linux cluster success stories in either of the follow configurations ...?

[cathryn_meyer@yale_edu]

Hello,

We are trying to get CCSM3 up and running on our linux cluster here at Yale, on nodes that are:

IBM Blade Server HS20
2 Intel Xeon 2.6GHz Processors 32-bit
2GB Ram
Gig-Ethernet
OS - Red Hat Linux release 9 (Shrike)
PGI 6.0-5
MPICH 1.2.5

Has anybody gotten CCSM3 to run in this same configuration? We can get the model to compile, but it runs for about and hour and then crashes, we think it has something to do with the coupler and communication between the different components. I can send error messages to anybody who might be of help.

Or ...

Has anybody gotten CCSM3 to successfully run in this other configuration, with nodes as follows:

IBM eServer xSeries 325
2 AMD 64-bit Opteron 1.4GHz Processors
2GB Ram
Networking - Myrinet
OS - SuSE SLES 8 64bit
PGI 6.0-5
MPICH 1.2.5 or 1.2.7

When we try to get CCSM to run on this cluster, it locks up during compilation.

ANY success stories, thoughts, helpful advice would be MUCH appreciated!

Thanks,
Cathy