PGI CDK Cluster Development Kit

Linux cluster development tools for 32-bit and 64-bit processor-based systems

Linux clusters now are easily affordable by any group interested in serious scientific computing. Getting hardware installed and a network up and running are just the first steps in building a cluster. The PGI CDK Cluster Development Kit contains all the software development and cluster management tools you need to get a Linux cluster running production applications.
The PGI CDK includes:

• Floating multi-user seats for PGI's parallel Fortran, C, and C++ compilers for Linux. Industry-leading single-processor performance and integrated native support for all three popular parallel programming models: HPF, OpenMP, and MPI.
• Graphical MPI and OpenMP Linux Cluster debugging (PGDBG®) and parallel performance profiling (PGPROF®) tools.
• Pre-compiled/pre-configured MPICH message-passing libraries and utilities.
• Pre-compiled/pre-configured TORQUE resource management and batch queueing system to manage the workload and throughput on your cluster.
• Pre-compiled ScaLAPACK parallel math library.
• Optimized BLAS and LAPACK serial math libraries.
• Tutorial examples and programs to help you get your codes up and running quickly using HPF, OpenMP, and MPI messaging.
• Installation utilities to simplify the setup and management of your Linux cluster.

The PGI CDK has the features you need whether you have legacy FORTRAN 77 that relies on Cray, DEC, or IBM extensions, or you are writing modern parallel codes in Fortran 95, High Performance Fortran (HPF), or C/C++. PGI's compilers are highly optimized for Intel Pentium, Xeon and Xeon-64, and AMD Athlon/AthlonXP and Opteron workstations and servers running Linux. Parallelize your applications using MPI or use explicit HPF or OpenMP directives.

In addition to the data parallel capabilities of the PGHPF® compiler, the PGI CDK package includes the PGF90? Fortran 95 compiler; the PGF77® FORTRAN 77 compiler; the PGCC® ANSI C compiler; and the PGC++? ISO/ANSI-compliant C++ compiler. All of these compilers support automatic parallelization for SMP workstations using a simple compiler switch, and full native support for OpenMP directive-based SMP programming. In addition to the full suite of parallel language compilers, the PGI CDK for Linux includes the PGDBG OpenMP/MPI Cluster Linux graphical debugger and the PGPROF OpenMP/MPI Cluster Linux performance profiler. Many vendors support parallelization using simple add-on pre-processing technology, but PGI offers the only intel Pentium, Xeon, Xeon-64 and AMD Athlon/AthlonXP and 64-bit Opteron parallel compilers, debugger and profiler available with integrated parallelization support. This results is faster development, higher performance and much higher reliability for the programmer. Also, with PGI CDK there are no parallel runtime fees of any kind--the executables produced by the PGI compilers can be executed on any compatible system regardless of whether the PGI compilers are installed.

MPICH, developed at the Argonne National Laboratory, is an open source implementation of the Message-Passing Interface (MPI) standard. MPICH is a full implementation of MPI, so your existing MPI applications will port easily to your Linux cluster using the PGI CDK. The TORQUE resource manager is based on OpenPBS and includes recent improvements and extensions.

Technical Features

A partial list of technical features supported includes the following:

• PGF90 OpenMP and auto-parallel Fortran 95 compiler
• PGF77 OpenMP and auto-parallel FORTRAN 77 compiler
• PGHPF data parallel compiler with native full HPF language support
• PGC++ OpenMP and auto-parallel ANSI C++ compiler
• PGCC OpenMP and auto-parallel ANSI/K&R C compiler
• PGDBG graphical Linux Cluster MPI and OpenMP debugger
• PGPROF graphical Linux Cluster MPI and OpenMP performance profiler
• Full 64-bit support on AMD Opteron, AMD Athlon64 and Intel Xeon-64 including full support for -mcmodel=medium and single data objects > 2GB
• Includes separate 32-bit x86 and 64-bit EM64T/AMD64 development environments and compilers
• Optimizing 64-bit code generators with automatic or manual platform selection
• Executables generated by PGI's 32-bit x86 compilers can run unchanged on AMD64 or EM64T processor-based systems
• Cray/DEC/IBM compatibility (including Cray POINTERs)
• Full support for Fortran 95 extensions
• Intel Pentium/Xeon and AMD Athlon/AthlonXP optimizations including SSE/SSE2 and prefetching where supported in hardware
• AMD Opteron optimizations including SSE/SSE2, prefetching, use of extended register sets, and 64-bit addressing
• Large file (> 2GB) support in Fortran on 32-bit x86 systems
• -r8/-i8 compilation flags, 64-bit integers
• Function inlining, loop unrolling, cache tiling
• Fully interoperable with gcc, g77, and gdb
• Byte-swapping I/O for RISC/UNIX interoperability
• Integrated cpp pre-processing
• Threads-based auto-parallelization using PGF77, PGF90, and PGCC C and C++
• Full support for OpenMP in PGF77, PGF90, and PGCC C and C++
• Support for SGI-compatible DOACROSS in PGF77 and PGF90, and for SGI-compatible parallelization pragmas in PGCC C and C++
• Fully upward compatible with PGHPF for high-end parallel systems
• Parallelization of irregular DO loops, FORALLs, and array assignments
• FORALL and F90 array assignment merging
• Re-use of communication schedules
• HPF parallelization using direct accesses to shared memory
• Complete implementation of the HPF Library
• Support for graphical HPF profiling and performance tuning
• Documentation PGI User's Guide, PGI CDK Release Notes
• Unconditional 30 day money back guarantee

System Requirements

• Front-end Node: 32-bit x86, 64-bit AMD64 technology or 64-bit IA32 EM64T processor-based workstation or server with one or more AMD Opteron, AMD Athlon/AthlonXP, Intel Pentium II/III/4, Intel Xeon/Xeon-64 microprocessors
• Cluster Nodes: 32-bit x86, 64-bit AMD64 technology or 64-bit IA32 EM64T processor-based workstation or server with one or more AMD Opteron, AMD Athlon/AthlonXP, Intel Pentium II/III/4, Intel Xeon/Xeon-64 microprocessors.
   
  Note: Hybrid 32-bit x86 and 64-bit (either AMD64 or EM64T) processor-based clusters are not supported, all cluster nodes must be of the same processor type (either 32-bit x86, 64-bit AMD64 or 64-bit EM64T). Clusters with 64-bit processors on the cluster nodes must also use like-technology 64-bit processor-based system for the front-end node.
   
• Operating System: Must be co-installed with a version of the Linux operating system with kernel revision 2.2.10 or higher. The PGI CDK is fully interoperable with newer versions of Linux which use kernel revision 2.2.10 and glibc 2.2.x. For 64-bit processor-based systems the following operating systems are supported:
  
  • SuSE Linux Enterprise Server 8 (SLES 8) (AMD64 only)
  • SuSE 9.0 and 9.1 64-bit Linux
  • Red Hat Enterprise Linux v3.0
  • Fedora Core 2
• Memory: Minimum 32 MB per Cluster Node. 128 MB recommended for Front-end Node
• Hard Disk: 400 MB on front-end node; 50 MB on each Cluster Node
• Peripherals: Mouse or compatible pointing device for use of optional graphical user interfaces. CD-ROM disk drive for installation
• Other: Web browser and Adobe Acrobat Reader for viewing online documentation.