cutil_inline.h

README.md

The original great cuda-convnet package is here: http://code.google.com/p/cuda-convnet/

To build the convnet using CUDA 5:

1. Change the build.sh to use correct pathes. For cuda SDK path, use the path contains file helper_cuda.h.
2. Make a dummyinclude folder, and put the cutil_inline.h into that folder.
3. For MAKEFILE, include the dummyinclude file.

Try ./build.sh and see whether that can compiles.

The functionality is not fully tested and this can just compiles the package.

cutil_inline.h

/* 
   Dummy Includes for cutil_inline.h
   Wei LI <kuantkid <at> gmail <dot> com>
 */
#include <helper_cuda.h>

#define cutilCheckMsg(a) getLastCudaError(a)
#define cutGetMaxGflopsDeviceId() gpuGetMaxGflopsDeviceId()

#define MIN(a,b) (a) < (b) ? (a) : (b)

Makefile

# Change the include line in Makefile
INCLUDES := -DCUDA_5 -I$(PYTHON_INCLUDE_PATH) -I$(NUMPY_INCLUDE_PATH) -I$(CUDA_SDK_PATH) -I./include -I./include/common -I./include/cudaconv2 -I./include/nvmatrix -I./dummyinclude/

This is great, but can anyone verify that it is in fact functional? I am unable to recreate the sample specified here: https://code.google.com/p/cuda-convnet/wiki/TrainingNet with no error output...

did you get it installed with cuda 6 on osx?