[MITgcm-support] RE: section 3.2 Obtaining the code

[Ed Hill]

Hi Enda,

Thanks for sending us a note -- I've subscribed you to the MITgcm-
support list and please feel free to send more questions to this list.

MITgcm runs on basically every "Unix-like" operating system including
Linux, MacOS X, and even Windows within the Cygwin environment.  We
routinely use it on x86, Mac G5s, x86_64, Alpha, and Itanium2-based
systems.  So yes, it works on all those architectures.

In terms of performance, we just don't have the time to run a lot of
benchmarks for the sake of platform comparisons.  The MITgcm code is
rather modular and the overall throughput is a function of many things
including which modules are activated, the volume of I/O performed, etc.

And, like many large fluid flow codes, MITgcm is usually memory-
bandwidth-limited on individual CPUs and rather sensitive to the
communications latency when run with MPI on clusters.  We've seen nearly
linear per-CPU speedups on individual Opteron systems where the memory
scales with the number of processors.  And we've seen huge improvements
with MPI scalability when using lower latency interconnects (eg. Altix
systems or clusters with Myrinet interconnects versus cheaper/slower
Gigabit Ethernet).

Ed

===  original email follows  ===

Hello,

Just to let you know that I have downloaded your MIT GCM code.  I have a
query from Banerjee Inst. Allahabad, India, as to whether this model
runs on Itanium2/Linux platforms, and if so how performance compares
with Opteron and/or Xeon platforms.  I don't know, so I thought I might
just try some quick tests.  However, if you have any such performance
information you can share, I'd be very grateful to see it!

Best wishes,
Enda
==================================================
Enda O'Brien
High-Performance Technical Computing
Hewlett-Packard Co. 
Ballybrit Business Park, Galway, Ireland 
Tel +353 91 754602            Fax +353 91 754435
enda.obrien at hp.com
==================================================

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464