Community
cancel
Showing results for 
Search instead for 
Did you mean: 
SC_Huang
Beginner
85 Views

super_resolution_demo with GPU/MYRIAD Failed

I've try the demo with error as below

200x200

with single-image-super-resolution-0063(FP32) GPU:

[ INFO ] Loading model to the plugin
[ ERROR ] network allocation failed: exceeded max size of memory object allocation

with single-image-super-resolution-0063(FP16) GPU:

[ INFO ] Loading model to the plugin
[ ERROR ] network allocation failed: /teamcity/work/scoring_engine_build/releases_2018_R5/thirdparty/clDNN/src/reshape.cpp at line: 64
Error has occured for: 352
Output layout count(=20480000) is not equal to: input layout count(=0)
Output layout of reshape pirmiti

with single-image-super-resolution-0063(FP16) MYRIAD:

[ INFO ] Loading model to the plugin
[usb_read:365] error=-1, total size is=36,leave size=36
E: [xLink] [         0] dispatcherEventReceive:304    dispatcherEventReceive() Read failed -1 | event 0x7f94d4ff8eb0 USB_CLOSE_STREAM_RESP

E: [xLink] [         0] eventReader:254    eventReader stopped
E: [xLink] [     52715] dispatcherWaitEventComplete:720    waiting is timeout, sending reset remote event

also similar problem in other model like single-image-super-resolution-1011,1021

so, what cause the error happened?

Is there any solution or it just not support plugin except mkldnn?

0 Kudos
2 Replies
nikos1
Valued Contributor I
85 Views

According to https://software.intel.com/en-us/openvino-toolkit/documentation/pretrained-models only CPU path ( -d CPU ) is supported for Super resolution samples ( Supported Samples Table : Reference this table for components that support the pretrained models. )

So yes, it looks only mkldnn for now as you said.

SC_Huang
Beginner
85 Views

nikos wrote:

According to https://software.intel.com/en-us/openvino-toolkit/documentation/pretrain... only CPU path ( -d CPU ) is supported for Super resolution samples ( Supported Samples Table : Reference this table for components that support the pretrained models. )

So yes, it looks only mkldnn for now as you said.

Hi Nikos,

Thanks for your kindly reply~

Reply