Different results between supported machines?

[andrew_hoell@student_uml_edu]

Hi,

I successfully ran the same CAM3.1 simulations on two different PC-Linux machines and the results were different. The first machine has 64-bit processors while the second has 32-bit processors. To the best of my knowledge, everything was configured properly before the runs and there were no error outputs. Can anyone provide guidance as to why this is happening? Thank you.

Andy

 

[rneale]

How different were they? The same exact run on different machines will give slightly different answers sue to the different precisions. It only takes the tiniest perturbation to the answers for it to propagate through the prognostics equations of the model and make two runs look different, but similar enough that they could be considered two ensemble members. As you run for longer averages of the runs should look increasingly similar.