OpenCV - Caffe - Object Detection - i3 NUC(8th Gen) - High CPU Util

[Mounagurusamy__Guruv]

Case 1: OpenCV - Caffe - Object Detection - i3 NUC(8th Gen) - High CPU Util

I am building an object detection solution using CCTV Cameras, using intel i3 NUC(8th Gen).

I have installed the following libraries: https://software.intel.com/content/www/us/en/develop/articles/intel-parallel-studio-xe-runtime-by-version.html

Also, I have installed Intel Neo OpenCL and I am calling the following to execute using OpenCL:

net.setPreferableBackend(cv.dnn.DNN_BACKEND_OPENCV)
net.setPreferableTarget(cv.dnn.DNN_TARGET_OPENCL)

But, I encounter the following warning while executing the code(with 76% CPU util for one CCTV Camera)

[ WARN:0] global /io/opencv/modules/dnn/src/dnn.cpp (1218) setUpNet DNN: OpenCL target is not supported with current OpenCL device (tested with GPUs only), switching to CPU.

I have also installed Beignet OpenCL

clinfo Details:

ioz@ioz-NUC8i3BEH:~/Desktop/CodeWrks/cam01$ clinfo
Number of platforms 1
Platform Name Intel Gen OCL Driver
Platform Vendor Intel
Platform Version OpenCL 2.0 beignet 1.3
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_byte_addressable_store cl_khr_3d_image_writes cl_khr_image2d_from_buffer cl_khr_depth_images cl_khr_spir cl_khr_icd cl_intel_accelerator cl_intel_subgroups cl_intel_subgroups_short cl_khr_gl_sharing
Platform Extensions function suffix Intel
beignet-opencl-icd: no supported GPU found, this is probably the wrong opencl-icd package for this hardware
(If you have multiple ICDs installed and OpenCL works, you can ignore this message)

Platform Name Intel Gen OCL Driver
Number of devices 0

NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) Intel Gen OCL Driver
beignet-opencl-icd: no supported GPU found, this is probably the wrong opencl-icd package for this hardware
(If you have multiple ICDs installed and OpenCL works, you can ignore this message)
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) Intel Gen OCL Driver
clCreateContext(NULL, ...) [default] No devices found in platform
beignet-opencl-icd: no supported GPU found, this is probably the wrong opencl-icd package for this hardware
(If you have multiple ICDs installed and OpenCL works, you can ignore this message)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
beignet-opencl-icd: no supported GPU found, this is probably the wrong opencl-icd package for this hardware
(If you have multiple ICDs installed and OpenCL works, you can ignore this message)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
beignet-opencl-icd: no supported GPU found, this is probably the wrong opencl-icd package for this hardware
(If you have multiple ICDs installed and OpenCL works, you can ignore this message)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) No devices found in platform

ICD loader properties
ICD loader Name OpenCL ICD Loader
ICD loader Vendor OCL Icd free software
ICD loader Version 2.2.11
ICD loader Profile OpenCL 2.1

Can you please help me understand, why do I encounter high CPU despite using intel OpenCL, Do i need to do anything else to bring the CPU Util low?

Case 2: OpenCV-Caffe - Object Detection using CCTV Camera - GNA Plugin - Intel Pentium Silver Processor

On contrary to the above mentioned case, I was able to execute the object detection code for five CCTV cameras in Intel Pentium Silver(with GNA Plugin installed) and Average CPU util for five cameras is 60%

I have also installed Beignet OpenCL

Is GNA Plugin makes the difference in CPU utilisation?

Note: I am not using the OpenVINO object detection code in both the systems

Can you please help me resolve this issue.

Thanks

Guru

[Luis_at_Intel]

Hi Mounagurusamy__Guruv,

Thanks for reaching out. Based on your issue description the problem seems to rely on "Case 1: OpenCV - Caffe - Object Detection - i3 NUC(8th Gen) - High CPU Util", as your Case 2 with OpenVINO™ Toolkit seems to be working fine. Is my understanding correct? If so, I may have to redirect this thread to OpenCL* forum where you will receive guidance.

Based on the clinfo details "no supported GPU found, this is probably the wrong opencl-icd package for this hardware" it looks like is not detecting a GPU supported by opencl-icd, and its falling back to CPU. Do you have any additional details about the i3 NUC(8th Gen) NUC/processor? SKU number, for example NUC8i3BEH or Intel® Core™ i3 8109U processor.

Best Regards,

Luis

[Mounagurusamy__Guruv]

Hi Luis,

1) For Case 2 - It's the Intel Silver Pentium J5005 processor and I have installed the OpenVINO GNA plugin, but I am not explicitly using OpenVINO for Code execution (I am using Caffe Object Detection in Python) - https://www.pyimagesearch.com/2017/09/18/real-time-object-detection-with-deep-learning-and-opencv/

But I have added the code to call Open CL (Beignet).

For Case 1 - I followed the same as Case 2, but GNA plugin couldn't be installed, but I have installed Beignet / Intel Neo.

Looks like both didn't work.

NUC Details - NUC8i3BEH, Intel® Core™ i3 8109U processor.

Thanks
Guru

[Mounagurusamy__Guruv]

Hi intel team,

Can you please let me know, if GNA Plugin is the reason for the performance optimization in J5005 Pentium Silver processor?, I am not using OpenVINO Models or Code.

I am using the following code, cv.dnn.DNN_TARGET_OPENCL or cv.dnn.DNN_TARGET_CPU

I am using the following Code:

https://www.pyimagesearch.com/2017/09/18/real-time-object-detection-with-deep-learning-and-opencv/

I am running the above code for five CCTV cameras, CPU Util is approx 60% on J5005 Pentium Silver.

If GNA Plugin Driver is the reason for the CPU performance optimization, Can you please confirm?

Thanks
Guru