PDA

View Full Version : OpenACC Complier: Fast, Portable, Just $199 from CAPS


News
06-05-12, 07:30 PM
http://5601-blogs-nvidia-com.voxcdn.com/wp-content/uploads/2012/06/Unknown1-650x366.pngDevelopers have a new ally in OpenACC compilers that can get their code up and running on GPUs in just minutes.

If you are intrigued, you are not alone. OpenACC has garnered significant mindshare and momentum in the HPC community as it promises to deliver a simple and portable programming platform by using a directive-based approach, similar to OpenMP. ¬*Consider that just a few weeks ago (http://blogs.nvidia.com/tag/gtc/) at the GPU Technology Conference (http://www.gputechconf.com/page/home.html), almost a thousand developers packed OpenACC sessions eager to learn more.

http://5601-blogs-nvidia-com.voxcdn.com/wp-content/uploads/2012/06/openacc-compiler-workflow-300x298.png (http://5601-blogs-nvidia-com.voxcdn.com/wp-content/uploads/2012/06/openacc-compiler-workflow.png) The OpenACC workflow
CAPS enterprises recently released their OpenACC 1.0 compliant compiler, and it‚??s available now with a limited-time discount ‚?? more below.

CAPS enterprise is a founding member of the OpenACC committee, and has been an innovator in directives-based compilers for GPU accelerators.¬* The many-core, hybrid computing revolution may be unfolding right before our eyes, but there are still developers looking to use simple directives-based constructs like OpenACC to parallelize their code.

Early results on the CAPS OpenACC compiler have been nothing short of spectacular. In a recent webinar (http://developer.download.nvidia.com/CUDA/training/OpenACC_forscience_CAPS.mp4), CAPS discussed how a large-scale structure and galaxy formation code, known as HydroC, was accelerated 2.3x with OpenACC directives on the Tesla C2070 GPU (compared with a quad-core CPU). This was done without modifying the original, internal algorithms.¬* They also demonstrated a 24x speedup (compared with serial CPU code) in an edge detection algorithm called Sobel Filter by inserting only six OpenACC directives.

If you‚??ve been waiting for a simpler, more portable approach to GPU computing, OpenACC is the programming model you‚??ve been looking for. And for a limited time, CAPS is offering their OpenACC compiler for $199. This offer includes an¬*OpenACC license and one year maintenance and support.

For more information, see the CAPS OpenACC page (http://www.caps-entreprise.com/openacc).

To learn more about OpenACC and CAPS, check out this¬*two-part series (http://www.gputechconf.com/gtcnew/on-demand-GTC.php?sessionTopic=&searchByKeyword=openacc&submit=&select=+&sessionEvent=2&sessionYear=&sessionFormat=#1578) on how to get your code running on GPUs with the CAPS compiler.

If you‚??ve tried OpenACC, we‚??d love to hear about your experience in the comments below.

http://feeds.feedburner.com/~ff/nvidiablog?d=yIl2AUoC8zA (http://feeds.feedburner.com/~ff/nvidiablog?a=u3n25rnOVDA:uLd0dbQHEgs:yIl2AUoC8zA) http://feeds.feedburner.com/~ff/nvidiablog?i=u3n25rnOVDA:uLd0dbQHEgs:V_sGLiPBpWU (http://feeds.feedburner.com/~ff/nvidiablog?a=u3n25rnOVDA:uLd0dbQHEgs:V_sGLiPBpWU)
http://feeds.feedburner.com/~r/nvidiablog/~4/u3n25rnOVDA

More... (http://feedproxy.google.com/~r/nvidiablog/~3/u3n25rnOVDA/)