Installing Pip Wheels for CUDA 12.0 fails

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
1

I followed the instructions from https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=aarch64-jetson&Compilation=Native&Distribution=Ubuntu&target_version=20.04&target_type=deb_local to install CUDA 12.0 on my Jetson AGX Orin. It went fine. The next step was to install Python modules following instructions at CUDA Installation Guide for Linux. It failed. This step:

python3 -m pip install --upgrade setuptools pip wheel

Produces an error:

nick@orin:~$ python3 -m pip install --upgrade setuptools pip wheel
Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com
/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl/urllib3/connectionpool.py:999: InsecureRequestWarning: Unverified HTTPS request is being made to host 'pypi.ngc.nvidia.com'. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl/urllib3/connectionpool.py:999: InsecureRequestWarning: Unverified HTTPS request is being made to host 'pypi.ngc.nvidia.com'. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
Collecting setuptools
  Downloading setuptools-65.6.3-py3-none-any.whl (1.2 MB)
     |████████████████████████████████| 1.2 MB 1.0 MB/s 
/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl/urllib3/connectionpool.py:999: InsecureRequestWarning: Unverified HTTPS request is being made to host 'pypi.ngc.nvidia.com'. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl/urllib3/connectionpool.py:999: InsecureRequestWarning: Unverified HTTPS request is being made to host 'pypi.ngc.nvidia.com'. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
Collecting pip
  Downloading pip-22.3.1-py3-none-any.whl (2.1 MB)
     |████████████████████████████████| 2.1 MB 4.4 MB/s 
/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl/urllib3/connectionpool.py:999: InsecureRequestWarning: Unverified HTTPS request is being made to host 'pypi.ngc.nvidia.com'. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl/urllib3/connectionpool.py:999: InsecureRequestWarning: Unverified HTTPS request is being made to host 'pypi.ngc.nvidia.com'. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings
Collecting wheel
  Downloading wheel-0.38.4-py3-none-any.whl (36 kB)
ERROR: launchpadlib 1.10.13 requires testresources, which is not installed.
Installing collected packages: setuptools, pip, wheel
  WARNING: The scripts pip, pip3, pip3.10 and pip3.8 are installed in '/home/nick/.local/bin' which is not on PATH.
  Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.
  WARNING: The script wheel is installed in '/home/nick/.local/bin' which is not on PATH.
  Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.
Successfully installed pip-22.3.1 setuptools-65.6.3 wheel-0.38.4

I installed missing testresources. The next steps failed completely:

nick@orin:~$ python3 -m pip install nvidia-pyindex
Defaulting to user installation because normal site-packages is not writeable
Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com
Collecting nvidia-pyindex
  Downloading nvidia-pyindex-1.0.9.tar.gz (10 kB)
  Preparing metadata (setup.py) ... done
Building wheels for collected packages: nvidia-pyindex
  Building wheel for nvidia-pyindex (setup.py) ... done
  Created wheel for nvidia-pyindex: filename=nvidia_pyindex-1.0.9-py3-none-any.whl size=8418 sha256=7a4b9795dfe2295a78e2f056de7f86e2b9f78de79e23ac5fcd178da9623a7647
  Stored in directory: /tmp/pip-ephem-wheel-cache-jnlv9z1v/wheels/11/a2/1a/2caaf9725a10ed91ab89274afabbfa70ab3db002f93600feae
Successfully built nvidia-pyindex
Installing collected packages: nvidia-pyindex
  WARNING: The script nvidia_pyindex is installed in '/home/nick/.local/bin' which is not on PATH.
  Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.
Successfully installed nvidia-pyindex-1.0.9
nick@orin:~$ python3 -m pip install nvidia-cuda-runtime-cu12
Defaulting to user installation because normal site-packages is not writeable
Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com
Collecting nvidia-cuda-runtime-cu12
  Downloading nvidia-cuda-runtime-cu12-0.0.1.dev5.tar.gz (4.3 kB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error
  
  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [17 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/tmp/pip-install-rdffxd6q/nvidia-cuda-runtime-cu12_cad2da2971e04a3ab98caed1f9a96920/setup.py", line 152, in <module>
          raise RuntimeError(open("ERROR.txt", "r").read())
      RuntimeError:
      ###########################################################################################
      The package you are trying to install is only a placeholder project on PyPI.org repository.
      This package is hosted on NVIDIA Python Package Index.
      
      This package can be installed as:
      ```
      $ pip install nvidia-pyindex
      $ pip install nvidia-cuda-runtime-cu12
      ```
      ###########################################################################################
      
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

Is there a solution to this problem? Preferably using Pip.

3

Hi,

Do you want to install CUDA Python on Orin?
Based on the document below, Jetson is not on the supported list.

https://nvidia.github.io/cuda-python/install.html#requirements

Thanks.

4

I have cuda-python 12.0.0 installed on Orin, and it seems to work fine. If you have a test, I can run it to verify.

5

Hi,

You can find some tests below:

https://nvidia.github.io/cuda-python/overview.html

Thanks.

6

I was wrong. It doesn’t work. Here is the top part of the example from https://nvidia.github.io/cuda-python/overview.html:

from cuda import cuda, nvrtc
import numpy as np

saxpy = """\
extern "C" __global__
void saxpy(float a, float *x, float *y, float *out, size_t n)
{
 size_t tid = blockIdx.x * blockDim.x + threadIdx.x;
 if (tid < n) {
   out[tid] = a * x[tid] + y[tid];
 }
}
"""

# Create program
err, prog = nvrtc.nvrtcCreateProgram(str.encode(saxpy), b"saxpy.cu", 0, [], [])
print(err)
# Compile program
opts = [b"--fmad=false", b"--gpu-architecture=compute_75"]
err, = nvrtc.nvrtcCompileProgram(prog, 2, opts)

# Get PTX from compilation
err, ptxSize = nvrtc.nvrtcGetPTXSize(prog)
ptx = b" " * ptxSize
err, = nvrtc.nvrtcGetPTX(prog, ptx)

# Initialize CUDA Driver API
err, = cuda.cuInit(0)

# Retrieve handle for device 0
err, cuDevice = cuda.cuDeviceGet(0)

# Create context
err, context = cuda.cuCtxCreate(0, cuDevice)
print(err)

# Load PTX as module data and retrieve function
ptx = np.char.array(ptx)
print(ptx[0][:500])

# Note: Incompatible --gpu-architecture would be detected here
err, module = cuda.cuModuleLoadData(ptx.ctypes.data)
print(err)
exit()

It produces:

nvrtcResult.NVRTC_SUCCESS
CUresult.CUDA_SUCCESS
b'//\n// Generated by NVIDIA NVVM Compiler\n//\n// Compiler Build ID: CL-31968024\n// Cuda compilation tools, release 12.0, V12.0.76\n// Based on NVVM 7.0.1\n//\n\n.version 8.0\n.target sm_75\n.address_size 64\n\n\t// .globl\tsaxpy\n.global .align 8 .u64 _ZN35_INTERNAL_00000000_8_saxpy_cu_saxpy2nv6target6detail9all_hostsE = 1;\n.global .align 8 .u64 _ZN35_INTERNAL_00000000_8_saxpy_cu_saxpy2nv6target6detail9sm_35_bitE = 2;\n.global .align 8 .u64 _ZN35_INTERNAL_00000000_8_saxpy_cu_saxpy2nv6target6detail9sm_37_bitE ='
CUresult.CUDA_ERROR_UNSUPPORTED_PTX_VERSION
7

Hi,

Does the CUDA C++ library work correctly?
It’s expected that the C++ library can work on Jetson.

For python, however, Jetson is not listed in the supported matrix yet.
Thanks.

9

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.