• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.5
• /
• pp.2003-2021
• /
• 2020

Today's computer systems are widely integrated with CPU and GPU to achieve considerable performance, but energy consumption of such system directly affects operational cost, maintainability and environmental problem, which has been aroused wide concern by researchers, computer architects, and developers. To cope with energy problem, we propose a task-scheduling framework to reduce energy under performance constraint by rationally allocating the tasks across the CPU and GPU. The framework first collects the estimated energy consumption of programs and performance information. Next, we use above information to formalize the scheduling problem as the 0-1 knapsack problem. Then, we elaborate our experiment on typical platform to verify proposed scheduling framework. The experimental results show that our proposed algorithm saves 14.97% energy compared with that of the time-oriented policy and yields 37.23% performance improvement than that of energy-oriented scheme on average.

• International journal of advanced smart convergence
• /
• v.7 no.3
• /
• pp.8-14
• /
• 2018

In the field of surveillance cameras, it is possible to increase the people detection accuracy by using depth information indicating the distance between the camera and the object. In general, depth information is obtained by calculating the parallax information of the stereo camera. However, this method is difficult to operate in real time in the embedded environment due to the large amount of computation. Jetson TX2, released by NVIDIA in March 2017, is a high-performance embedded board with a GPU that enables parallel processing using the GPU. In this paper, a stereo camera is installed in Jetson TX2 to acquire depth information in real time, and we proposed a people counting method using acquired depth information. Experimental results show that the proposed method had a counting accuracy of 98.6% and operating in real time.

• Annual Conference of KIPS
• /
• 2012.11a
• /
• pp.1086-1087
• /
• 2012

GPU를 이용하여 병렬 처리 연산을 하는 연구는 활발히 진행되고 있고, 이미 많은 곳에서 사용되고 있다. 본 논문에서는 엔비디아에서 개발한 CUDA를 사용하여 DES 알고리즘을 고속으로 구현하기 위해 CUDA overlapping을 이용했다. 이것은 GPU 에서 연산을 하는 동시에 연산 결과를 바로 Host로 보내어 연산시간과 전송시간을 Overlap 하여 시간을 더 단축 하도록 하는 구현방법이다. 그 결과 Overlap 하기 전보다 약 30%의 성능향상을 확인 할 수 있었다. 향후 DES 뿐만 아니라 3DES, AES, SEED 등 여러 암호화 알고리즘들도 적용할 예정이다.

• Journal of information and communication convergence engineering
• /
• v.20 no.2
• /
• pp.96-102
• /
• 2022

We aim to build predictive models for Airbnb's prices using a GPU-accelerated RAPIDS in a big data cluster. The Airbnb Listings datasets are used for the predictive analysis. Several machine-learning algorithms have been adopted to build models that predict the price of Airbnb listings. We compare the results of traditional and big data approaches to machine learning for price prediction and discuss the performance of the models. We built big data models using Databricks Spark Cluster, a distributed parallel computing system. Furthermore, we implemented models using multiple GPUs using RAPIDS in the spark cluster. The model was developed using the XGBoost algorithm, whereas other models were developed using traditional central processing unit (CPU)-based algorithms. This study compared all models in terms of accuracy metrics and computing time. We observed that the XGBoost model with RAPIDS using GPUs had the highest accuracy and computing time.

• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.104-123
• /
• 2022

Style transfer based on neural network provides very high quality results by reflecting the high level structural characteristics of images, and thereby has recently attracted great attention. This paper deals with the problem of resolution limitation due to GPU memory in performing such neural style transfer. We can expect that the gradient operation for style transfer based on partial image, with the aid of the fixed size of receptive field, can produce the same result as the gradient operation using the entire image. Based on this idea, each component of the style transfer loss function is analyzed in this paper to obtain the necessary conditions for partitioning and padding, and to identify, among the information required for gradient calculation, the one that depends on the entire input. By structuring such information for using it as auxiliary constant input for partition-based gradient calculation, this paper develops a recursive algorithm for super high-resolution image style transfer. Since the proposed method performs style transfer by partitioning input image into the size that a GPU can handle, it can perform style transfer without the limit of the input image resolution accompanied by the GPU memory size. With the aid of such super high-resolution support, the proposed method can provide a unique style characteristics of detailed area which can only be appreciated in super high-resolution style transfer.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.9
• /
• pp.21-31
• /
• 2014

Singular value decomposition (SVD) has been widely used to identify unique features from a data set in various fields. However, a complex matrix calculation of SVD requires tremendous computation time. This paper improves the performance of a representative one-sided block Jacoby algorithm using a two-dimensional (2D) multi-core system. In addition, this paper explores an optimal multi-core system by varying the number of processing elements in the 2D multi-core system with the same 400MHz clock frequency and TSMC 28nm technology for each matrix-based one-sided block Jacoby algorithm ($128{\times}128$, $64{\times}64$, $32{\times}32$, $16{\times}16$). Moreover, this paper demonstrates the potential of the 2D multi-core system for the one-sided block Jacoby algorithm by comparing the performance of the multi-core system with a commercial high-performance graphics processing unit (GPU).

• Journal of IKEEE
• /
• v.25 no.4
• /
• pp.634-643
• /
• 2021

In this paper, an FPGA-based acceleration scheme of range Doppler algorithm (RDA) is proposed for the real time synthetic aperture radar (SAR) imaging. Hardware architectures of matched filter based on systolic array architecture and a high speed sinc interpolator to compensate range cell migration (RCM) are presented. In addition, the proposed hardware was implemented and accelerated on Xilinx Alveo FPGA. Experimental results for 4096×4096-size SAR imaging showed that FPGA-based implementation achieves 2 times acceleration compared to GPU-based design. It was also confirmed the proposed design can be implemented with 60,247 CLB LUTs, 103,728 CLB registers, 20 block RAM tiles and 592 DPSs at the operating frequency of 312 MHz.

• KIPS Transactions on Computer and Communication Systems
• /
• v.2 no.2
• /
• pp.67-74
• /
• 2013

Approximate string matching problems have been studied in diverse fields. Recently, fast approximate string matching algorithms are being used to reduce the time and costs for the next generation sequencing. To measure the amounts of errors between two strings, we use a distance function such as the edit distance. Given two strings X(|X| = m) and Y(|Y| = n) over an alphabet ${\Sigma}$, the edit distance between X and Y is the minimum number of edit operations to convert X into Y. The edit distance between X and Y can be computed using the well-known dynamic programming technique in O(mn) time and space. The edit distance also can be computed using the Four-Russians' algorithm whose preprocessing step runs in $O((3{\mid}{\Sigma}{\mid})^{2t}t^2)$ time and $O((3{\mid}{\Sigma}{\mid})^{2t}t)$ space and the computation step runs in O(mn/t) time and O(mn) space where t represents the size of the block. In this paper, we present a parallelized version of the computation step of the Four-Russians' algorithm. Our algorithm computes the edit distance between X and Y in O(m+n) time using m/t threads. Then we implemented both the sequential version and our parallelized version of the Four-Russians' algorithm using CUDA to compare the execution times. When t = 1 and t = 2, our algorithm runs about 10 times and 3 times faster than the sequential algorithm, respectively.

• Journal of the Korea Computer Graphics Society
• /
• v.15 no.1
• /
• pp.17-23
• /
• 2009

We present a simple new idea to improve the quality of exemplar based texture synthesis using multiple rotated input exemplars. Our algorithm successfully obtain rotational synthesis feature variations and manages to reduce the artifacts in the results, especially patch seams due to the structure of the exemplars provided which have been inappropriate for previous neighborhood matching synthesis algorithms. Our algorithm is parallel in nature, thus it is possible to implement our algorithm using GPU or multi-core CPU to accelerate synthesis process.

• Korean Journal of Computational Design and Engineering
• /
• v.19 no.3
• /
• pp.294-304
• /
• 2014

This paper presents a real-time rendering algorithm of large-scale geometric data using GLSL (OpenGL shading language). It details the VAO (vertex array object) and VBO(vertex buffer object) to be used for up-loading the large-scale point clouds and polygon meshes to a graphic video memory, and describes the shader program composed by a vertex shader and a fragment shader, which manipulates those large-scale data to be rendered by GPU. In addition, we explain the global rendering procedure that creates and runs the shader program with the VAO and VBO. Finally, a rendering performance will be measured with application examples, from which it will be demonstrated that the proposed algorithm enables a real-time rendering of large amount of geometric data, almost impossible to carry out by previous techniques.