[Starpu-devel] StarPU Configuration/Build error

[Pushkar Ratnalikar <pratnali@umail.iu.edu>]

Hi,

I am trying to build StarPU 1.3 from source on a machine with NVIDIA GPU and CUDA 5.5. This is the command I am using

../configure --without-hwloc --disable-opencl --with-cuda-dir=/opt/nvidia/cudatoolkit/default --with-cuda-include-dir=/opt/nvidia/cudatoolkit/default/include --with-cuda-lib-dir=/opt/nvidia/cudatoolkit/default/lib64

I get an error during configuration 

-----------------------------------------------------------------------------------------------------

checking for library containing set_mempolicy... -lnuma

checking whether libnuma is available... yes

checking maximum number of sched_ctxs... 10

checking maximum number of CPUs... 64

checking whether CPUs should be used... yes

checking maximum number of CUDA devices... 4

checking whether CUDA is available in /opt/nvidia/cudatoolkit/default... 

checking cuda.h usability... yes

checking cuda.h presence... yes

checking for cuda.h... yes

checking whether CUDA is available with: -L/opt/nvidia/cudatoolkit/default/lib64... 

checking for main in -lcuda... no

checking whether CUDA is available... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /usr/local/cuda... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /c/cuda... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /cygdrive/c/cuda... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /opt/cuda... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /opt/nvidia/cudatoolkit/default... 

checking cuda.h usability... yes

checking cuda.h presence... yes

checking for cuda.h... yes

checking whether CUDA is available with: -L/opt/nvidia/cudatoolkit/default/lib64... 

checking for main in -lcuda... no

checking whether CUDA is available with: -L/opt/nvidia/cudatoolkit/default/lib... 

checking for main in -lcuda... no

checking whether CUDA is available with: -L/opt/nvidia/cudatoolkit/default/lib/x64... 

checking for main in -lcuda... no

checking whether CUDA is available with: -L/opt/nvidia/cudatoolkit/default/lib/Win32... 

checking for main in -lcuda... no

checking whether CUDA is available in /..... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /..... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /..... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is available in /..... 

checking cuda.h usability... no

checking cuda.h presence... no

checking for cuda.h... no

checking whether CUDA is working... no

configure: error: cannot find CUDA

-----------------------------------------------------------------------------------------------------

It seems that it initially finds CUDA installation in the path specified, but continues to look at other locations. 

I have attached that config.log file.

Please let me know if I am missing something.

Thanks,

Pushkar

Attachment: config.log
Description: Binary data