{"title":"Division and Overflow Detection in Residue Number Systems","authors":"Yates A. Keir, P. W. Cheney, M. Tannenbaum","doi":"10.1109/TEC.1962.5219389","DOIUrl":null,"url":null,"abstract":"Residue arithmetic has some intriguing characterThe sign of an integer I will be represented implicistics which could possibly be exploited in a special purpose, or even itly: a general purpose, computer. However, simple mechanizations of the operations of division and overflow detection are not inherent in _ < I <-¢ I is ositive the structure of a residue number system. Methods of handling these 2 P operations are discussed in this paper. In the first section, a relatively straightforward division process is presented whose execution time M is comparable to those of existing binary machines. The next section -< I < M X I is negative. shows how the division process can be cut short under certain condi2 M-I. tions. The amount of equipment required for this is not insignificant; however, this equipment can also be used to speed up the normal diviThe integer value of a number W will be designated sion procedure and to detect multiplicative overflow. The multiplicaby [W]. tive overflow detection scheme proposed in the concluding section has Er o the following desirable features: E(I) = 1092(I 1) It does not require a redundant number system. T(I) -2E 2) It is compatible with the division process and requires no Y| X signifies that Y divides X evenly; i.e., X is special circuitry. divisible by Y. If X is not divisible by Y, this is 3) It is faster than the brute-force approaches which either redenoted by Y| /X. quire residue division or essentially check the residue multi(Y, mi) =1 signifies that Y and mi are relatively plication by a multiplication in a more conventional number v system. prime.","PeriodicalId":177496,"journal":{"name":"IRE Trans. Electron. Comput.","volume":"2 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1962-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"37","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IRE Trans. Electron. Comput.","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/TEC.1962.5219389","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 37
Abstract
Residue arithmetic has some intriguing characterThe sign of an integer I will be represented implicistics which could possibly be exploited in a special purpose, or even itly: a general purpose, computer. However, simple mechanizations of the operations of division and overflow detection are not inherent in _ < I <-¢ I is ositive the structure of a residue number system. Methods of handling these 2 P operations are discussed in this paper. In the first section, a relatively straightforward division process is presented whose execution time M is comparable to those of existing binary machines. The next section -< I < M X I is negative. shows how the division process can be cut short under certain condi2 M-I. tions. The amount of equipment required for this is not insignificant; however, this equipment can also be used to speed up the normal diviThe integer value of a number W will be designated sion procedure and to detect multiplicative overflow. The multiplicaby [W]. tive overflow detection scheme proposed in the concluding section has Er o the following desirable features: E(I) = 1092(I 1) It does not require a redundant number system. T(I) -2E 2) It is compatible with the division process and requires no Y| X signifies that Y divides X evenly; i.e., X is special circuitry. divisible by Y. If X is not divisible by Y, this is 3) It is faster than the brute-force approaches which either redenoted by Y| /X. quire residue division or essentially check the residue multi(Y, mi) =1 signifies that Y and mi are relatively plication by a multiplication in a more conventional number v system. prime.
余数运算有一些有趣的特点:整数的符号将被表示成隐含的形式,这些隐含的形式可能在特殊用途的计算机中被利用,甚至可能在通用计算机中被利用。然而,简单的除法和溢出检测操作的机械化并不是_ < I <-ⅱ是一个剩余数系统的正结构所固有的。本文讨论了处理这些2p操作的方法。在第一部分中,给出了一个相对简单的除法过程,其执行时间M与现有二进制机相当。下一节-< I < M X I是负的。说明了在一定条件下如何缩短分割过程。规划设计。为此所需的设备数量并非微不足道;不过,这种设备也可以用来加速正常除法,将一个数字W的整数值指定为运算程序,并检测乘法溢出。可乘性[W]。结论部分提出的主动溢出检测方案具有以下可取特征:E(I) = 1092(I)不需要冗余的数字系统。T(I) -2E 2)与除法过程兼容,不需要Y| X表示Y能将X整除;即,X是特殊电路。如果X不能被Y整除,这是3)它比用Y b| /X表示的暴力方法更快。(Y, mi) =1表示Y和mi在更常规的数字v系统中是相对的乘法运算。主要的