Avoiding Discrete NVIDIA GPU

[Both__Michael]

Hi,

I installed OpenVino on Ubuntu 18.04 using the Apt method. My system has a discrete NVIDIA GPU and an Intel® Core™ i7-7700K CPU @ 4.20GHz × 8

I ran the CPU examples as suggested and they ran successfully:

> source /opt/intel/openvino/bin/setupvars.sh
> ./demo_squeezenet_download_convert_run.sh
> ./demo_security_barrier_camera.sh
> ./demo_benchmark_app.sh

When I run manually, it still works fine:

> ./classification_sample_async -i /opt/intel/openvino/deployment_tools/demo/car.png -m /home/michael/openvino_examples/models/FP16/squeezenet1.1.xml -d CPU
. . .
classid probability label
------- ----------- -----
817     0.8364178   sports car, sport car
511     0.0945683   convertible
479     0.0419195   car wheel
751     0.0091233   racer, race car, racing car
436     0.0068038   beach wagon, station wagon, wagon, estate car, beach waggon, station waggon, waggon
656     0.0037315   minivan
586     0.0025940   half track
717     0.0016044   pickup, pickup truck
864     0.0012045   tow truck, tow car, wrecker
581     0.0005833   grille, radiator grille
. . .

If I try to run on the GPU, it fails:

> ./classification_sample_async -i /opt/intel/openvino/deployment_tools/demo/car.png -m /home/michael/openvino_examples/models/FP16/squeezenet1.1.xml -d GPU
[ INFO ] InferenceEngine:
    API version ............ 2.1
    Build .................. custom_releases/2019/R3_ac8584cb714a697a12f1f30b7a3b78a5b9ac5e05
    Description ....... API
[ INFO ] Parsing input parameters
[ INFO ] Parsing input parameters
[ INFO ] Files were added: 1
[ INFO ]     /opt/intel/openvino/deployment_tools/demo/car.png
[ INFO ] Creating Inference Engine
[ ERROR ] Failed to create plugin /opt/intel/openvino_2019.3.376/deployment_tools/inference_engine/lib/intel64/libclDNNPlugin.so for device GPU
Please, check your environment
No OpenCL device found which would match provided configuration:
    GeForce GTX 1080 Ti: invalid vendor type

To verify there is an Intel GPU, I did this:

lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation HD Graphics 630 (rev 04)
01:00.0 VGA compatible controller: NVIDIA Corporation GP102 [GeForce GTX 1080 Ti] (rev a1)

I also verified the OpenCL DNN shared object is installed:

ls -l /opt/intel/openvino_2019.3.376/deployment_tools/inference_engine/lib/intel64
total 59812
-rw-r--r-- 1 root root  5654861 Oct 21 13:22 cache.json
drwxr-xr-x 2 root root     4096 Dec  4 12:11 cldnn_global_custom_kernels
-rw-r--r-- 1 root root  7802408 Oct 21 13:22 libclDNNPlugin.so
-rw-r--r-- 1 root root  3267072 Oct 21 13:22 libcpu_extension_avx2.so
-rw-r--r-- 1 root root  3262632 Oct 21 13:22 libcpu_extension_avx512.so
-rw-r--r-- 1 root root  3184352 Oct 21 13:22 libcpu_extension_sse4.so
-rw-r--r-- 1 root root  1074424 Oct 21 13:22 libGNAPlugin.so
-rw-r--r-- 1 root root   464912 Oct 21 13:22 libHeteroPlugin.so
-rw-r--r-- 1 root root  4966520 Oct 21 13:22 libinference_engine.so
-rw-r--r-- 1 root root 20792952 Oct 21 13:22 libMKLDNNPlugin.so
-rw-r--r-- 1 root root   260104 Oct 21 13:22 libMultiDevicePlugin.so
-rw-r--r-- 1 root root  4449688 Oct 21 13:22 libmyriadPlugin.so
-rw-r--r-- 1 root root  1847476 Oct 21 13:22 MvNCAPI-ma2450.mvcmd
-rw-r--r-- 1 root root  2033816 Oct 21 13:22 MvNCAPI-ma2x8x.mvcmd
-rw-r--r-- 1 root root  1833728 Oct 21 13:22 MvNCAPI-mv0262.mvcmd
-rwxr-xr-x 1 root root   227496 Oct 21 13:22 myriad_compile
-rwxr-xr-x 1 root root    80048 Oct 21 13:22 myriad_perfcheck
-rw-r--r-- 1 root root      654 Oct 21 13:22 plugins.xml
drwxr-xr-x 2 root root     4096 Dec  4 12:11 vpu_custom_kernels

If I ran the example with just the -h, it shows it can only see the CPU:

> ./classification_sample_async -h
[ INFO ] InferenceEngine: 
	API version ............ 2.1
	Build .................. custom_releases/2019/R3_ac8584cb714a697a12f1f30b7a3b78a5b9ac5e05
	Description ....... API
[ INFO ] Parsing input parameters
. . .
Available target devices:  CPU

How do I get the above example to run on the Intel GPU, and stop it from trying to run on the NVIDIA GPU?

Thanks for you help,

Michael

[Luis_at_Intel · Accepted solution]

Hi Michael,

I was able to see the same issue, a couple of things I can suggest that worked for me:

1. Make sure you have Internal/Integrated Graphics enabled in the BIOS. For my system once in BIOS menu, the setting was under Chipset > Internal Graphics > [Enabled].
2. Follow required steps to enable the toolkit components to use the processor graphics (GPU ). Find the steps for Linux here.

I was able to see the same error you reported for Linux, after performing these steps I can run the OpenVINO* demo scripts just fine using the GPU plugin. I didn't have to remove the discrete graphics at all, both discrete and integrated are running. I hope this works for you and let us know if it does.

Note for users on Windows: for those running into the same issue on Windows* 10, for Step #2 above follow these steps instead. FYI I had to update my graphics driver using Device Manager > Display adapters > Intel(R) HD Graphics > Search automatically for updated driver software.

Best Regards,

Luis