The Earth Simulator (ES), developed under the Japanese government’s initiative “Earth Simulator project”, is a highly parallel vector supercomputer system. In this paper, an overview of ES, its architectural features, hardware technology and the result of performance evaluation are described.
In May 2002, the ES was acknowledged to be the most powerful computer in the world: 35.86 teraflop/s for the LINPACK HPC benchmark and 26.58 teraflop/s for an atmospheric general circulation code (AFES). Such a remarkable performance may be attributed to the following three architectural features; vector processor, shared-memory and high-bandwidth non-blocking interconnection crossbar network.
The ES consists of 640 processor nodes (PN) and an interconnection network (IN), which are housed in 320 PN cabinets and 65 IN cabinets. The ES is installed in a specially designed building, 65 m long, 50 m wide and 17 m high. In order to accomplish this advanced system, many kinds of hardware technologies have been developed, such as a high-density and high-frequency LSI, a high-frequency signal transmission, a high-density packaging, and a high-efficiency cooling and power supply system with low noise so as to reduce whole volume of the ES and total power consumption.
For highly parallel processing, a special synchronization means connecting all nodes, Global Barrier Counter (GBC), has been introduced. 相似文献
The effect of nonuniform magnetic field on the linear and nonlinear wave propagation phenomena in two-phase pipe flow of magnetic fluid is investigated theoretically to realize the effective energy conversion system using boiling two-phase flow of magnetic fluid. Firstly, the governing equations of two-phase flow based on the unsteady thermal nonequilibrium two-fluid model are presented and the linear void wave propagation phenomena in boiling two-phase flow are numerically analyzed by using the finite volume method. Next, the nonlinear pressure wave propagation in gas-liquid two-phase flow is numerically analyzed by using the finite different method. According to these theoretical studies on the wave propagation phenomena in two-phase flow of magnetic fluid, it seems to be a reasonable proposal that the precise control of the wave propagation in two-phase flow is possible by effective use of the magnetic force. 相似文献
The recent increase in the demand for power in urban areas has caused changes in the way power is supplied. Underground distribution systems have been introduced for convenience in urban areas and to prevent injuries in case of a crisis. As a result, many new types of power cables are being used. One type, XPLE cables (also called CV cables) have been rapidly taking the place of conventional cables such as OF or SL cables since the 1960s because of their relatively easy installation and maintainability. However, distribution cables used under severe environmental conditions deteriorate rapidly (due mainly to water treeing), and some cables experience insulation breakdown. Therefore, many diagnostic methods have been developed, though they have not been very effective. By combining methods, a high reliability can be achieved in comparison to the use of a single method. However, processing the data is difficult for on-site personnel, so many experts are required. But there is a lack of qualified experts, hence we have turned to the development of expert systems. 相似文献
Skeletal form of KnbO3 crystals growing in Li2B4O7 solvent was in-situ observed at 900℃ and it was found that shallow depression started to develop on the surface of KnbO3 crystals when the crystal size exceeded several micron,typically 7 micron.Based on the quantitative criterion derived by Chernov,the estimated critical size of KNbO3 crystals was 1 micron,which was consistent with the experimental measurement.The kinetic coefficients,Kcorner and Kcr,in the criterion were experimentally obtained in the diffusive-convective and diffusive-advective flow states respectively. 相似文献
In this paper, bulk-Si metal–oxide–semiconductor field effect transistors (MOSFETs) are fabricated using the catalytic chemical vapor deposition (Cat-CVD) method as an alternative technology to the conventional high-temperature thermal chemical vapor deposition. Particularly, formation of low-resistivity phosphorus (P)-doped poly-Si films is attempted by using Cat-CVD-deposited amorphous silicon (a-Si) films and successive rapid thermal annealing (RTA) of them. Even after RTA processes, neither peeling nor bubbling are observed, since hydrogen contents in Cat-CVD a-Si films can be as low as 1.1%. Both the crystallization and low resistivity of 0.004 Ω·cm are realized by RTA at 1000 °C for only 5 s. It is also revealed that Cat-CVD SiNx films prepared at 250 °C show excellent oxidation resistance, when the thickness of films is larger than approximately 10 nm for wet O2 oxidation at 1100 °C. It is found that the thickness required to stop oxygen penetration is equivalent to that for thermal CVD SiNx prepared at 750 °C. Finally, complementary MOSFETs (CMOSs) of single-crystalline Si were fabricated by using Cat-CVD poly-Si for gate electrodes and SiNx films for masks of local oxidation of silicon (LOCOS). At 3.3 V operation, less than 1.0 pA μm−1 of OFF leakage current and ON/OFF ratio of 107–108 are realized, i.e. the devices can operate similarly to conventional thermal CVD process. 相似文献
In this paper, we propose a framework for human action analysis from video footage. A video action sequence in our perspective is a dynamic structure of sparse local spatial–temporal patches termed action elements, so the problems of action analysis in video are carried out here based on the set of local characteristics as well as global shape of a prescribed action. We first detect a set of action elements that are the most compact entities of an action, then we extend the idea of Implicit Shape Model to space time, in order to properly integrate the spatial and temporal properties of these action elements. In particular, we consider two different recipes to construct action elements: one is to use a Sparse Bayesian Feature Classifier to choose action elements from all detected Spatial Temporal Interest Points, and is termed discriminative action elements. The other one detects affine invariant local features from the holistic Motion History Images, and picks up action elements according to their compactness scores, and is called generative action elements. Action elements detected from either way are then used to construct a voting space based on their local feature representations as well as their global configuration constraints. Our approach is evaluated in the two main contexts of current human action analysis challenges, action retrieval and action classification. Comprehensive experimental results show that our proposed framework marginally outperforms all existing state-of-the-arts techniques on a range of different datasets. 相似文献