Binary arithmetic for discretely variable word length in a serial computer

ACM '58 Pub Date : 1958-06-11 DOI:10.1145/610937.610961
P. Ercoli, R. Vacca
{"title":"Binary arithmetic for discretely variable word length in a serial computer","authors":"P. Ercoli, R. Vacca","doi":"10.1145/610937.610961","DOIUrl":null,"url":null,"abstract":"The serial computer FINAC, a Ferranti l~lark I machine installed at the Consiglio Nazionale delle Ricerche in Rome, has been frequently used for the solution of systems of algebraic linear equations and in general of problems of linear algebra. As the size and the number of ill-conditioned systems to be solved had a tendency to increase the problem arose of performing the computation with a greater accuracy, in order not to lose all of the approximation during the execution of the programs. It was decided that efficient programs could be written using only the machine fixed point instructions, that is, without pro-grassing the machine for floating point operation, if the maximum length of 40 bits of the numbers employed could be doubled. This could be done by programme, but in this case the machine would be slowed down to a considerable extent. Two different possibilities were open : that of increasing the speed of multiplication on numbers of 40 bits by doubling the parallelism of the series-parallel multiplication circuit ,ioe° using the multiplier's digits in groups of two and staticising the whole multiplier in a single step, and that of installing an arithmetic unit operating on 40 or 80 bits words. In the first case fixed-point 80-bits operation would always have been possible by programming every single arithmetic operation, in the second case 80-bit instructions would have been available in the machine order code. Of course an increase in the speed of multiplication would have had also favourable effects on machine speed when operating in fixed-point on 40 bits numbers. The speed on FINAC measured in Gills for operation on 40-bit numbers is 12,96 ms and would be 12 ms with a fast multiplier.","PeriodicalId":174899,"journal":{"name":"ACM '58","volume":"232 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1958-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACM '58","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/610937.610961","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2

Abstract

The serial computer FINAC, a Ferranti l~lark I machine installed at the Consiglio Nazionale delle Ricerche in Rome, has been frequently used for the solution of systems of algebraic linear equations and in general of problems of linear algebra. As the size and the number of ill-conditioned systems to be solved had a tendency to increase the problem arose of performing the computation with a greater accuracy, in order not to lose all of the approximation during the execution of the programs. It was decided that efficient programs could be written using only the machine fixed point instructions, that is, without pro-grassing the machine for floating point operation, if the maximum length of 40 bits of the numbers employed could be doubled. This could be done by programme, but in this case the machine would be slowed down to a considerable extent. Two different possibilities were open : that of increasing the speed of multiplication on numbers of 40 bits by doubling the parallelism of the series-parallel multiplication circuit ,ioe° using the multiplier's digits in groups of two and staticising the whole multiplier in a single step, and that of installing an arithmetic unit operating on 40 or 80 bits words. In the first case fixed-point 80-bits operation would always have been possible by programming every single arithmetic operation, in the second case 80-bit instructions would have been available in the machine order code. Of course an increase in the speed of multiplication would have had also favourable effects on machine speed when operating in fixed-point on 40 bits numbers. The speed on FINAC measured in Gills for operation on 40-bit numbers is 12,96 ms and would be 12 ms with a fast multiplier.
串行计算机中离散可变字长的二进制算法
串行计算机FINAC是安装在罗马国立科学院的一台ferrananti - larlark I型计算机,它经常被用于求解代数线性方程组和一般线性代数问题。由于待解决的病态系统的大小和数量有增加的趋势,因此出现了以更高的精度执行计算的问题,以便在程序执行期间不丢失所有近似值。他们决定,如果所使用的40位数字的最大长度可以增加一倍,那么可以只使用机器的定点指令来编写高效的程序,也就是说,不需要为机器的浮点操作提供支持。这可以通过程序来完成,但在这种情况下,机器将在相当程度上减慢速度。有两种不同的可能性:一种是通过将串并联乘法电路的并行度加倍来提高40位数字的乘法速度,另一种是将乘数的数字分成两组,并在一步中使整个乘数静态化,另一种是安装一个运算单元,处理40位或80位的单词。在第一种情况下,通过编程每一个算术运算,定点80位操作总是可能的,在第二种情况下,80位指令将在机器命令代码中可用。当然,当对40位数字进行定点运算时,乘法速度的提高也会对机器速度产生有利的影响。在gils中测量的FINAC对40位数字的操作速度为12.96 ms,使用快速乘法器将为12 ms。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信
小红书