Problem with installing CUDA on Quadro

Hi all,

I’m having trouble building the sample projects with CUDA. Here’s the error message I get when I’m trying to do make.

===========================================================
make[1]: Entering directory /home/maryam/NVIDIA_CUDA_SDK/common' src/cutil.cpp:46:27: error: builtin_types.h: No such file or directory make[1]: *** [obj/release/cutil.cpp_o] Error 1 make[1]: Leaving directory /home/maryam/NVIDIA_CUDA_SDK/common’
make: *** [lib/libcutil.so] Error 2

This is not a PATH problem. There is no file named “builtin_types.h” in my entire system.
I would appreciate your help.

Thanks

You need to install the CUDA toolkit. The SDk is just a collection of examples.

I have CUDA toolkit installed !

builtin_types.h is part of the cuda toolkit.

ls -l /usr/local/cuda/include/builtin_types.h
-rw-r–r-- 1 root root 2367 Jan 3 11:11 /usr/local/cuda/include/builtin_types.h

The problem was that the CUDA_INSTALL_PATH wasn’t set right on common.mk, I think that’s a bug for CUDA 1.1. But, now I get the following error when I’m trying to build the projects:

===============================================================

make -C projects/fluidsGL/

make[1]: Entering directory `/home/maryam/NVIDIA_CUDA_SDK/projects/fluidsGL’

/usr/lib/gcc/i586-suse-linux/4.1.2/…/…/…/…/i586-suse-linux/bin/ld: cannot find -lglut

collect2: ld returned 1 exit status

make[1]: *** […/…/bin/linux/release/fluidsGL] Error 1

make[1]: Leaving directory `/home/maryam/NVIDIA_CUDA_SDK/projects/fluidsGL’

make: *** [projects/fluidsGL/Makefile.ph_build] Error 2

================================================================

Any help?

Thanks

No one else has reported problems with CUDA_INSTALL_PATH. How do you believe it should have been set?

In your build failure above, you’re missing libGLUT, which ships with the OS.

This is how CUDA_INSTALL_PATH was set in common.mk:

CUDA_INSTALL_PATH ?= /usr/local/cuda (with the question mark ! )
instead of CUDA_INSTALL_PATH := /usr/local/cuda/cuda

I have installed CUDA 1.0 on a different machine with the same OS (SUSE 10.2) that I’m using now, and didn’t have problems. Doesn’t libGLUT come with display driver?

Thanks.

CUDA_INSTALL_PATH is correct in common.mk. If you look at the Makefile, you’ll it has:

CUDA_INSTALL_PATH ?= /usr/local/cuda

ifdef cuda-install
CUDA_INSTALL_PATH := $(cuda-install)
endif

This serves to set CUDA_INSTALL_PATH to /usr/local/cuda only if you accepted the default install location when you installed the toolkit. The ?= serves as a conditional assignment.

libGL comes with the display driver. libGLUT does not.

Thanks, installed the the libGLUT and it’s working fine now.