-
Notifications
You must be signed in to change notification settings - Fork 626
Description
🐛 Describe the bug
torchao build fails on macOS arm64 due to missing symbol c10::StorageImpl::throw_data_ptr_access_error()
🧩 Summary
When running install_requirements.sh
to install dependencies for Executorch on macOS (Apple Silicon), the installation fails while building torchao
. The failure is due to a missing symbol during linking:
Undefined symbols for architecture arm64:
"c10::StorageImpl::throw_data_ptr_access_error() const", referenced from:
void const* c10::TensorImpl::data_impl<void const, ...>(...) const in op_linear_8bit_act_xbit_weight_aten.cpp.o
void* c10::TensorImpl::data_impl<void, ...>(...) const in op_embedding_xbit_aten.cpp.o
ld: symbol(s) not found for architecture arm64
c++: error: linker command failed with exit code 1
...
ERROR: Failed to build installable wheels for some pyproject.toml based projects (torchao)
💻 Environment
- OS: macOS 15.2 (Apple Silicon / arm64)
- Python: 3.10.17 (via pyenv)
- CMake: 3.31.6
- PyTorch:
2.8.0.dev20250524
- TorchVision:
0.22.0.dev20250524
- TorchAudio:
2.6.0.dev20250524
- Installation method:
./install_requirements.sh
Versions
Collecting environment information...
PyTorch version: 2.8.0.dev20250524
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A
OS: macOS 15.5 (arm64)
GCC version: Could not collect
Clang version: 17.0.0 (clang-1700.0.13.5)
CMake version: version 3.31.6
Libc version: N/A
Python version: 3.10.17 (main, Apr 29 2025, 13:28:11) [Clang 17.0.0 (clang-1700.0.13.3)] (64-bit runtime)
Python platform: macOS-15.5-arm64-arm-64bit
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
CPU:
Apple M4 Pro
Versions of relevant libraries:
[pip3] numpy==2.2.6
[pip3] torch==2.8.0.dev20250524
[pip3] torchaudio==2.6.0.dev20250524
[pip3] torchdata==0.11.0
[pip3] torchsr==1.0.4
[pip3] torchtune==0.6.1
[pip3] torchvision==0.22.0.dev20250524
[conda] Could not collect