GPU not supported Windows DLS 2.0.6

Hello
I install the new version of DLS (V2.0.6) but the GPU support is not working. I test it with a GTX 1050 and a GTX 1060. I get a segmentation fault (it seems that python crashed).
My os is windows 10.

Thanks in advance

Stephan

Hi Stephan,

Did you complete the CUDA 9 installation during the Deep Learning Studio installation?

In any case, can you reinstall CUDA 9 and the driver using the following link ?

https://developer.nvidia.com/compute/cuda/9.0/Prod/network_installers/cuda_9.0.176_win10_network-exe

Also, do you have another python installation in the system PATH ? Could you try by removing your existing python from system PATH and restart the Deep Learning Studio Manager.

Regards
Rajendra

1 Like

Yes I have another python in the path and it was the problem.
Thanks for the reply

We have fixed the PATH issue in 2.0.7 version which is now available on our website.

Hello. I installed new version 2.0.7 with CUDA but still got “GPU not supported”. No Python variables in Path and both 8 and 9 CUDA installed. Any idea why GPU not working?

Can you try running below commands from command prompt and provide us the output

cd <Deep Learning Studio install folder>
.\conda3\python .\home\app\gpucheck.py

Thank you for the quick reply.

Result:
Using TensorFlow backend.
Not supported
No module named ‘tensorflow’

It would be better if you modify .\home\app\start.sh file and replace following line:

python -u /home/app/gpucheck.py > gpucheck.log 2>&1

with:

python -u /home/app/gpucheck.py

After the above change, you can start deep learning studio and send us the log output

Thanks.

Log:

Starting Deep Learning Studio…
Checking GPU support…
Using MXNet backend.
[11:12:04] C:\Users\prakash\Desktop\incubator-mxnet\dmlc-core\include\dmlc/logging.h:308: [11:12:04] C:\Users\prakash\Desktop\incubator-mxnet\src\storage\storage.cc:109: Check failed: e == cudaSuccess || e == cudaErrorCudartUnloading CUDA: CUDA driver version is insufficient for CUDA runtime version
Checking GPU support…Not supported
simple_bind error. Arguments:
Input_1: (10, 1)
[11:12:04] C:\Users\prakash\Desktop\incubator-mxnet\src\storage\storage.cc:109: Check failed: e == cudaSuccess || e == cudaErrorCudartUnloading CUDA: CUDA driver version is insufficient for CUDA runtime version
GPU not supported
skipping cifar dataset download!
skipping IMDB dataset download!
skipping MNIST dataset download!
skipping reuters dataset download!
No changes detected
Using MXNet backend.
D:\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\account\templatetags\account_tags.py:4: DeprecationWarning: {% load account_tags %} is deprecated, use {% load account %}
DeprecationWarning)
D:\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\socialaccount\templatetags\socialaccount_tags.py:4: DeprecationWarning: {% load socialaccount_tags %} is deprecated, use {% load socialaccount %}
" {% load socialaccount %}", DeprecationWarning)
Operations to perform:
Apply all migrations: account, admin, auth, authtoken, automl, contenttypes, environments, project, projects, reversion, sessions, sites, socialaccount
Running migrations:
No migrations to apply.
Using MXNet backend.
D:\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\account\templatetags\account_tags.py:4: DeprecationWarning: {% load account_tags %} is deprecated, use {% load account %}
DeprecationWarning)
D:\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\socialaccount\templatetags\socialaccount_tags.py:4: DeprecationWarning: {% load socialaccount_tags %} is deprecated, use {% load socialaccount %}
" {% load socialaccount %}", DeprecationWarning)
Using MXNet backend.
Using MXNet backend.

Seems like there is version mismatch between CUDA 9.0 and the current driver installed.

Is it possible for you to uninstall existing NVIDIA driver and any existing CUDA installation and rerun the CUDA 9.0 installer?

Cuda 9.0 installer is present in the <Deep Learning Studio Install location>.

Thanks. I tried but it doesn’t work unfortunately

Starting Deep Learning Studio…
Checking GPU support…
Using MXNet backend.
[07:53:14] [07:53:14] C:\Users\prakash\Desktop\incubator-mxnet\dmlc-core\include\dmlc/logging.h:308C:\Users\prakash\Desktop\incubator-mxnet\dmlc-core\include\dmlc/logging.h:308: [07:53:14] c:\users\prakash\desktop\incubator-mxnet\mshadow\mshadow./cuda/tensor_gpu-inl.cuh:110: Check failed: err == cudaSuccess (48 vs. 0) Name: MapPlanKernel ErrStr:no kernel image is available for execution on the device
: [07:53:14] c:\users\prakash\desktop\incubator-mxnet\mshadow\mshadow./cuda/tensor_gpu-inl.cuh:110: Check failed: err == cudaSuccess (48 vs. 0) Name: MapPlanKernel ErrStr:no kernel image is available for execution on the device
[07:53:14] C:\Users\prakash\Desktop\incubator-mxnet\dmlc-core\include\dmlc/logging.h[07:53:14] C:\Users\prakash\Desktop\incubator-mxnet\dmlc-core\include\dmlc/logging.h:308: [07:53:14] c:\users\prakash\desktop\incubator-mxnet\src\engine./threaded_engine.h:370: [07:53:14] c:\users\prakash\desktop\incubator-mxnet\mshadow\mshadow./cuda/tensor_gpu-inl.cuh:110: Check failed: err == cudaSuccess (48 vs. 0) Name: MapPlanKernel ErrStr:no kernel image is available for execution on the device
A fatal error occurred in asynchronous engine operation. If you do not know what caused this error, you can try set environment variable MXNET_ENGINE_TYPE to NaiveEngine and run with debugger (i.e. gdb). This will force all operations to be synchronous and backtrace will give you the series of calls that lead to this error. Remember to set MXNET_ENGINE_TYPE back to empty after debugging.
:308: [07:53:14] c:\users\prakash\desktop\incubator-mxnet\src\engine./threaded_engine.h:370: [07:53:14] c:\users\prakash\desktop\incubator-mxnet\mshadow\mshadow./cuda/tensor_gpu-inl.cuh:110: Check failed: err == cudaSuccess (48 vs. 0) Name: MapPlanKernel ErrStr:no kernel image is available for execution on the device
A fatal error occurred in asynchronous engine operation. If you do not know what caused this error, you can try set environment variable MXNET_ENGINE_TYPE to NaiveEngine and run with debugger (i.e. gdb). This will force all operations to be synchronous and backtrace will give you the series of calls that lead to this error. Remember to set MXNET_ENGINE_TYPE back to empty after debugging.
GPU not supported
skipping cifar dataset download!
skipping IMDB dataset download!
skipping MNIST dataset download!
skipping reuters dataset download!

Can you download following file?

https://s3-us-west-2.amazonaws.com/deep-learning-studio-manager/downloads/mxnet.zip

It has mxnet.dll and opencv.dll. Copy these dlls to ‘\mxnet’ folder.

After copying, restart deep learning studio.

Thank you very much. It works