jupiter notebook file, RuntimeError: Error while calling cudaGetDevice(&the_device_id)

Question

jupiter notebook file, RuntimeError: Error while calling cudaGetDevice(&the_device_id)

Shavkat Shoniyozov 0

Hello folks, i kept running into this issue when i run my python script on jupiter notebook in ml studio. "RuntimeError: Error while calling cudaGetDevice(&the_device_id) in file /tmp/pip-install-d074pk05/dlib_f1c0e8e51fe740e1849b08eccc100dcf/dlib/cuda/gpu_data.cpp:204. code: 100" I bought computing instance cpu and could not get gpu one. im trying to use face-detection library in my code. i installed required libraries successfully, but at the end, it kept saying the same error. the script code run well on my local environment.

Thanks in advance ~~~

romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-08-11T09:55:19.2333333+00:00

@Shavkat Shoniyozov Does the local environment support GPU? I think for a CPU based compute instance if the script's libraries are still using a GPU then this is an expected error. Although the installation of the libraries is successful the actual runtime call might still need a GPU, I think this might be causing the error.
Shavkat Shoniyozov 0 Reputation points

2023-08-11T13:07:32.11+00:00

Thank for your reply @romungi-MSFT . I dont think gpu is being used in local environment. i changed the code to run only on cpu but it kept repeating the same error again and again ㅠㅠ
romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-08-13T13:57:53.6666667+00:00

@Shavkat Shoniyozov If the script is only using the CPU, this should not be the case. How did you install the libraries on the compute instance through the jupyter notebook? Try to install using the following format and check again under a new kernel session?

%pip install <library>

Your answer

romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-08-11T09:55:19.2333333+00:00

@Shavkat Shoniyozov Does the local environment support GPU? I think for a CPU based compute instance if the script's libraries are still using a GPU then this is an expected error. Although the installation of the libraries is successful the actual runtime call might still need a GPU, I think this might be causing the error.
Shavkat Shoniyozov 0 Reputation points

2023-08-11T13:07:32.11+00:00

Thank for your reply @romungi-MSFT . I dont think gpu is being used in local environment. i changed the code to run only on cpu but it kept repeating the same error again and again ㅠㅠ
romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-08-13T13:57:53.6666667+00:00

@Shavkat Shoniyozov If the script is only using the CPU, this should not be the case. How did you install the libraries on the compute instance through the jupyter notebook? Try to install using the following format and check again under a new kernel session?

%pip install <library>

Share via

jupiter notebook file, RuntimeError: Error while calling cudaGetDevice(&the_device_id)

Your answer