Unknown Engine error

[pullaraomaddu]

Hi,

I am trying to run the flownet demos. I was able to compile the Caffe successfully. After compiling Caffe, I tried to run the demo application inside models/flownet. It is giving "Layer conv1 has unknown engine" error. Any help on what I am doing wrong? Below are tail of the log for "./demo_flownet.py S data/0000000-img0.ppm data/0000000-img1.ppm"

I0526 09:29:06.400105 29264 net.cpp:161] Top shape: 1 6 384 512 (1179648)
I0526 09:29:06.400121 29264 net.cpp:169] Memory required for data: 18874368
I0526 09:29:06.400136 29264 layer_factory.hpp:76] Creating layer conv1
F0526 09:29:06.400158 29264 layer_factory.cpp:37] Layer conv1 has unknown engine.
*** Check failure stack trace: ***
@ 0x7f3c6ae0fbcd google::LogMessage::Fail()
@ 0x7f3c6ae11b48 google::LogMessage::SendToLog()
@ 0x7f3c6ae0f763 google::LogMessage::Flush()
@ 0x7f3c6ae1247e google::LogMessageFatal::~LogMessageFatal()
@ 0x7f3c6b4316ce caffe::GetConvolutionLayer<>()
@ 0x7f3c6b44169c caffe::LayerRegistry<>::CreateLayer()
@ 0x7f3c6b4473d5 caffe::Net<>::Init()
@ 0x7f3c6b449c33 caffe::Net<>::Net()
@ 0x406d35 test()
@ 0x405831 main
@ 0x7f3c6a122ec5 __libc_start_main
@ 0x405f41 (unknown)

The resulting FLOW is stored in flownets-pred-NNNNNNN.flo

Appreciate your response.

Thanks,
Pullarao

[pullaraomaddu]

The error is gone after changing "engine: CUDNN" to "engine: CAFFE" in prototxt files. Thanks @waldol1 comments in the post BVLC/caffe#2682