Re: NetCDF on Connection Machines

• To: "Glenn P. Davis" <davis@xxxxxxxxxxxxxxxx>
• Subject: Re: NetCDF on Connection Machines
• From: dwf@xxxxxxxxxxxx (David W. Forslund)
• Date: Tue, 7 Sep 93 16:37:18 -0600

"Glenn P. Davis" writes:
 > > Has anyone ported netCDF to either a CM-200 or a CM-5?
 > I believe netcdf has been ported and used on CM-5, a numbre
 > of years ago.
 > 
 > Of course, since netCDF is essentially I/O bound, it
 > doesn't benefit significantly from massive parallelism.
 > As I recall, it essentially runs on the "front end" machine
 > in a CM architecture. This reduces your question to 
 > "Has anyone ported netCDF to a CM front end machine, eg,
 > a vax or sun" and the answer is yes.
 > 
 > -glenn

But there is an effort underway here at Los Alamos to port netCDF to
the parallel disk system (SDA) provided by TMC.   I/O can be done
massively parallel just like computing.  We will keep people posted
when it is in a useable form.  This is critical if one is to use it
effectively on the CM-5.

David Forslund
Advanced Computing Laboratory
MS B287
Los Alamos National Laboratory
Los Alamos, NM 87545

Voice:(505) 665-1907
FAX: (505) 665-4939
EMAIL: dwf@xxxxxxxx

• References:
  • Re: NetCDF on Connection Machines
    • From: Glenn P. Davis