Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.
5527 Discussions

Copy net to GPU manually to speed up load network

Nemes__Adam
Beginner
186 Views

Hey

I have a relativle small network which loads to the GPU really slowly (more then 20 minutes! )

What can cause such behaviour?

Is there a way to speed this up? 

Maybe manually copying the network install give xmlpath or something like this?

 

Thanks 

Adam

0 Kudos
2 Replies
Rizal_Intel
Moderator
177 Views

Hi Adam,

 

Loading your input model in Intermediate Representation (IR) format to GPU takes much longer than loading the same model to a CPU because the GPU stack is based on OpenCL. The load time depends on the compilation time of OpenCL kernels.

 

You could try to reduce this by enabling cl_cache in the application's directory.

When you enable the cl_cache, the first time you load the model it will still take a long time because the OpenCL kernel will compile. However, each subsequent load of the same model will be much faster. Please ensure that you have write permission at the directory to write the cache.

 

Read the Optimization Guide GPU Checklist for more information on how to optimize a model for running on a GPU.

 

Regards,

Rizal


Rizal_Intel
Moderator
159 Views

Hi Adam,


Intel will no longer monitor this thread since we have provided a solution. If you need any additional information from Intel, please submit a new question.


Regards,

Rizal


Reply