differences between the results of one processor and that of multi processors

[arash]

Hi everyone,

Does anybody know why the results of a simulation run by a single
processor differ from those of a simulation run by multi processors.
For example the max difference between wave heights is 0.001 but for
suspended sediment is twice, one is 0.0004 m^2/s and another one is
0.0008 m^2/s .

Any advice would be appreciated.

Cheers,
Arash

[Jaap]

Hi, Arash

Thanx for your email. You're making some pretty detailed comparisons!

About the difference in wave height, I wouldn't worry to much about it
(assuming your imposed wave height is much higher thaan 0.001
meter :-)). The sediment transport differs a factor 2 you're saying.
I have two questions:

1. Are you comparing time series or time averaged values?
2. Did you set random = 0 (allows you to get the same random seed
over and over again). If not please do so, for this detailed
comparison.

Probably you want to make sure that mpi simulations give you the same
(comparable) results as single core simulations? How does the overall
moprhological response compare? Are the overall wave height
transformation, flow pattern, sediment transport field comparable?

Good luck!

Jaap