Re: [AMBER] CUDA MPI compilation issue

From: Jason Swails <>
Date: Thu, 11 Jul 2013 23:26:12 -0400

On Jul 11, 2013, at 11:06 PM, "Wang Chern Hoe (Dr)" <> wrote:

> I ran into this issue upon issuing:
> ./configure -cuda -mpi intel
> make -j 4 install
> home/amber12-cuda/lib -lnetcdf -shared-intel
> ld: cannot find -ltorque
> make[3]: *** [pmemd.cuda.MPI] Error 1
> How can I correct this ?

This indicates to me that you are installing this on a cluster and using an MPI installed on the cluster that the admins linked to the torque library (which is the scheduler). This is a good thing, in my opinion since it takes a lot of the labor out of launching MPI programs by letting torque and MPI communicate regarding the resources that should be used (and where they should be used).

However, it appears that the torque developer files are not in your environment. Oftentimes, clusters manage their development environment using the "module" program/command. In this case, you can search for a torque module and make sure it's loaded.

Your cluster should have some documentation that describes what you should do to compile, for instance, an MPI program. I would suggest consulting the cluster documentation or submitting a help ticket with your issue. They will certainly know where and how to set up libraries on their system better than I would ;).


> Thanks
> CONFIDENTIALITY:This email is intended solely for the person(s) named and may be confidential and/or privileged.If you are not the intended recipient,please delete it,notify us and do not copy,use,or disclose its content.
> Towards A Sustainable Earth:Print Only When Necessary.Thank you.
> _______________________________________________
> AMBER mailing list

Jason Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Thu Jul 11 2013 - 20:30:08 PDT
Custom Search