Re: [AMBER-Developers] CUDA build fails for PBSA

From: Ray Luo <rluo.uci.edu> Date: Wed, 8 Mar 2017 13:36:08 -0800

--
Ray Luo, Ph.D.
Professor of Structural Biology/Biochemistry/Biophysics,
Chemical Physics, Biomedical Engineering, and Chemical Engineering
Department of Molecular Biology and Biochemistry
University of California, Irvine, CA 92697-3900
On Tue, Mar 7, 2017 at 9:48 AM, Daniel Roe <daniel.r.roe.gmail.com> wrote:
> Clean GIT master: 0494517ae737e465fdd737684454526f7ad14d26
>
> ./configure --with-python `which python` -cuda gnu
>
> ```
> make[2]: Entering directory
> `/mnt/raid0/droe/Amber/GIT/amber-test/AmberTools/src/pbsa'
> Error: CUSP_HOME is not properly set. This must point to your CUSP
> library installation.
> make[2]: *** [.cuda_precheck] Error 2
> make[2]: Leaving directory
> `/mnt/raid0/droe/Amber/GIT/amber-test/AmberTools/src/pbsa'
> make[1]: *** [cuda_serial] Error 2
> ```
>
> The message offers no clues as to what CUSP_HOME is supposed to refer
> to but I assume it's this: https://developer.nvidia.com/cusp
>
> CUSP appears to come with CUDA 8.0 (at least there is a libcusparse.so
> in the lib64 subdir). However, some fiddling around shows PBSA expects
> CUSP_HOME to point to something that has a /cusp subdirectory. I'm
> guessing previously this was an add-on install for CUDA?
>
> Bottom line is that currently it seems that CUDA-enabled PBSA depends
> on this, but it is not properly checked for during configure. Probably
> configure should search for CUSP and set a config.h variable pointing
> to it, or leave it blank if not found. The entire CUDA install
> shouldn't fail just because this one piece isn't there.
>
> -Dan
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers