Getting Error in installing vllm on Nvidia Jetson AGX ORIN

Getting Error in installing vllm on Nvidia Jetson AGX ORIN

I have tried installation using pip , source git source and building setup.py also but could not succeed.

Below are the details about my setup -

  • Operating System: Linux - 5.15.136-tegra - aarch64

  • Compiler: /usr/bin/c++

  • Compiler Version: 11.4.0 (Ubuntu 11.4.0-1ubuntu1~22.04)

With pip -

(bot_venv) lftds@ubuntu:~/Documents/copilot$ pip install vllm

Collecting vllm

Using cached vllm-0.5.1.tar.gz (790 kB)

Installing build dependencies … done

Getting requirements to build wheel … error

error: subprocess-exited-with-error

× Getting requirements to build wheel did not run successfully.

│ exit code: 1

╰─> [16 lines of output]

Traceback (most recent call last):

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py”, line 353, in

main()

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py”, line 335, in main

json_out[‘return_val’] = hook(**hook_input[‘kwargs’])

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py”, line 118, in get_requires_for_build_wheel

return hook(config_settings)

File “/tmp/pip-build-env-wsi_ghvs/overlay/lib/python3.10/site-packages/setuptools/build_meta.py”, line 327, in get_requires_for_build_wheel

return self._get_build_requires(config_settings, requirements=)

File “/tmp/pip-build-env-wsi_ghvs/overlay/lib/python3.10/site-packages/setuptools/build_meta.py”, line 297, in _get_build_requires

self.run_setup()

File “/tmp/pip-build-env-wsi_ghvs/overlay/lib/python3.10/site-packages/setuptools/build_meta.py”, line 313, in run_setup

exec(code, locals())

File “”, line 432, in

File “”, line 353, in get_vllm_version

RuntimeError: Unknown runtime environment

[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.

error: subprocess-exited-with-error

× Getting requirements to build 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.

From git source -

bot_venv) lftds@ubuntu:~/Documents/copilot$ git clone GitHub - vllm-project/vllm: A high-throughput and memory-efficient inference and serving engine for LLMs

Cloning into ‘vllm’…

remote: Enumerating objects: 22588, done.

remote: Total 22588 (delta 0), reused 0 (delta 0), pack-reused 22588

Receiving objects: 100% (22588/22588), 21.67 MiB | 8.61 MiB/s, done.

Resolving deltas: 100% (16741/16741), done.

(bot_venv) lftds@ubuntu:~/Documents/copilot$ cd vllm/

(bot_venv) lftds@ubuntu:~/Documents/copilot/vllm$ pip install .

Processing /home/lftds/Documents/copilot/vllm

Installing build dependencies … done

Getting requirements to build wheel … error

error: subprocess-exited-with-error

× Getting requirements to build wheel did not run successfully.

│ exit code: 1

╰─> [16 lines of output]

Traceback (most recent call last):

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py”, line 353, in

main()

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py”, line 335, in main

json_out[‘return_val’] = hook(**hook_input[‘kwargs’])

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py”, line 118, in get_requires_for_build_wheel

return hook(config_settings)

File “/tmp/pip-build-env-x2sad19s/overlay/lib/python3.10/site-packages/setuptools/build_meta.py”, line 327, in get_requires_for_build_wheel

return self._get_build_requires(config_settings, requirements=)

File “/tmp/pip-build-env-x2sad19s/overlay/lib/python3.10/site-packages/setuptools/build_meta.py”, line 297, in _get_build_requires

self.run_setup()

File “/tmp/pip-build-env-x2sad19s/overlay/lib/python3.10/site-packages/setuptools/build_meta.py”, line 313, in run_setup

exec(code, locals())

File “”, line 432, in

File “”, line 353, in get_vllm_version

RuntimeError: Unknown runtime environment

[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.

error: subprocess-exited-with-error

× Getting requirements to build 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 from source.py -

(bot_venv) lftds@ubuntu:~/Documents/copilot/vllm$ python setup.py bdist_wheel

running bdist_wheel

running build

running build_py

creating build

creating build/lib.linux-aarch64-cpython-310

creating build/lib.linux-aarch64-cpython-310/vllm

copying vllm/tracing.py → build/lib.linux-aarch64-cpython-310/vllm

copying vllm/model_executor/layers/fused_moe/configs/E=8,N=7168,device_name=AMD_Instinct_MI300X.json → build/lib.linux-aarch64-cpython-310/vllm/model_executor/layers/fused_moe/configs

copying vllm/model_executor/layers/fused_moe/configs/E=8,N=14336,device_name=AMD_Instinct_MI300X.json → build/lib.linux-aarch64-cpython-310/vllm/model_executor/layers/fused_moe/configs

running build_ext

– The CXX compiler identification is GNU 11.4.0

– Detecting CXX compiler ABI info

– Detecting CXX compiler ABI info - done

– Check for working CXX compiler: /usr/bin/c++ - skipped

– Detecting CXX compile features

– Detecting CXX compile features - done

– Build type: RelWithDebInfo

– Target device: cuda

– Could NOT find Python (missing: Python_INCLUDE_DIRS Interpreter Development.Module Development.SABIModule) (found version “3.10.12”)

CMake Error at cmake/utils.cmake:10 (message):

Unable to find python matching:

/home/lftds/Documents/copilot/bot_venv/bin/python.

Call Stack (most recent call first):

CMakeLists.txt:43 (find_python_from_executable)

– Configuring incomplete, errors occurred!

See also “/home/lftds/Documents/copilot/vllm/build/temp.linux-aarch64-cpython-310/CMakeFiles/CMakeOutput.log”.

Traceback (most recent call last):

File “/home/lftds/Documents/copilot/vllm/setup.py”, line 430, in

setup(

File “/home/lftds/Documents/copilot/bot_venv/lib/python3.10/site-packages/setuptools/init.py”, line 103, in setup

return distutils.core.setup(**attrs) .

.

.

.

.

.

.

.

.

.

.

File “/home/lftds/Documents/copilot/vllm/setup.py”, line 175, in configure

subprocess.check_call(

File “/usr/lib/python3.10/subprocess.py”, line 369, in check_call

raise CalledProcessError(retcode, cmd)

subprocess.CalledProcessError: Command ‘[‘cmake’, ‘/home/lftds/Documents/copilot/vllm’, ‘-DCMAKE_BUILD_TYPE=RelWithDebInfo’, ‘-DCMAKE_LIBRARY_OUTPUT_DIRECTORY=/home/lftds/Documents/copilot/vllm/build/lib.linux-aarch64-cpython-310/vllm’, ‘-DCMAKE_ARCHIVE_OUTPUT_DIRECTORY=build/temp.linux-aarch64-cpython-310’, ‘-DVLLM_TARGET_DEVICE=cuda’, ‘-DVLLM_PYTHON_EXECUTABLE=/home/lftds/Documents/copilot/bot_venv/bin/python’, ‘-DNVCC_THREADS=1’]’ returned non-zero exit status 1.

(bot_venv) lftds@ubuntu:~/Documents/copilot/vllm$ cmake -DPYTHON_EXECUTABLE=/home/lftds/Documents/copilot/bot_venv/bin/python …

CMake Error: The source directory “/home/lftds/Documents/copilot” does not appear to contain CMakeLists.txt.

Specify --help for usage, or press the help button on the CMake GUI.

(bot_venv) lftds@ubuntu:~/Documents/copilot/vllm$ cd /home/lftds/Documents/copilot/vllm

mkdir -p build

cd build

cmake -DPYTHON_EXECUTABLE=/home/lftds/Documents/copilot/bot_venv/bin/python -DPYTHON_INCLUDE_DIR=/usr/include/python3.10 -DPYTHON_LIBRARY=/usr/lib/x86_64-linux-gnu/libpython3.10.so …

– The CXX compiler identification is GNU 11.4.0

– Detecting CXX compiler ABI info

– Detecting CXX compiler ABI info - done

– Check for working CXX compiler: /usr/bin/c++ - skipped

– Detecting CXX compile features

– Detecting CXX compile features - done

– Build type:

– Target device: cuda

CMake Error at CMakeLists.txt:45 (message):

Please set VLLM_PYTHON_EXECUTABLE to the path of the desired python version

before running cmake configure.

– Configuring incomplete, errors occurred!

See also “/home/lftds/Documents/copilot/vllm/build/CMakeFiles/CMakeOutput.log”.

Hi,

Based on their guide, they seem to only support desktop Linux environments.

https://docs.vllm.ai/en/latest/getting_started/installation.html

Could you check with the vllm team to see if their library can run on the Jetson platform?

Thanks.

@mausam.jain I’ve not had luck in the past with getting vLLM to compile on Jetson, and would recommend trying other LLM inferencing libraries that we have working on Jetson and pre-built container images for, like MLC, AWQ, llama.cpp, and exllama (all of which support quantization)

It would seem unlikely at this point of another library being faster than MLC, as it is already close to peak theoretical performance on Orin, and I support it and AWQ as preferred backends in the NanoLLM library.

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