This User's Guide is an extension to the NCAR CCM2 User's Guide [3] for execution in a distributed memory parallel environment. Many of the procedures described are installation and machine specific to the Intel Paragon at Oak Ridge National Laboratory or the IBM SP2 at Argonne National Laboratory. The implementations of the PCCM2 have left as much as possible of the original NCAR coding and procedures in place so that anyone familiar with the use of the CCM2 code can make the transition to use of the parallel code with minimal effort. The ``physics'' routines are essentially untouched allowing easy modification of PCCM2 by users in pursuit of their scientific research with little regard to parallelism.
Following the NCAR CCM2 User's Guide [3] we first present the details necessary for running the code on various parallel platforms. Next we describe PCCM2 internals and material that might be useful for making modifications to the code and setting it up on a particular system. The first subsection presents execution scripts for use on the Intel Paragon running OSF/1 Version 1.3 or later. Environment variables that must be set for the execution are described and pertinent libraries are named for the standard control simulation.
Section 5.2 describes the setup of the history tape output in the parallel environment. To efficiently execute in a parallel environment the large volume of code output must be processed in parallel, striping the output across several RAID's and using multiple I/O nodes. High bandwidth is achieved in this way. Accessing the history tape outputs and converting them to a format readable by other machines is accomplished using the filters described in section 5.3.5.
The final section discusses post processing tools for the PCCM2 model output.