From NWChem
You are viewing a single post from the thread title above
-
Edoapra Forum:Admin, Forum:Mod, bureaucrat, sysop
|
|
|
Forum Vet
Threads 7
Posts 1289
|
|
| 10:55:39 AM PST - Wed, Feb 8th 2012 |
|
Please set FC=ifort
|
Christof,
I see from your compilation settings that you have supplied a long list of compiler options with the variable FC. I see pleny of potential problems with this since 1) the -openmp option is likely to cause runtime conflicts with the Global Arrays parallelization and 2) the makefile structure will be confused by this long variable for FC. If you really want to change the compiler options, the recommended way would be (for example)
make FC=ifort FOPTIMIZE="-O2 -xHost"
I will try to answer your questions next
1. The "Linux workstation platform" compilation instructions result in a binary to be used with mpirun, since USE_MPI is set equal to y
2. testtask should not fail for the 3-D Global Arrays test
3. If no high-speed network is present, you can let ARMCI_NETWORK undefined. Another option, recently introduced in GA/ARMCI and that we have not thoroughly tested yet, is MPI-MT. I would suggest, first to try to get the vanilla ARMCI compilation to work,
and then you might try the ARMCI_NETWORK=MPI-MT setting
4. I am not sure how to answer this one, since I can see conflicting details in your question. If you compiler NWChem with USE_MPI=y, tests have to be run with doqmtests.mpi, since only doqmtests.mpi uses the needed mpirun
Edo
|
|
|
AWC's:
2.5.10 MediaWiki - Stand Alone Forum Extension
Forum theme style by: AWC
