Open medhini opened 6 years ago
Fixed this by specifying paths using -I while running the nvcc command. Another error popped up which required a cuda_config.h file. Removing -D GOOGLE_CUDA=1 solved this.
Thanks @medhini - removing -D GOOGLE_CUDA=1
fixed this for me too.
@medhini can you explain more what you linked with the -l argument, please? Which paths did you specify?
Hi, I am trying to compile the op by myself and when I run nvcc -std=c++11 -c -o roi_pooling_op_gpu.cu.o roi_pooling_op_gpu.cu.cc -I \ $TF_INC -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC
I get the following error: In file included from roi_pooling_op_gpu.cu.cc:10:0: /home/medhini2/.local/lib/python2.7/site-packages/tensorflow/include/tensorflow/core/util/cuda_kernel_helper.h:24:31: fatal error: cuda/include/cuda.h: No such file or directory
include "cuda/include/cuda.h"
How do I go about fixing this?