[AMBER] Problem building Amber16 with -cuda

From: Ryan Novosielski <novosirj.rutgers.edu>
Date: Wed, 11 Apr 2018 07:28:19 +0000

Good morning,

I’m attempting to build Amber16 with AmberTools17 and running into a problem that I don’t believe I previously hit building the same version of AMBER some time back. The error I’m getting is related to the linker not finding -lcuda, first when attempting to build pbsa.cuda and then later trying to build cpptraj.cuda. My understanding is that -lcuda is for the NVIDIA driver API and -lcudart is for the NVIDIA runtime API. As I’m fairly certain I’ve never had the driver installed on the machine where I’m attempting this build, it seems to me that this -lcuda was added to some places in the code at some point. However, if I remove this -lcuda from $AMBERHOME/config.h and $AMBERHOME/AmberTools/src/cpptraj/config.h, the software appears to build fine and passes all of the cuda_serial “make test” and all but one of the cuda_parallel “make test” (the lone difference is at the end of my message — I’m skeptical that one has anything to do with the other).

Does anyone happen to know what the reason is for this change, or at least can confirm that it is a change? So far, I’ve only come across one other software package that has insisted on having the driver installed (though I understand some earlier versions of GROMACS also required it, and there are software packages that do). We generally don’t build on systems that have the actual hardware available.

Thanks in advance!

PS: here’s the one failure with pmemd.cuda.MPI:

[novosirj.perceval2 16]$ more /opt/sw/packages/intel_18.0/mvapich2-2.2/cuda-8.0/amber/16/logs/test_amber_cuda_parallel/2018-04-11_03-16-52.diff
possible FAILURE: check mdout.cellulose_nvt.dif
/opt/sw/packages/intel_18.0/mvapich2-2.2/cuda-8.0/amber/16/test/cuda/cellulose
212c212
< Etot = 5.8651 EKtot = 273.2191 EPtot = 276.4278
> Etot = 5.8650 EKtot = 273.2191 EPtot = 276.4278
### Maximum absolute error in matching lines = 1.00e-04 at line 212 field 3
### Maximum relative error in matching lines = 1.71e-05 at line 212 field 3
---------------------------------------

--
____
|| \\UTGERS,  	 |---------------------------*O*---------------------------
||_// the State	 |         Ryan Novosielski - novosirj.rutgers.edu
|| \\ University | Sr. Technologist - 973/972.0922 (2x0922) ~*~ RBHS Campus
||  \\    of NJ	 | Office of Advanced Research Computing - MSB C630, Newark
     `'



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Wed Apr 11 2018 - 00:30:03 PDT
Custom Search