Assessment of SIMD programming on graphics card

Nagendran, S and Ashrafulla, Mohammad and Bakka, Jagadevi (2008) Assessment of SIMD programming on graphics card. In: Proceedings of the International Conference on Aerospace Science and Technology (INCAST 2008-128), 26-28 Jun 2008, Bangalore, India.

[img] PDF

Download (676kB)
[img] Indexer Terms (Generate index codes conversion from application/pdf to indexcodes)

Download (4kB)


Recent advance of the technologies incorporated in graphics hardware has enabled general-purpose computations on graphics hardware, which can further be used for high-performance computation in low cost. Over the past six years GPUs have shown a marked increase in its performance and capabilities. Modern GPUs are now fully programmable massively parallel floating point processors demonstrating a performance/cost ratio superior to central processing units (CPUs) with computations of high arithmetic intensity. This effort in general-purpose computing on the GPU, also known as GPU computing, has positioned the GPU as a compelling alternative to traditional microprocessors in high-performance computer systems of the future. GPU is massively multithreaded parallel computing SIMD platform and NVIDIA is best known for a line of outstanding graphics processors that have become popular as the basis for graphics cards. Modern NVIDIA GPUs are not single processors but rather are parallel supercomputers on a chip that consist of very many, very fast processors. NVIDIA has supported this trend by releasing the CUDA (Compute Unified Device Architecture) interface library to allow intrepid applications developers to write code that can be uploaded for execution by NVIDIA's massively parallel GPUs. In this paper we analyze the performance of SIMD applications which are accelerated by GPU comparing that to their CPU counterparts. We initially design a set of SIMD programs using C language which are executed on the CPU, the same programs are data parallelized with the help of CUDA API and executed with GPU to find the performance gain and thereby analyze what potential benefits GPU can offer.

Item Type: Conference or Workshop Item (Paper)
Additional Information: Copyright for this article belongs to National Aerospace Laboratories
Uncontrolled Keywords: Graphics processing unit;CUDA (Compute unified device architecture);FFT (Fast fourier transform);SIMD programming
Subjects: MATHEMATICAL AND COMPUTER SCIENCES > Computer Programming and Software
Depositing User: Ms. Alphones Mary
Date Deposited: 25 Mar 2009
Last Modified: 17 Jun 2010 09:25

Actions (login required)

View Item View Item