AMD announces its support for open GPGPU API OpenACC
In December last year, Nvidia in partnership with The Portland Group (PGI), cray and caps introduced the new massively parallel computing API for gpu accelerated OpenACC API which its creators call the Post-CUDA API, for unlike owner CUDA API, OpenACC is open and also provides a programming interface much easier.
So far OpenACC was supported by Nvidia GPUs and CPU emulation mode, but surprisingly amd announces support for this API, making it the second largest manufacturer of GPUs in supporting this new GPGPU API-driven open three of the major manufacturers supercomputers (CAPS, CRAY and PGI).
OpenACC not the only new API that provides support AMD, as it also announced its support for standard programming OpenHMPP . With the support of these two new API, AMD claims that remain committed to industry and developer community focused on developing GPU accelerated applications, allowing developers to easily transition their applications to benefiting from the great computing power of GPUs.
According to AMD, the performance of your GPU with this new API is very promising, thanks to the directives of these API facilitates the work of programmers maintaining C or FORTRAN code of existing applications, facilitating their migration. Furthermore CAPS technology help users take advantage of their math libraries AMD Accelerated Parallel Processing API.
At the moment AMD FirePro GPUs are the first to support and OpenHMPP OpenACC and AMD announced that it will demonstrate at the event International Supercomputing Conference (ISC) 2012, which will be held from 17 to 21 June this year.
Link: CAPS Compiler now adds support for AMD’s Heterogeneous Compute Platforms (Planet 3DNow!)Tags: accelerated, amd, API, caps, computing, cray, directive, FirePro, GPGPU, gpu, Nvidia, OpenACC, OpenHMPP, parallel, PGI, programming, support