[cam-users] A problem with CAM2.0.1 restart run

Jim Rosinski rosinski@ucar.edu
Thu, 11 Sep 2003 09:40:17 -0600 (MDT)


On Thu, 11 Sep 2003, Ravi S Nanjundiah wrote:

>  Is it possible that different number of processors are being used for the
> two runs? Then is it likely that order of summation in Gaussian quadrature
> is different leading to different results?

Differing numbers of MPI tasks should also give identical results, since
the dynamics only uses a 1-d decomposition and all sums are done on-processor.

Wan Hui wrote:

> My cases were run with both SPMD and SMP DISABLED ,and we had made some
> changes in the Makefile to use pgf90 and cc as Fortran and C compiler
> respectively. Do you think that may be the cause of the problem?  I am
> planning to run the model on another system to see if I can get the right
> result.

Combining pgf90 and cc should not be a problem.  I know others with access to
pgf90 but not pgcc have done this as well.  

Jim Rosinski