scholarly journals A reconfigurable computer system based on FPGAs with liquid cooling

Author(s):  
И.И. Левин ◽  
А.И. Дордопуло ◽  
Ю.И. Доронченко ◽  
М.К. Раскладкин ◽  
А.М. Федоров

Статья посвящена проблемам построения перспективных реконфигурируемых вычислительных систем c жидкостным охлаждением для программируемых логических интегральных схем семейства Xilinx Virtex UltraScale. Рассматриваются архитектура, компоновка и сравнительные технические характеристики систем погружного жидкостного охлаждения. Приводятся результаты расчетов, макетирования и экспериментальной проверки основных технических решений созданного вычислительного модуля нового поколения для построения высокопроизводительных вычислительных систем с жидкостным охлаждением с производительностью 1 Пфлопс в стандартном вычислительном шкафу высотой 47U при потребляемой мощности 150 кВт. Реконфигурируемая вычислительная система с жидкостным охлаждением обеспечивает существенное преимущество по таким технико-экономическим параметрам, как реальная и удельная производительность, энергоэффективность, массогабаритные характеристики и другим по сравнению с аналогичными системами. The paper deals with problems of design of promising reconfigurable computer systems with liquid cooling for Xilinx Virtex UltraScale FPGAs. Architecture, placement and comparative technical parameters of systems with immersion liquid cooling are considered. Results of design, prototyping and experimental testing of the principal technical solutions of the designed computational module of the next generation are discussed. The computational module is intended for the creation of high-performance computer systems with liquid cooling with performance of 1 PFlops in a standard 47U computer rack with power of 150 kWatt. The reconfigurable computer system with liquid cooling provides considerable advantage in such technical and economical parameters as the real performance and the specific performance, power efficiency, mass and dimension parameters, etc. in comparison with similar systems.

1993 ◽  
Vol 26 (3) ◽  
pp. 76-79
Author(s):  
R Clarke ◽  
F Humphries

Nuclear Electric have commissioned Ferranti International to develop a replacement system for their on-line computer systems in earlier AGR power stations. This system represents a step forward both in software and performance and takes advantage of investment made for military use in the Ada Language, CASE tools and a high-performance real-time relational database.


Author(s):  
Sukhvinder Kang ◽  
David Miller ◽  
John Cennamo

The power dissipation levels in high performance personal computers continue to increase rapidly while the silicon die temperature requirements remain unchanged or have been lowered. Advanced air cooling solutions for the major heat sources such as CPU and GPU modules use heat pipes and high flow rate fans to manage the heat load at the expense of significant increases in the sound power emitted by the computer system. Closed loop liquid cooling systems offer an excellent means to efficiently meet the combined challenges of high heat loads, low thermal resistance, and low noise while easily managing die level heat fluxes in excess of 500 W/cm2. This paper describes the design and attributes of an advanced liquid cooling system that can cool single or multiple heat sources within the computer system. The cooling system described use copper cold plates with meso scale channels to pick up heat from CPU and GPU type heat sources and highly efficient liquid-to-air heat exchangers with flat copper tubes and plain fins to transfer the heat to air by forced convection. A water based coolant is used for high thermal performance and additives are used to provide burst protection to the cooling system at temperatures down to −40 °C and corrosion protection to critical components. A highly reliable compact pump is used to circulate the fluid in a closed loop. The overall system is integrated using assembly methods and materials that enable very low fluid permeation for long life.


Author(s):  
D. A. Dolgov ◽  
K. S. Nozdrin

The paper discusses the practical issues of creation cluster-type computing equipment and switching equipment, which are based on Russian technologies and components developed for the segment of high-performance teraflop-class servers. A number of technical solutions are proposed, aimed at forming in the shortest possible time with minimal costs using a limited set of components (clusters), a model range of cluster-type computing equipment. The latter should ensure the creation of technical means of automation systems that have a performance parity with technical means of foreign production, as well as surpass it in a number of important operational and technical parameters, including the duration of the products and technical means life cycle, failure stability and external influencing factors durability.


2020 ◽  
Author(s):  
S Mukhtar Ayubi Simatupang

AbstrakBrainware adalah istilah yang digunakan untuk manusia yang digunakan untuk manusia yang berhubungan dengan sistem komputer. Manusia merupakan suatu elemen dari sistem komputer yang merancang bagaimana suatu mesin dapat bekerja sesuai dengan hasil yang diinginkan. Tingkatan brainware terdiri atas system analyst, programmer, administrator, dan operator. Bagian bagian brainware terdiri atas operator komputer, teknisi, trainer, konsultan, project manager, programmer, grapic designer, spesialis jaringan, database administrator, dan system analitis. Kata Kunci : Brainware (Perangkat Sumber Daya Manusia)AbstractBrainware is a term used for humans that is used for humans related to computer systems. Humans are an element of a computer system that designs how a machine can work in accordance with the desired results. The brainware level consists of system analysts, programmers, administrators, and operators. The brainware section consists of computer operators, technicians, trainers, consultants, project managers, programmers, grapic designers, network specialists, database administrators, and system analytics.Keywords: Brainware (Human Resources Tool)


Author(s):  
Joshua A. Kroll

This chapter addresses the relationship between AI systems and the concept of accountability. To understand accountability in the context of AI systems, one must begin by examining the various ways the term is used and the variety of concepts to which it is meant to refer. Accountability is often associated with transparency, the principle that systems and processes should be accessible to those affected through an understanding of their structure or function. For a computer system, this often means disclosure about the system’s existence, nature, and scope; scrutiny of its underlying data and reasoning approaches; and connection of the operative rules implemented by the system to the governing norms of its context. Transparency is a useful tool in the governance of computer systems, but only insofar as it serves accountability. There are other mechanisms available for building computer systems that support accountability of their creators and operators. Ultimately, accountability requires establishing answerability relationships that serve the interests of those affected by AI systems.


1985 ◽  
Vol 10 (2) ◽  
pp. 79-86 ◽  
Author(s):  
Anne Costigan ◽  
Frances E. Wood ◽  
David Bawden

A comparative evaluation of three implementations of a large databank, the NIOSH Registry of Toxic Effects of Chem ical Substances, has been carried out. The three implementa tions are: a printed index, a text searching computer system, and a computerised chemical databank system, with substruc ture searching facilities. Seven test queries were used, with the aim of drawing conclusions of general relevance to chemical databank searching. The computer systems were shown to have advantages over printed indexes for several of the queries, including those involving an element of browsing. Substructure search facilities were especially advantageous. Aspects of indexing of data present, and the criteria for inclusion of types of data, were also highlighted.


1988 ◽  
Vol 13 (1) ◽  
pp. 25-32 ◽  
Author(s):  
T V Seshadri ◽  
N Kinra

Who, in the organization, buys the computer system? How are various departments involved in the organizational decision process? T V Seshadri and N Kinra analyse the decision processes of 30 organizations that had bought a computer system—mini, mainframe, or macro. Based on a questionnaire study and factor analysis, the authors conclude that the EDP department and Board of Directors are critical in the buying grids of the purchasing organizations. They draw implications of their findings for managers marketing computer systems.


2014 ◽  
Vol 555 ◽  
pp. 665-672
Author(s):  
Adriana Şerban Târgoveţ ◽  
Dragoş Ionescu-Bondoc

During swimming competitions starting from the block-start platform, a potential hypothesis was noticed, through an active multimodal process, which can make the swimming start efficient, especially in the case of sprint races, by improving the propulsive force parameters of the inferior limbs. The swimming start research from interdisciplinary perspective: biomechanical, kinematic, informational and statistical can consolidate and improve the specific technique in accordance with the abilities and psycho-motor qualities of the swimmers. The present study is based on an experiment where the spatial-temporal and kinematic parameters were processed with the help of a Dartfish program. The evolution of parameters is researched as a result of a motor training program with the purpose to increase the propulsive force off the block-start. The improvement of spatial-temporal parameters influences the performance and evolution of technical parameters. Initial and final recordings were made on an MLD Station Evo5 and MLD software MuskelLeistungs Diagnose, fromSPSport, SPSportdiagnosegeräte, in order to evaluate the force, the power and the propulsive force. The argumentation of the experimental research is based on the statement: “the spatial characteristics of the motions and actions can be studied for themselves as parameters, characteristics or as a reference method for defining other characteristics, such as velocity or push-off force [1]. The main purpose of this study is to identify the influences of the specific start training upon the force improvement and kick power of the support foot from the block-start, during the classic track start. Given that the track start technique is the same as the one of the kick start executed from the international block-start of Omega, OSB11, developed in 2009, one assumes that the improvement of the classic track start leads by default to the improvement of the kick start. Lack of training to practice this type of start leads to deficient use during competitions, thus obtaining poor performances. There are no kick block-starts in Romania in order to train high performance athletes participating in international competitions and as a consequence, poor results are obtained at sprint races. One assumes that training for this type of start can be succesfully made only from a block-start similar to the kick one. The block-start model adapted by us under the same biomechanical conditions as the ones of the international kick start, is called “athletic kick”. The training specific to the kick start is carried out only with the optimum use of the kick block-start, the reasons for this being presented by N, Houel, A. Charliac, JL.Rey, Phellardin the paper: “How the swimmer could improve his track start using new Olympic plot” [2].


Author(s):  
Ilya Levin ◽  
Alexey Dordopulo ◽  
Alexander Fedorov ◽  
Yuriy Doronchenko

Sign in / Sign up

Export Citation Format

Share Document