March 2012

Michael Wolfe's Programming Guide
First Look: The PGI Accelerator Compilers with OpenACC
Intended for those developers already familiar with the PGI Accelerator programming model, this article introduces PGI support for the OpenACC API and describes characteristics of and differences between the two models.
For developers new to GPU programming, Michael has updated the first part of his popular PGI Programming Model on NVIDIA GPU article to include OpenACC. At the same time he has split the article into two; one article focusing on PGI Accelerator Fortran and the other on PGI Accelerator C.
Also in this issue…
The New PGC++ Compiler for Linux
In this article we present user-visible changes coming soon in the PGI C++ compiler for Linux. The new compiler uses the system C++ libraries, avoiding problems that arise from dealing with another C++ and Standard Template Library.
5x in 5 Hours: Porting a 3D Elastic Wave Simulator to GPUs Using OpenACC
This step-by-step guide shows how we achieved a 5x speed-up of the SEISMIC_CPML application in just 5 hours by porting it to run on GPUs using OpenACC directives. The article describes the basic steps involved, some optimization tips, identifies several potential pitfalls, and most of all, demonstrate how you can apply the OpenACC directives to your own code.
Using PGDBG to Debug Programs on Remote Systems
Until now, the PGI debugger PGDBG®, has only supported debugging a locally running program. The PGI 2012 release of PGDBG introduces support for debugging a program running on a different system than the debugger.
Upcoming Events
In cooperation with NVIDIA, Michael Wolfe will be presenting a webinar on "Getting the Most Out of OpenACC—Directive Optimization with PGI Accelerator" on Thursday, 29 March at 9:00 PDT. Registration is free.
PGI will be presenting a paper on OpenACC and participating in a full day OpenACC tutorial at the NVIDIA GPU Technology Conference in San Jose, California, 14-17 May.
PGI is exhibiting in booth #855 at ISC'12 in Hamburg, Germany, 17-21 June.
Resources
PGI OpenACC Getting Started Guide
Recent News
First PGI Compilers to Support OpenACC Now Available
New 2012 Release of PGI Compilers and Development Tools
PGI Ships OpenCL Compiler for Multi-core ARM
PGI Ships CUDA C/C++ Compiler for Multi-core x86
nCore and PGI Team to Produce GPU Programming Training
Next Issue Highlights
Advanced OpenACC Features
Multiple GPU Support with OpenACC
Comparing CUDA and OpenMP Performance on x86
PGI Accelerator Programming Model v2.0

PGI OpenCL Compiler for ARM
PGCL is an OpenCL compiler framework for ARM processor-based Android devices. With it, developers can create Android applications that run in parallel across multiple ARM CPU cores. PGCL gives Android developers an avenue to create applications that are multi-core enabled, inherently scalable and forward-portable to the coming accelerator-enabled mobile platforms.
OpenCL is a trademark of Apple Inc. used by permission by Khronos.
- The Portland Group, Inc.
- Suite 320
- Two Centerpointe Drive
- Lake Oswego, OR 97035