• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.05a
• /
• pp.1651-1654
• /
• 2004

파이프라인과 슈퍼스칼라 방식이 일반화된 시스템 구조 하에서, 분기 명령어는 시스템 전체적인 성능에 중요한 영향을 미친다. 특히 분기 예측이 실패했을 경우, 잘못된 분기 예측으로 인한 페널티가 발생한다는 점에서 분기 예측의 정확도에 대한 중요성은 크다고 할 수 있다. 본 논문에서는 분기 예측의 정확도를 높이기 위해서, 분기 예측과 관련된 신경망을 구축하여 이를 통해 분기 예측에 필요한 각 요소별 가중치의 변화를 분석하고, 이를 분기 예측에 새롭게 반영하고자 한다. 본 논문에서는 이를 위해 실행 구동 방식의 시뮬레이터인 SimpleScalar를 통하여 모의 실험을 수행하였으며, 실험 결과 본 논문에서 제시한 새로운 기법이 기존의 일반적인 이단계 적응형 분기 예측 기법이나 gshare 기법에 비하여 더 우수한 결과를 보였다.

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.12
• /
• pp.653-662
• /
• 2016

This paper describes a test case generation algorithm for Simulink/Stateflow models based on the Rapidly exploring Random Tree (RRT) algorithm that has been successfully applied to path finding. An important factor influencing the performance of the RRT algorithm is the metric used for calculating the distance between the nodes in the RRT space. Since a test case for a Simulink/Stateflow (SL/SF) model is an input sequence to check a specific condition (called a test target in this paper) at a specific status of the model, it is necessary to drive the model to the status before checking the condition. A status maps to a node of the RRT. It is usually necessary to check various conditions at a specific status. For example, when the specific status represents an SL/SF model state from which multiple transitions are made, we must check multiple conditions to measure the transition coverage. We propose a unique distance calculation metric, based on the observation that the test targets are gathered around some specific status such as an SL/SF state, named key nodes in this paper. The proposed metric increases the probability that an RRT is extended from key nodes by imposing penalties to non-key nodes. A test case generation algorithm utilizing the proposed metric is proposed. Three models of Electrical Control Units (ECUs) embedded in a commercial vehicle are used for the performance evaluation. The performances are evaluated in terms of penalties and compared with those of the algorithm using a typical RRT algorithm.

• Journal of the Mineralogical Society of Korea
• /
• v.30 no.3
• /
• pp.103-112
• /
• 2017

In order to transform arsenopyrite into pyrrhotite and to decrease As content by less than 2,000 mg/kg, pulp sample and non-magnetic pulp sample were heated in a microwave oven at different heating times and were separated through wet-magnetic separation. As the microwave heating time increased, the phase of pyrrhotite was extended to become arsenopyrite entirely. The melting pores and micro-cracks occurred on the pyrrhotite due to hot spot phenomenon with microwave heating. The heated raw pulp sample (As content : 19,970.13 mg/kg) and non-magnetic pulp sample (As content : 19,970.13 mg/kg) which were heated in a microwave oven for 10 minutes were separated through wet-magnetic separation and magnetic fraction containing less than 2,000 mg/kg of As content was recovered only from the heated sample of magnetic separation. It was discovered that for the sulfide complex ore with As penalty imposed on, if microwave heating and wet-magnetic separation are effectively utilized, magnetic fraction. We expect to be able to obtain ore minerals with an arsenic content below the penalty charge.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.42 no.1
• /
• pp.9-26
• /
• 2005

In the pursuit of ever higher levels of performance, recent computer systems have made use of deep pipeline, dynamic scheduling and multi-issue superscalar processor technologies. In this situations, branch prediction schemes are an essential part of modem microarchitectures because the penalty for a branch misprediction increases as pipelines deepen and the number of instructions issued per cycle increases. In this paper, we propose a novel branch prediction scheme, direction-gshare(d-gshare), to improve the prediction accuracy. At first, we model a neural network with the components that possibly affect the branch prediction accuracy, and analyze the variation of their weights based on the neural network information. Then, we newly add the component that has a high weight value to an original gshare scheme. We simulate our branch prediction scheme using Simple Scalar, a powerful event-driven simulator, and analyze the simulation results. Our results show that, compared to bimodal, two-level adaptive and gshare predictor, direction-gshare predictor(d-gshare. 3) outperforms, without additional hardware costs, by up to 4.1% and 1.5% in average for the default mont of embedded direction, and 11.8% in maximum and 3.7% in average for the optimal one.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.21 no.8
• /
• pp.81-86
• /
• 2007

This paper proposes the scheme to improve the current direct load control(DLC) program by inspecting the problems of the DLC program. In order to increase the number of the customers participating DLC program, a reduction of the base incentive and an increase of the direct load control incentive are suggested based on the interruption cost of electric power considering the characteristics of load types and the introduction of demand side bidding is recommended. Secondly, the standards of power system operations is required to control DLC program efficiently for the penalty, interruption times, the number of interrupting loads, notice time for the load interruption and the periods of the DLC program contract.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.7 no.3
• /
• pp.567-571
• /
• 2012

To design the long-haul optical communication system, we need to decide the type of optical fiber and optical amplifier, span length of optical amplifier, dispersion compensation method, optical signal power, etc. Therefore, we need to predict the performance of optical communication system when we change one of the system parameters. In this paper, we investigate the method of predicting the maximum transmission length of the designed optical communication system and finding the optimum optical signal power to obtain the maximum transmission length.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.11B
• /
• pp.1607-1611
• /
• 2001

The dependence of the performance of 40-Gb/s optical transmission using short pulses on the fiber chromatic dispersion is numerically studied. When very short pulses are used, the wide spectrum of the optical signal and the chromatic dispersion of the fiber interact in such a way that results in the reduction of nonlinear impairments of the transmission performance. The degree of this reduction is determined by the combined effects of chromatic dispersion of the fiber and the strength of the optical signal and the transmission distance. When 3ps-long pulses were used for the transmission, the eye-closure penalty was highest with the dispersion D=4ps/nm/km.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2000.04a
• /
• pp.544-549
• /
• 2000

슈퍼스칼라 프로세서의 성능을 향상시키기 위해서는 데이터 종속성에 의한 장애를 제거해야 한다. 최근 여러 논문들은 이러한 데이터 종속성을 제거하기 위해서 명령의 결과 값을 예상하는 메커니즘이 연구되고 있다. 결과 값 예상 메커니즘 중 여러 예측기를 하이브리드해서 사용하는 방법은 각각 하나의 예측기만을 사용하는 방법보다 더 좋은 성능을 얻을 수 있다. 그러나 종전의 하이브리드 예측기는 명령어를 중복해서 저장하여 많은 하드웨어 크기를 요구한다. 본 논문에서는 여러 예측기의 장점을 이용하여 높은 성능을 얻을 수 있는 새로운 하이브리드 예측 메커니즘을 제안한다. 또한 예상하기 어려운 명령어를 동적으로 찾아내어 예상하지 않음으로서 잘못 예상한 misprediction 페널티를 줄이고 예상 정확도를 높인다. 시뮬레이션 결과 SPECint95 벤치마크 프로그램에 대해 제안한 하이브리드 예측기에서 예측율은 평균 79%에서 90%로 향상하였고, misprediction rate는 평균 12%에서 2%로 낮추었다

• Korean Journal of Optics and Photonics
• /
• v.12 no.2
• /
• pp.121-128
• /
• 2001

We examine the performance of a Raman amplifier as a function of fiber structure with respect to amplifier gain and double Rayleigh crosstalk penalty. Variations on fiber core radius or index affect both the Raman gain efficiency and Rayleigh backscattering. Contrary to the common concept, the penalty from the doubly amplified Rayleigh scattering could exceed the benefits of higher gain efficiency of small effective area fibers. Appropriate fiber designing parameters are required to increase Raman amplifier efficiency without system penalties. lties.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.25 no.6
• /
• pp.469-476
• /
• 2012

In this paper, a finite element domain decomposition method using local and mixed Lagrange multipliers for a large scal structural analysis is presented. The proposed algorithms use local and mixed Lagrange multipliers to improve computational efficiency. In the original FETI method, classical Lagrange multiplier technique was used. In the dual-primal FETI method, the interface nodes are used at the corner nodes of each sub-domain. On the other hand, the proposed FETI-local analysis adopts localized Lagrange multipliers and the proposed FETI-mixed analysis uses both global and local Lagrange multipliers. The numerical analysis results by the proposed algorithms are compared with those obtained by dual-primal FETI method.