Today Cray introduced new OpenACC 2.0 support for directives-based programming of accelerators and coprocessors in the company’s supercomputers. Leveraging years of successful production accelerator experience and supercomputing domain knowledge, the Cray OpenACC programming implementation now supports Nvidia Tesla GPU accelerators and Intel Xeon Phi coprocessors on the Cray XC30 series of supercomputers.
Congratulations to Cray for releasing the first compiler with complete support for OpenACC 2.0,” said Duncan Poole, president of OpenACC. “The objective of OpenACC is to provide a programming API that performs well on all types of accelerators and CPUs. The OpenACC organization is responding quickly to create an open standard that is architected to ensure developer code is portable, with good performance on all types of accelerators and CPUs. This rapid pace of leadership further distances the OpenACC specification from similar efforts.”
In an ongoing effort to make it easier for programmers to take advantage of parallel computing, the OpenACC parallel programming standard is a simple, directives-based approach enabling the acceleration of applications on hybrid many-core and multi-core processing systems. The OpenACC approach provides simplified programming of accelerators and coprocessors, wide application support, and flexible and standards-compliant results. OpenACC 2.0 support is delivered for the Cray compiler through the latest release of the Cray Programming Environment.
Read the Full Story.