>
>I have just joined this mailing group as I am considering using CCM3 to
>model the effects of Saharan dust on North African climate. 
>
>My first concern is whether I will be able to run the model using the
>facilities
>available at my university, which operates a "workstation farm" on Digital
>Alphas
>running PVM. Does CCM3 have machine-specific code, and if so is an Alpha 
>version available or planned?
>

CCM3 is supported on: SGI, IBM-RS6k, Cray, Cray-T3D, and sun.  NCAR doesn't
support a alpha version.  Has anyone else done the port?

The main "machine-specific" code are Cray pointers, which are supported under
F77 on all the above machines and also F90 under the Cray.  If the Alpha doesn't
have Cray pointers in it's implementation of F77 or F90 you would have a hard
time porting the code.  Other than that I suspect you won't really have any
problems.

What are Cray pointers?  Here's a sample from the code...

      pointer (phbuf,buf1)   ! history buffer: dynamically allocated

This allows dynamic allocation of arrays sizes.  They are used quite extensively
in the code.  F90 also has a stricter type of pointers, but the syntax is quite
a bit different.  I think eventually CCM3 will convert over to F90 pointers as
F90 only supports Cray pointers on the Cray now.  But, that will be some time.
See "http://www.scd.ucar.edu/tcg/consweb/Fortran90/F90.html" for more details on
F90 and the difference between Cray pointers and F90 pointers.

Good luck.  If you do get it to work on the Alpha let us know.

Erik
-- 
Erik Kluzek, (CGD at NCAR)
National Center for Atmospheric Research
Boulder CO, (off) (303)497-1326 (fax) (303)497-1324
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!