Design and Performance Evaluation of Image Processing Algorithms on GPUs

Park, In Kyu; Singhal, Nitin; Lee, Man Hee; Cho, Sungdae; Kim, Chris W

doi:10.1109/tpds.2010.115

Cited by 124 publications

(50 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…When kernel functions run N times, they work in parallel on N channels. This property of kernel separates from familiar C functions [15], [16]. The primary parallel construct is a data-parallel, SPMD kernel function.…”

Section: Parallel Implementationmentioning

confidence: 99%

Effective Gaussian Blurring Process on Graphics Processing Unit with CUDA

Bozkurt¹,

Yağanoğlu²,

Günay³

2015

IJMLC

View full text Add to dashboard Cite

Abstract-Single threaded applications cannot catch up with parallel systems when it comes to scalability mainly because of the clock frequency limitation of modern CPUs and economical reasons. Using shared memory or distributed memory architectures parallel systems can provide tremendous speed up compared to single threaded systems. CUDA, a shared memory parallel software, is considered to be a powerful language because of its easy thread management aspect and support for GPUs. Gauss blurring is a well-known image processing technique which reduces image noise and detail. Because of the high computation requirement of this technique, single threaded applications exhibit poor performance. In this paper we show that orders of magnitude speed up can be achieved by carrying out this operation on CUDA architecture with the help of high parallelism.

show abstract

Section: Parallel Implementationmentioning

confidence: 99%

Effective Gaussian Blurring Process on Graphics Processing Unit with CUDA

Bozkurt¹,

Yağanoğlu²,

Günay³

2015

IJMLC

View full text Add to dashboard Cite

show abstract

“…Second, the performance and scalability of the proposed approach will be evaluated by performing a comparison between the CSX600-based implementation and the fastest host only version using OpenMP on a state-of-theart multiprocessor server system featuring a total of 8 system core. Similar experiments using OpenMP have been used in image processing algorithms on GPUs [15]. Third, the experiments will provide some initial evidence for the claim that the proposed AP-based ATC system implementation exhibits greater efficiency and a huge increase in the degree of predictability than the MIMD-based solution.…”

Section: Resultsmentioning

confidence: 85%

Scalable and Efficient Associative Processor Solution to Guarantee Real-Time Requirements for Air Traffic Control Systems

Yuan

Baker

Meilander

et al. 2012

2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops &Amp; PhD Forum

View full text Add to dashboard Cite

Abstract-This paper proposes a solution to air traffic control (ATC) using an enhanced SIMD machine model called an Associative Processor (AP). Our solution differs from previous ATC systems that are designed for MIMD computers and have a great deal of difficulty meeting the predictability requirements for ATC, which are critical for meeting the strict certification standards required for safety critical software components. The proposed AP solution supports accurate predictions of worst case execution times and guarantees all deadlines are met. Furthermore, the software developed based on the AP model is much simpler and smaller in size than the current corresponding ATC software. As the associative processor is built from SIMD hardware, it is considerably cheaper and simpler than the MIMD hardware currently used to support ATC. We have designed a prototype for eight ATC real-time tasks on ClearSpeed CSX600 accelerator that is used to emulate AP. Performance is evaluated in terms of execution time and predictability and is compared to the fastest host-only version implemented using OpenMP on an 8-core multiprocessor (MIMD). Our extensive experiments show that the AP implementation meets all deadlines that can be statically scheduled. To the contrary, some tasks miss their deadlines when implemented on MIMD. It is shown that the proposed AP solution will support accurate and meaningful predictions of worst case execution times and will guarantee that all deadlines are met.

show abstract

“…However, the particular case of windowed computation has not received much attention in an analytical framework. On the other hand there is a lot of work on implementing image and windowed computations on particular platforms (notably FPGAs and GPUs) [1,2,5,8,11,12]. To our knowledge, ours is the first work to analytically study windowed computations and their execution on a pipelined platform of computation.…”

Section: Introductionmentioning

confidence: 99%

Pipelined Execution of Windowed Image Computations

Vaidyanathan

Vinukonda

Lessing

2013

IJNC

View full text Add to dashboard Cite

Many image processing operations manipulate an individual pixel using the values of other pixels in the neighborhood. Such operations are called windowed operations. The size of the windowed operation is a measure of the size of the given pixel's neighborhood. A windowed computation applies a windowed operation on all pixels of the image. An image processing application is typically a sequence of windowed computations. While windowed computations admit high parallelism, the cost of inputting and outputting the image often restricts the computation to a few computational units.In this paper we analytically study the running of a sequence of z windowed computations, each of size w, on a z-stage pipelined computational model. For an N × N image and n × n input/output bandwidth per stage, we show that the sequence of windowed computations can be run in at most over a single stage; δ, the overhead is quite small. We also show that the memory requirement per stage is O(wN + n 2 ). With values of N , n and w that reflect the current stateof-the-art, over 20 pipeline stages can be sustained with less than 5% overhead for a 10M-pixel image. Each of these stages would require less than 128 Kbytes of storage.

show abstract

Design and Performance Evaluation of Image Processing Algorithms on GPUs

Cited by 124 publications

References 28 publications

Effective Gaussian Blurring Process on Graphics Processing Unit with CUDA

Effective Gaussian Blurring Process on Graphics Processing Unit with CUDA

Scalable and Efficient Associative Processor Solution to Guarantee Real-Time Requirements for Air Traffic Control Systems

Pipelined Execution of Windowed Image Computations

Contact Info

Product

Resources

About