Fatal error: cuda/cuda_config.h about flownet2-tf HOT 5 OPEN

@JohnnieXDU If you installed your tensorflow r1.3 from source, you could copy cuda_config.h to your python dist-packages,

sudo cp tensorflow/third_party/toolchains/gpus/cuda/cuda/cuda_config.h /usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/stream_executor/cuda/

then you may still have the following compilation issues:

/usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/core/util/cuda_kernel_helper.h(620): error: identifier "__shfl" is undefined
/usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/core/util/cuda_kernel_helper.h(640): error: identifier "__shfl_up" is undefined
/usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/core/util/cuda_kernel_helper.h(660): error: identifier "__shfl_down" is undefined
/usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/core/util/cuda_kernel_helper.h(680): error: identifier "__shfl_xor" is undefined

The solution is to add the following compiler flags back to the Makefile
GPUCFLAGS = -c -gencode=arch=compute_30,code=sm_30

This worked on my machine.

from flownet2-tf.

you have to use tensorflow 1.2.1.

from flownet2-tf.

Thanks, finally I fix this by using TF 1.2.1 @MatthewD1993 @cubeye

from flownet2-tf.

@cubeye Thank you very much~~

from flownet2-tf.

thanks!

from flownet2-tf.