[AMBER] Compiling AMBER on Windows 7 with CUDA

From: Starr Hazard <hazards.musc.edu>
Date: Wed, 11 May 2011 14:14:16 -0400

Folks,

I'm a GPU newbie. I am a WINDOWS Vista novice as well.

I am trying to get AMBER compiled on a DELL Workstation with an NVIDIA
c2050 Tesla GPU.
I hope to learn somethings about GPUs as compute engines.

I have installed AMBER on non-GPU linux systems but never with CUDA.

I am trying to make this work within CYGWIN on a Dell workstation
running WINDOWS 7.

I can compile AmberTools without cuda.

However when I try ./configure -cygwin -cuda gnu

I read that I must define CUDA_HOME in cygwin eg

export CUDA_HOME='c:\\Program Files\\NVIDIA GPU Computing
Toolkit\\CUDA\\v4.0'

I can then echo the $CUDA_HOME variable but

  ls $CUDA_HOME

ls: cannot access c:\\Program: No such file or directory
ls: cannot access Files\\NVIDIA: No such file or directory
ls: cannot access GPU: No such file or directory
ls: cannot access Computing: No such file or directory
ls: cannot access Toolkit\\CUDA\\v4.0: No such file or directory

Which suggests that I have not yet figured out how to tell cygwin about
a WINDOWS path to nvcc? cygpath has not helped so far.

Is there a preferred way to tell cygwin about WINDOWS path with white
spaces?

Starr



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 11 2011 - 12:00:03 PDT
Custom Search