[MITgcm-support] How to build a regional ocean and sea ice model for Arctic

liuyu ly_elvis at yahoo.com.cn
Sat Jul 15 20:35:00 EDT 2006


Hi Chris,

     Your suggestion of working with some part of face 3 of 510*510*6*50 cube is a very good idea.
     We have an IBM 690 with 3 nodes. Node 1 and 2 both have 32 cpus and approximately 45G available memory, but we have some operational models to run during some fixed hours every day, so other tasks will be suspended. Node 3 is always available for us, but it has only 16 cpus and 5 G memory.
     I think a 400*300*50 sub-region will be more practical for me. But how to setup this region, and if the boundary conditions come from the global cube, all the needed data will be a great demand for hard disk space.
  
     Thanks for your help,
				

        liuyu
        ly_elvis at yahoo.com.cn
          2006-07-16




More information about the MITgcm-support mailing list