in RETURN_VALUE
    self.output.compile_subgraph(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 931, in compile_subgraph
    self.compile_and_call_fx_graph(tx, list(reversed(stack_values)), root)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1102, in compile_and_call_fx_graph
    compiled_fn = self.call_user_compiler(gm)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1175, in call_user_compiler
    raise BackendCompilerFailed(self.compiler_fn, e).with_traceback(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1156, in call_user_compiler
    compiled_fn = compiler_fn(gm, self.example_inputs())
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/repro/after_dynamo.py", line 117, in debug_wrapper
    compiled_gm = compiler_fn(gm, example_inputs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/__init__.py", line 1730, in __call__
    return compile_fx(model_, inputs_, config_patches=self.config)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1321, in compile_fx
    return aot_autograd(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/backends/common.py", line 57, in compiler_fn
    cg = aot_module_simplified(gm, example_inputs, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 891, in aot_module_simplified
    compiled_fn = create_aot_dispatcher_function(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 604, in create_aot_dispatcher_function
    compiled_fn = compiler_fn(flat_fn, fake_flat_args, aot_config, fw_metadata=fw_metadata)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/runtime_wrappers.py", line 434, in aot_wrapper_dedupe
    return compiler_fn(flat_fn, leaf_flat_args, aot_config, fw_metadata=fw_metadata)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/runtime_wrappers.py", line 639, in aot_wrapper_synthetic_base
    return compiler_fn(flat_fn, flat_args, aot_config, fw_metadata=fw_metadata)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/jit_compile_runtime_wrappers.py", line 97, in aot_dispatch_base
    compiled_fw = compiler(fw_module, updated_flat_args)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1249, in fw_compiler_base
    return inner_compile(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/repro/after_aot.py", line 83, in debug_wrapper
    inner_compiled_fn = compiler_fn(gm, example_inputs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/debug.py", line 304, in inner
    return fn(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 423, in compile_fx_inner
    compiled_graph = fx_codegen_and_compile(
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 689, in fx_codegen_and_compile
    compiled_fn = graph.compile_to_fn()
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/graph.py", line 1227, in compile_to_fn
    return self.compile_to_module().call
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/graph.py", line 1175, in compile_to_module
    self.codegen_with_cpp_wrapper() if self.cpp_wrapper else self.codegen()
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/graph.py", line 1150, in codegen
    self.scheduler = Scheduler(self.buffers)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 1217, in __init__
    self.nodes = [self.create_scheduler_node(n) for n in nodes]
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 1217, in <listcomp>
    self.nodes = [self.create_scheduler_node(n) for n in nodes]
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 1308, in create_scheduler_node
    group_fn = self.get_backend(node.get_device()).group_fn
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 2225, in get_backend
    self.backends[device] = self.create_backend(device)
  File "/home/arl/miniconda3/envs/torch/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 2217, in create_backend
    raise RuntimeError(
torch._dynamo.exc.BackendCompilerFailed: backend='inductor' raised:
RuntimeError: Cannot find a working triton installation. More information on installing Triton can be found at https://github.com/openai/triton

Set TORCH_LOGS="+dynamo" and TORCHDYNAMO_VERBOSE=1 for more information

You can suppress this exception and fall back to eager by setting:
    import torch._dynamo
    torch._dynamo.config.suppress_errors = True

Package version:

(torch) arl@ubuntu:~$ conda list
# packages in environment at /home/arl/miniconda3/envs/torch:
#
# Name                    Version                   Build  Channel
_libgcc_mutex             0.1                        main  
_openmp_mutex             5.1                      51_gnu  
_sysroot_linux-aarch64_curr_repodata_hack 4                   h57d6b7b_14    conda-forge
binutils_impl_linux-aarch64 2.38                 h0c9fd12_1  
bzip2                     1.0.8                h998d150_6  
ca-certificates           2024.6.2             hcefe29a_0    conda-forge
filelock                  3.15.4                   pypi_0    pypi
fsspec                    2024.6.0                 pypi_0    pypi
gcc                       12.1.0              h75b0f6e_10    conda-forge
gcc_impl_linux-aarch64    12.1.0              h9c21524_17    conda-forge
jinja2                    3.1.4                    pypi_0    pypi
kernel-headers_linux-aarch64 4.18.0              h5b4a56d_14    conda-forge
ld_impl_linux-aarch64     2.38                 h8131f2d_1  
libffi                    3.4.4                h419075a_1  
libgcc-devel_linux-aarch64 12.1.0              hf2ffb8d_17    conda-forge
libgcc-ng                 13.2.0              he277a41_13    conda-forge
libgomp                   13.2.0              he277a41_13    conda-forge
libsanitizer              12.1.0              hd01590b_17    conda-forge
libstdcxx-ng              13.2.0              h3f4de04_13    conda-forge
libuuid                   1.41.5               h998d150_0  
markupsafe                2.1.5                    pypi_0    pypi
mpmath                    1.3.0                    pypi_0    pypi
ncurses                   6.4                  h419075a_0  
networkx                  3.3                      pypi_0    pypi
numpy                     1.26.4                   pypi_0    pypi
openssl                   3.3.1                h68df207_0    conda-forge
pip                       24.0            py310hd43f75c_0  
python                    3.10.14              h4bb2201_1  
readline                  8.2                  h998d150_0  
setuptools                69.5.1          py310hd43f75c_0  
sqlite                    3.45.3               h998d150_0  
sympy                     1.12.1                   pypi_0    pypi
sysroot_linux-aarch64     2.17                h5b4a56d_14    conda-forge
tk                        8.6.14               h987d8db_0  
torch                     2.3.0a0+40ec155e58.nv24.3          pypi_0    pypi
typing-extensions         4.12.2                   pypi_0    pypi
tzdata                    2024a                h04d1e81_0  
wheel                     0.43.0          py310hd43f75c_0  
xz                        5.4.6                h998d150_1  
zlib                      1.2.13               h998d150_1

JetPack version:

Package: nvidia-jetpack
Source: nvidia-jetpack (6.0)
Version: 6.0+b106
Architecture: arm64
Maintainer: NVIDIA Corporation

This works fine on x86-64 machines with the same PyTorch version.

AastaLLL · June 25, 2024, 5:44am

Hi,

Could you help to check if the same error occurs outside of the virtual environment?

Thanks.

too_ym86 · June 25, 2024, 9:19am

The same error occurs outside of the virtual environment:

>>> test_opt(a, b)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 454, in _fn
    return fn(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 904, in catch_errors
    return callback(frame, cache_entry, hooks, frame_state, skip=1)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 769, in _convert_frame
    result = inner_convert(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 398, in _convert_frame_assert
    return _compile(
  File "/usr/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 669, in _compile
    guarded_code = compile_inner(code, one_graph, hooks, transform)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 542, in compile_inner
    out_code = transform_code_object(code, transform)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/bytecode_transformation.py", line 1033, in transform_code_object
    transformations(instructions, code_options)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 163, in _fn
    return fn(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 507, in transform
    tracer.run()
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 2122, in run
    super().run()
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 785, in run
    and self.step()
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 748, in step
    getattr(self, inst.opname)(inst)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 2241, in RETURN_VALUE
    self.output.compile_subgraph(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 931, in compile_subgraph
    self.compile_and_call_fx_graph(tx, list(reversed(stack_values)), root)
  File "/usr/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1102, in compile_and_call_fx_graph
    compiled_fn = self.call_user_compiler(gm)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1175, in call_user_compiler
    raise BackendCompilerFailed(self.compiler_fn, e).with_traceback(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1156, in call_user_compiler
    compiled_fn = compiler_fn(gm, self.example_inputs())
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/repro/after_dynamo.py", line 117, in debug_wrapper
    compiled_gm = compiler_fn(gm, example_inputs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/__init__.py", line 1730, in __call__
    return compile_fx(model_, inputs_, config_patches=self.config)
  File "/usr/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1321, in compile_fx
    return aot_autograd(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/backends/common.py", line 57, in compiler_fn
    cg = aot_module_simplified(gm, example_inputs, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 891, in aot_module_simplified
    compiled_fn = create_aot_dispatcher_function(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 604, in create_aot_dispatcher_function
    compiled_fn = compiler_fn(flat_fn, fake_flat_args, aot_config, fw_metadata=fw_metadata)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/runtime_wrappers.py", line 434, in aot_wrapper_dedupe
    return compiler_fn(flat_fn, leaf_flat_args, aot_config, fw_metadata=fw_metadata)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/runtime_wrappers.py", line 639, in aot_wrapper_synthetic_base
    return compiler_fn(flat_fn, flat_args, aot_config, fw_metadata=fw_metadata)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/jit_compile_runtime_wrappers.py", line 97, in aot_dispatch_base
    compiled_fw = compiler(fw_module, updated_flat_args)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1249, in fw_compiler_base
    return inner_compile(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/repro/after_aot.py", line 83, in debug_wrapper
    inner_compiled_fn = compiler_fn(gm, example_inputs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/debug.py", line 304, in inner
    return fn(*args, **kwargs)
  File "/usr/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/usr/lib/python3.10/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 423, in compile_fx_inner
    compiled_graph = fx_codegen_and_compile(
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 689, in fx_codegen_and_compile
    compiled_fn = graph.compile_to_fn()
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/graph.py", line 1227, in compile_to_fn
    return self.compile_to_module().call
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/graph.py", line 1175, in compile_to_module
    self.codegen_with_cpp_wrapper() if self.cpp_wrapper else self.codegen()
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/graph.py", line 1150, in codegen
    self.scheduler = Scheduler(self.buffers)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_dynamo/utils.py", line 249, in time_wrapper
    r = func(*args, **kwargs)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 1217, in __init__
    self.nodes = [self.create_scheduler_node(n) for n in nodes]
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 1217, in <listcomp>
    self.nodes = [self.create_scheduler_node(n) for n in nodes]
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 1308, in create_scheduler_node
    group_fn = self.get_backend(node.get_device()).group_fn
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 2225, in get_backend
    self.backends[device] = self.create_backend(device)
  File "/home/arl/.local/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 2217, in create_backend
    raise RuntimeError(
torch._dynamo.exc.BackendCompilerFailed: backend='inductor' raised:
RuntimeError: Cannot find a working triton installation. More information on installing Triton can be found at https://github.com/openai/triton

Set TORCH_LOGS="+dynamo" and TORCHDYNAMO_VERBOSE=1 for more information


You can suppress this exception and fall back to eager by setting:
    import torch._dynamo
    torch._dynamo.config.suppress_errors = True

too_ym86 · June 25, 2024, 9:26am

FYI, it works fine on CPU.

import torch

def test(a, b):
    return a * b

a = torch.randn(3,3,3)
b = torch.tensor([[[1,-1,-1]]], device=a.device, dtype=a.dtype)
test_opt = torch.compile(test)
test_opt(a, b)

AastaLLL · June 26, 2024, 7:00am

Hi,

Confirm that we can reproduce the same error on JetPack 6.0 + Pytorch v2.4.0.
But if setting the below command, the test_opt(a, b) can work as expected.

Does this work for you?

>>> torch._dynamo.config.suppress_errors = True
>>> test_opt(a, b)
W0626 02:48:29.236000 281473232717504 torch/_dynamo/convert_frame.py:843] WON'T CONVERT test <stdin> line 1 
...
W0626 02:48:29.236000 281473232717504 torch/_dynamo/convert_frame.py:843] Set TORCH_LOGS="+dynamo" and TORCHDYNAMO_VERBOSE=1 for more information
W0626 02:48:29.236000 281473232717504 torch/_dynamo/convert_frame.py:843] 
tensor([[[ 0.4441,  1.5841,  0.3711],
         [ 1.5443,  0.6918, -0.0790],
         [-0.3570, -0.1881,  0.0323]],

        [[ 2.1000,  0.8786,  0.5348],
         [ 0.7633,  1.4262, -2.7466],
         [-0.2026,  0.0186,  1.2799]],

        [[-0.5236,  0.4396,  1.6877],
         [ 1.1878,  0.7496, -0.6581],
         [ 0.7863,  0.7889,  0.4728]]], device='cuda:0')
>>>

Thanks.

too_ym86 · June 26, 2024, 10:54am

Thanks. It works (with a lot of warnings) after adding the setting.

>>> torch._dynamo.config.suppress_errors = True
>>> torch.equal(test_opt(a, b), test(a, b))
True

What is this setting and how does this affect performance?

AastaLLL · June 27, 2024, 9:57am

Hi,

Please check below document:

https://pytorch.org/docs/stable/torch.compiler_deepdive.html

The setting seems to change the JIT compiling back to the traditional way.
Thanks.