Note: These tutorials are incomplete. More complete versions are being made available for our members. Sign up for free.

GPU Computing

Due to enormous cost of semiconductor design and fabrication, it is not possible to find too many alternate architectures for high-end computing apart from the dominant Intel architecture. Programmable graphics processor units (GPU) remain to be a notable exception.

Where do GPUs come from? Our original picture of computere architecture in Section 1 did not show an important component of the computer, namely its graphic display. Users interact with the computer through the display and they like to the display to be the most pleasant to human eye. Therefore, a large amount of money has been invested to make CPU interact with the display properly and present visually appealing images. Especially the gamers are ready to pay premium to obtain a better look and feel of the animated object.

In earlier days of computing, the display was controlled and operated by the CPU itself, and then another supporting chip working to speed up the display of

another chip. Slowly that chip morphed into another board and so on. The board got its own programmability aside from CPU, because it was far easier to send small pieces of information and code to be displayed than full pixels for each spot.

Looking whether they could go the other way. Eventually, others recognized the immense computing power of the GPUs and investigated whether they could be explored for computation. For these people, Nvidia introduced CUDA programming language.

http://en.wikipedia.org/wiki/Graphics_processing_unit GPU - http://www.nvidia.com/object/what-is-gpu-computing.html CUDA - http://www.nvidia.com/object/cuda_home_new.html

OpenCL http://en.wikipedia.org/wiki/OpenCL

GPU-based Bioinformatics applications

GPU companies Many companies have produced GPUs under a number of brand names. In 2008, Intel, Nvidia and AMD/ATI were the market share leaders, with 49.4%, 27.8% and 20.6% market share respectively. However, those numbers include Intel’s integrated graphics solutions as GPUs. Not counting those numbers, Nvidia and ATI control nearly 100% of the market(as of 2008). In addition, S3 Graphics [10] (owned by VIA Technologies) and Matrox produce GPUs.

Stream Processing and General Purpose GPUs (GPGPU)

GPU SIMD

http://code.google.com/p/stanford-cs193g-sp2010/wiki/TutorialMultidimensionalKernelLaunch http://pdsgroup.hpclab.ceid.upatras.gr/files/CUDA-Parallel-Programming-Tutorial.pdf

GPU drawback

http://stackoverflow.com/questions/124222/what-are-the-advantages-and-disadvantages-of-gpgpu-general-purpose-gpu-develop http://gpgpu.org/ http://www.scribd.com/doc/5687/The-Next-Mainstream-Programming-Language-A-Game-Developers-Perspective-by-Tim-Sweeney http://arstechnica.com/gaming/2008/09/gpu-sweeney-interview/ http://people.maths.ox.ac.uk/~gilesm/hpc/NVIDIA/NVIDIA_CUDA_Tutorial_No_NDA_Apr08.pdf http://en.wikipedia.org/wiki/Stream_processing http://people.maths.ox.ac.uk/~gilesm/hpc/NVIDIA/NVIDIA_CUDA_Tutorial_No_NDA_Apr08.pdf http://www.cs.wm.edu/~kemper/cs654/slides/nvidia.pdf

Architecture

http://www.cs.rochester.edu/~kshen/csc258-spring2011/lectures/student_Tang.pdf http://pdsgroup.hpclab.ceid.upatras.gr/files/CUDA-Parallel-Programming-Tutorial.pdf http://techreport.com/review/17670/nvidia-fermi-gpu-architecture-revealed