Does IntelArBB allow user to create applications targeteed for HPP, inculding GPU?

Does IntelArBB allow user to create applications targeteed for HPP, inculding GPU?

Imagen de Xiang Ma

I understood that previously the RapidMind Development Platform allows developers to use standard C++ programming to easily create applications targeted for highperformance
processors, including GPUs, the CellBE, and multi-core CPUs. In the case of the GPU the RapidMind platform
can be used for both shaders and general purpose processing.

In http://software.intel.com/en-us/articles/intel-array-building-blocks/, it states that "Intel Array Building Blocks is the combination of Intels Ct Technology and RapidMind technology."
My questions:
1). Does IntelArBB still allow user to create applications targeteed for HPP, inculding GPU?
2). If not, how about performace of IntelArBB comparing with using NVIDIA GPU for image processing application?
3). How about performace of IntelArBB comparing with using IPP for image processing application?

Thanks very much for help!

publicaciones de 3 / 0 nuevos
Último envío
Para obtener más información sobre las optimizaciones del compilador, consulte el aviso sobre la optimización.
Imagen de Noah Clemons (Intel)

Please take a look at our Knowledge Base Posting for a more thorough answer to your question.

Intel ArBB currently works for Intel's multicore and forthcoming many core processors. Intel ArBB is currently in Beta and gathering feedback on performance. Many Beta testers are comparing Nvidia GPU implementations with ArBB, but there are a wide variety of algorithms and workloads that do not fit within the GPU model of programming at all. So there are many situations wherean Nvidia implementation cannot be compared to ArBB because they cannotprogram it for a GPU. ArBB is being designed so thatyou code as if it is serial, single-core code and you gain advantages of multi and many core throughput computing without having to fit any algorithm in a GPU model of programming.

Please also take a look at my Knowledge Base article about comparing ArBB with other Intel internal tools. ArBB is meant for coding custom algorithms where the canned functions of IPP and MKL do not apply. Why would you code the same algorithm for IPP when it is already optimized at the maximum level, with a different hardware-specific implementation for each processor?

Imagen de

Inicie sesión para dejar un comentario.