The administrator for the dept C870 recently downloaded CUDA v2.0. I’m trying to compile some CUDA code which uses the cutil.h functions. I went to the install dir/common to make (which I thought built cutil), but after compiling my source code I get three errors,

undefined reference to cutCheckCmdLineFlag and cutGetCmdLineArgumenti.

In my compile command I use -I/home/NVIDIA_CUDA_SDK2.0/common/inc to include cutil.h

Do I need to do anything else?

Being a linux moron, I created a new dir in the NVIDIA_CUDA_SDK2.0/projects directory and copied my file into it, and copied and amended a makefile from the matrixmult example. When I ran make from that directory the above errors were not reported, but when Itried to run what I thought was the exe in the obj/release dir I was told I did not have permission, even though I have read, write and execute pernmissions for that directory!

So I copied the makefile from that dir in to the original working dir and amended it so it included, called make from the original working dir and was told cutil.h could not be found.

Suggestions welcome.

Please ignore all of the above. I am new to linux/makefiles. I just found the exe in a completely different dir (but it was not where the release notes said it would be).

However, I still don’t understand why from my original working directory cutil.h the compiler could not find cutil.h? What if I don’t want to build all my projects in the install/project directory? How do I amned the makefile(s) so that cutil.h can be found?