PGI                   PGI系列 編譯軟體

The Portland Group, Inc. (a.k.a. PGI) is a premier supplier of software compilers and tools for parallel computing, known as PGI® products. The Portland Group offers optimizing parallel FORTRAN 2003, C99 and C++ compilers and tools for workstations, servers and clusters running Linux, MacOS or Windows operating systems based on the following microprocessors:

˙64-bit Intel 64 technology CPU (Intel Core/Xeon/i3,i5,i7 families and related processors)
˙64-bit AMD64 technology CPU (AMD Opteron family and related processors)
˙NVIDIA CUDA GPGPU (Tesla 10/20 product series)

The focus of the Portland Group is to provide the highest performance, production quality cross-platform compilers and software development tools to the High Performance Computing (HPC) market.

PGI Workstation
PGI Workstation™ compilers and tools are for 64-bit x64 and 32-bit x86 and processor-based workstations or servers with or without GPU accelerators. PGI Workstation tools support debugging and profiling up to eight local MPI processes. PGI Workstation includes a single user node-lock license for Linux, Mac OS X or Windows. Licenses are available individually or in multi-packs of 5, 10, 25 or 50.

PGI Server
PGI Server™ compilers and tools are for 64-bit x64 and 32-bit x86 processor-based workstations or servers with or without GPU accelerators. PGI Server tools support debugging and profiling up to 16 local MPI processes. PGI Server includes a multi-user network floating license for 2, 5, 10, 25 or 50 simultaneous users. Licenses are availbe for Linux or Windows. A multi-platform license (Linux, MacOS and Windows) is also available.

PGI CDK
The PGI CDK® Cluster Development Kit® includes compilers and tools specifically optimized for developing, debugging and tuning MPI or hybrid MPI/OpenMP applications running on clusters of x64 or x86 processor-based servers with or without GPU accelerators. The PGI CDK includes support for MPICH-1, MPICH-2, OpenMPI, HP-MPI and MSMPI over Ethernet and MVAPICH over InfiniBand. The PGI CDK includes a multi-user network floating licences for 2, 5, 10, 25 or 50 simultaneous users and is offered in three configurations supporting remote debugging and profiling of up to 16, 64 or 256 MPI processes. Licenses are available for Linux or Windows HPC Server 2008.

PGI Visual Fortran
PGI Visual Fortran® fully integrates the PGI suite of high-performance 64-bit and 32-bit parallel Fortran compilers and tools including PGI Accelerator™ Fortran and CUDA Fortran into Microsoft Windows using Microsoft Visual Studio. PVF license options include a single user node-locked license or a multi-user network floating license.

PGI Compilers and Tools for Mobile and Embedded Platforms
PGI OpenCL Compiler for ARM
PGCL™ is an OpenCL™ framework for compiling and running OpenCL 1.1 embedded profile applications on the ST-Ericsson NovaThor™ U8500 and follow-on platforms using a single ARM core as the OpenCL host and multiple ARM cores as an OpenCL computing device

The following table lists features, capabilities and licensing options that differ among the PGI product families.

　

Feature	PGI Workstation	PGI Server	PGI CDK			PGI Visual Fortran
			16	24	256	PVF	PVF Server
Language Options
Fortran only
C/C++ only
Fortran & C/C++
Operating Systems
Linux only
Mac OS X only
Windows only
Multi-platform
MPI Processes
Debugging with PGDBG®	8	8	16	64	256	8	8
Profiling with PGPROF®	8	8	16	64	256	8	8
Licensing Type
Node-locked Single User
Multi-user Network Floating

The PGI Workstation, PGI Server and PGI CDK license differences are;

PGI Workstation—For deployment on a single node-locked system. This license allows up to eight MPI processes to be debugged using PGDBG as in mpirun -np 8 -dbg=pgdbg foo. All eight processes must be running on the same system as PGDBG. In addition, up to eight MPI processes running on the same system can be profiled using PGPROF with the results presented collectively.
PGI Server—For deployment on any number of network nodes. Options available for up to 2, 5, 10, 25 or 50 simultaneous users. This license allows up to 16 MPI processes to be debugged using PGDBG (for example mpirun -np 16 -dbg=pgdbg foo). All 16 processes must be running on one node. In addition, up to 16 MPI processes running on the same node can be profiled using PGPROF with the results presented collectively.
PGI CDK—For deployment on any number of networked nodes. Options available for up to 2, 5, 10, 25 or 50 simultaneous users. This licenses allows for up to 16, 64, or 256 MPI processes to be debugged using PGDBG and the processes can run on different cluster nodes as designated by the machines.LINUX file but only if the PGI CDK daemons have been installed. In addition, up to 16/64/256 processes can be profiled across the cluster using PGPROF with the information presented collectively.