• Journal of the Korea Society of Computer and Information
• /
• v.16 no.9
• /
• pp.1-10
• /
• 2011

Unfortunately, in current microprocessors, increasing the frequency causes increased power consumption and reduced reliability whereas it improves the performance. To overcome the power and thermal problems in the singlecore processors, multicore processors has been widely used. For 2D multicore processors, interconnection is regarded as one of the major constraints in performance and power efficiency. To reduce the performance degradation and the power consumption in 2D multicore processors, 3D integrated design technique has been studied by many researchers. Compared to 2D multicore processors, 3D multicore processors get the benefits of performance improvement and reduced power consumption by reducing the wire length significantly. However, 3D multicore processors have serious thermal problems due to high power density, resulting in reliability degradation. Detailed thermal analysis for multicore processors can be useful in designing thermal-aware processors. In this paper, we analyze the impact of workload distribution, distance to the heat sink, and number of stacked dies on the processor temperature. We also analyze the effects of the temperature on overall system performance. Especially, this paper presents the guideline for thermal-aware multicore processor design by analyzing the thermal problems in 2D multicore processors and 3D multicore processors.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.44 no.1
• /
• pp.74-84
• /
• 2007

This paper presents a new dual-loop Delay Locked Loop(DLL) to expand the delay lock range of a conventional DLL. The proposed dual-loop DLL contains a Coarse_loop and a Fine_loop, and its operation utilizes one of the loops selected by comparing the initial time-difference among the reference clock and 2 internal clocks. The 2 internal clock signals are taken, respectively, at the midpoint and endpoint of a VCDL and thus are $180^{\circ}$ separated in phase. When the proposed DLL is out of the conventional lock range, the Coarse_loop is selected to push the DLL in the conventional lock range and then the Fine_loop is used to complete the locking process. Therefore, the proposed DLL is always stably locked in unless it is harmonically false-locked. Since the VCDL employed in the proposed DLL needs two control voltages to adjust the delay time, it uses TG-based inverters, instead of conventional, multi-stacked, current-starved inverters, to compose the delay line. The new VCDL provides a wider delay range than a conventional VCDL In overall, the proposed DLL demonstrates a more than 2 times wider lock range than a conventional DLL. The proposed DLL circuits have been designed, simulated and proved using 0.18um, 1.8V TSMC CMOS library and its operation frequency range is 100MHz${\sim}$1GHz. Finally, the maximum phase error of the DLL locked in at 1GHz is less than 11.2ps showing a high resolution and the simulated power consumption is 11.5mW.

• Progress in Medical Physics
• /
• v.29 no.2
• /
• pp.59-65
• /
• 2018

This paper describes the clinical use of the dose verification of multileaf collimator (MLC)-based CyberKnife plans by combining the Octavius 1000SRS detector and water-equivalent RW3 slab phantom. The slab phantom consists of 14 plates, each with a thickness of 10 mm. One plate was modified to support tracking by inserting 14 custom-made fiducials on surface holes positioned at the outer region of $10{\times}10cm^2$. The fiducial-inserted plate was placed on the 1000SRS detector and three plates were additionally stacked up to build the reference depth. Below the detector, 10 plates were placed to avoid longer delivery times caused by proximity detection program alerts. The cross-calibration factor prior to phantom delivery was obtained by performing with 200 monitor units (MU) on the field size of $95{\times}92.5mm^2$. After irradiation, the measured dose distribution of the coronal plane was compared with the dose distribution calculated by the MultiPlan treatment planning system. The results were assessed by comparing the absolute dose at the center point of 1000SRS and the 3-D Gamma (${\gamma}$) index using 220 patient-specific quality assurance (QA). The discrepancy between measured and calculated doses at the center point of 1000SRS detector ranged from -3.9% to 8.2%. In the dosimetric comparison using 3-D ${\gamma}$-function (3%/3 mm criteria), the mean passing rates with ${\gamma}$-parameter ${\leq}1$ were $97.4%{\pm}2.4%$. The combination of the 1000SRS detector and RW3 slab phantom can be utilized for dosimetry validation of patient-specific QA in the CyberKnife MLC system, which made it possible to measure absolute dose distributions regardless of tracking mode.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2011.08a
• /
• pp.293-294
• /
• 2011

It is known that semiconductor quantum-dot (QD) heterostructures have superior zero-dimensional quantum confinement, and they have been successfully applied to semiconductor laser diodes (QDLDs) for optical communication and infrared photodetectors (QDIPs) for thermal images [1]. The self-assembled QDs are normally formed at Stranski-Krastanov (S-K) growth mode utilizing the accumulated strain due to lattice-mismatch existing at heterointerfaces between QDs and cap layers. In order to increase the areal density and the number of stacks of QDs, recently, sub-monolayer (SML)-thick QDs (SQDs) with reduced strain were tried by equivalent thicknesses thinner than a wetting layer (WL) existing in conventional QDs (CQDs) by S-K mode. Despite that it is very different from CQDs with a well-defined WL, the SQD structure has been successfully applied to QDIP[2]. In this study, optical characteristics are investigated by using photoluminescence (PL) spectra taken from self-assembled InAs/GaAs QDs whose coverage are changing from submonolayer to a few monolayers. The QD structures were grown by using molecular beam epitaxy (MBE) on semi-insulating GaAs (100) substrates, and formed at a substrate temperature of 480$^{\circ}C$ followed by covering GaAs cap layer at 590$^{\circ}C$. We prepared six 10-period-stacked QD samples with different InAs coverages and thicknesses of GaAs spacer layers. In the QD coverage below WL thickness (~1.7 ML), the majority of SQDs with no WL coexisted with a small amount of CQDs with a WL, and multi-peak spectra changed to a single peak profile. A transition from SQDs to CQDs was found before and after a WL formation, and the sublevel of SQDs peaking at (1.32${\pm}$0.1) eV was much closer to the GaAs bandedge than that of CQDs (~1.2 eV). These revealed that QDs with no WL could be formed by near-ML coverage in InAs/GaAs system, and single-mode SQDs could be achieved by 1.5 ML just below WL that a strain field was entirely uniform.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.11
• /
• pp.12-20
• /
• 2013

Even though extensive research results have been applied to wireless cellular systems to improve their capacity and coverage, severe performance degradation experienced in cell boundary areas still remains as a major limiting factor to prohibit further improvement of user equipment (UE) throughput. In the Long Term Evolution-Advanced (LTE-A) standard of the Third Generation Partnership Project (3GPP), Some advanced techniques have been introduced to overcome this "cell-edge problem", including coordinated multipoint transmission and reception (CoMP) and inter-cell interference coordination (ICIC). In this paper, we propose yet another strategy to improve the performance of low-tier UEs by using the concept of multiple beam direction patterns (BDPs). Such multiple BDPs can be implemented using multi-layer antenna arrays stacked vertically at base station (BS) sites to transmit signals in different main beam directions. In comparison to conventional three-sector antennas with a fixed beam pattern, the proposed methods makes signal transmission in a rotational fashion to significantly enhance the reception quality of UEs located near sector (or cell) edge areas, preventing the situation where certain UEs are marginally covered by the BS for the whole transmission time. Performance evaluation results show that the proposed scheme outperforms the conventional three-sector transmission by 171% in low 5% UEs in terms of the UE throughput.

• Geophysics and Geophysical Exploration
• /
• v.17 no.2
• /
• pp.74-87
• /
• 2014

Analysis of multi-channel seismic reflection profiles collected from the Ulleung Basin reveals that the Quaternary sequence consists of four stratigraphic units separated by erosional unconformities. Individual stratigraphic unit includes eighteen mass transport deposits which are variable in geometric characteristics and spatial distribution. Each mass transport deposit on the seismic profile is acoustically characterized by chaotic or transparent seismic facies, and shows wedge or lens-shaped external geometry. The mass transport deposits, which comprise a succession of stacked wedges, mainly occur on the southern slope, and their thickness gradually decreases toward the basin plain. The time structure map of erosional unconformities shows that a tectonic-induced structural high and troughs toward the northwest and northeast are developed at the central part of the basin. Based on the isochron map, the mass transport deposits, originated from southern part of the study area, transported to the basin plain and can be divided into two groups by the structural high. Consequently, the mass transport deposits within the Quaternary sequence in the Ulleung Basin are largely controlled by the large amounts of sediment supply, dissociation of gas hydrate during the lowstands, and central structural high.