You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I try the Instant-NGP backbone colab, but when I install the cuda extensions, I got these errors:
Processing ./raymarching
Preparing metadata (setup.py) ... done
Building wheels for collected packages: raymarching
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for raymarching (setup.py) ... error
ERROR: Failed building wheel for raymarching
Running setup.py clean for raymarching
Failed to build raymarching
ERROR: Could not build wheels for raymarching, which is required to install pyproject.toml-based projects
Processing ./shencoder
Preparing metadata (setup.py) ... done
Building wheels for collected packages: shencoder
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for shencoder (setup.py) ... error
ERROR: Failed building wheel for shencoder
Running setup.py clean for shencoder
Failed to build shencoder
ERROR: Could not build wheels for shencoder, which is required to install pyproject.toml-based projects
Processing ./freqencoder
Preparing metadata (setup.py) ... done
Building wheels for collected packages: freqencoder
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for freqencoder (setup.py) ... error
ERROR: Failed building wheel for freqencoder
Running setup.py clean for freqencoder
Failed to build freqencoder
ERROR: Could not build wheels for freqencoder, which is required to install pyproject.toml-based projects
Processing ./gridencoder
Preparing metadata (setup.py) ... done
Building wheels for collected packages: gridencoder
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for gridencoder (setup.py) ... error
ERROR: Failed building wheel for gridencoder
Running setup.py clean for gridencoder
Failed to build gridencoder
ERROR: Could not build wheels for gridencoder, which is required to install pyproject.toml-based projects
Here is the nvidia-smi:
Thu Mar 14 05:26:11 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.104.05 Driver Version: 535.104.05 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Tesla T4 Off | 00000000:00:04.0 Off | 0 |
| N/A 36C P8 9W / 70W | 0MiB / 15360MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+
@kwin120 I think it's because you are trying to use CUDA 12. It doesn't work for me as well. And I don't know if there any plans to support CUDA 12 here.
I am grappling with this as well. I am trying to figure out what's wrong with the most recent PyTorch torch-2.4.0.dev20240330+cu121 . If anyone has any tips please share.
Description
I try the Instant-NGP backbone colab, but when I install the cuda extensions, I got these errors:
Processing ./raymarching
Preparing metadata (setup.py) ... done
Building wheels for collected packages: raymarching
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for raymarching (setup.py) ... error
ERROR: Failed building wheel for raymarching
Running setup.py clean for raymarching
Failed to build raymarching
ERROR: Could not build wheels for raymarching, which is required to install pyproject.toml-based projects
Processing ./shencoder
Preparing metadata (setup.py) ... done
Building wheels for collected packages: shencoder
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for shencoder (setup.py) ... error
ERROR: Failed building wheel for shencoder
Running setup.py clean for shencoder
Failed to build shencoder
ERROR: Could not build wheels for shencoder, which is required to install pyproject.toml-based projects
Processing ./freqencoder
Preparing metadata (setup.py) ... done
Building wheels for collected packages: freqencoder
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for freqencoder (setup.py) ... error
ERROR: Failed building wheel for freqencoder
Running setup.py clean for freqencoder
Failed to build freqencoder
ERROR: Could not build wheels for freqencoder, which is required to install pyproject.toml-based projects
Processing ./gridencoder
Preparing metadata (setup.py) ... done
Building wheels for collected packages: gridencoder
error: subprocess-exited-with-error
× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for gridencoder (setup.py) ... error
ERROR: Failed building wheel for gridencoder
Running setup.py clean for gridencoder
Failed to build gridencoder
ERROR: Could not build wheels for gridencoder, which is required to install pyproject.toml-based projects
Here is the nvidia-smi:
Thu Mar 14 05:26:11 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.104.05 Driver Version: 535.104.05 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Tesla T4 Off | 00000000:00:04.0 Off | 0 |
| N/A 36C P8 9W / 70W | 0MiB / 15360MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+
Steps to Reproduce
Expected Behavior
successfully install cuda extensions
Environment
google colab, T4 GPU, python 3
The text was updated successfully, but these errors were encountered: