公开(公告)号：US4025773A

公开(公告)日：1977-05-24

申请号：US639516

申请日：1975-12-15

Applicant: Walter Scott Bennett

Inventor： Walter Scott Bennett

IPC: G06F7/52

CPC classification number: G06F7/5272 ,  G06F7/535 ,  G06F2207/3852 ,  G06F2207/5353 ,  G06F2207/5356

Abstract: Three modular arrays structured from a common module are connected together a first way to form a binary quotient by successive approximations, or a second way to form a binary product. Any one of the three modular arrays may be used to add or subtract two binary numbers. To divide, one array is utilized to effectively form the reciprocal of the binary divisor, most significant bit first, by successive approximation. Control circuitry, including a carry detector, dictates the formation of the shift and add sequence that effectively represents the reciprocal of the divisor by controlling the positioning of the divisor before each addition step so that the product is a series of binary ones. The add and shift sequence utilized to generate the series of binary ones, as it is evolving, is also being utilized to manipulate the dividend, thereby forming the quotient, most significant bit first. In effect, the dividend is being multiplied by the reciprocal of the divisor so as to form an approximate product of the dividend and reciprocal of the divisor, most significant bit first. This product is actually an increasingly precise approximation of the quotient of the dividend and divisor. The binary product of two numbers is formed, most significant bit first, by manipulating the multiplicand according to an add and shift sequence determined by use of the multiplier.

Abstract translation: 由公共模块构成的三个模块化阵列通过逐次逼近形成二进制商的第一种方式连接在一起，或者形成二进制产品的第二种方法。 三个模块阵列中的任何一个可用于加或减两个二进制数。 为了划分，利用一个阵列通过逐次近似有效地形成二进制除数的倒数，最高有效位。 包括进位检测器的控制电路决定了通过在每个加法步骤之前控制除数的位置使得该乘积成为一系列二进制的方法，来形成有效地表示除数的倒数的移位和加法序列。 用于生成二进制序列的加法和移位序列随着其进化，也被用于操纵分红，从而形成商，最重要的是位。 实际上，红利乘以除数的倒数，从而形成除数的近似乘积和最大有效位。 这个产品实际上是一个越来越精确的近似股息和除数的商。 通过根据通过使用乘法器确定的加法和移位序列来操纵被乘数，形成两个数字的二进制积，最高有效位。

公开(公告)号：US4047011A

公开(公告)日：1977-09-06

申请号：US639517

申请日：1975-12-15

Applicant: Walter Scott Bennett

Inventor： Walter Scott Bennett

IPC: G06F7/52 ,  G06F7/57

CPC classification number: G06F7/5272 ,  G06F7/535 ,  G06F7/57 ,  G06F2207/3852 ,  G06F2207/5353 ,  G06F2207/5356

Abstract: Three modular arrays structured from a common module are connected together a first way to form a binary quotient by successive approximations, or a second way to form a binary product. Any one of the three modular arrays may be used to add or subtract two binary numbers. To divide, one array is utilized to generate a shift and add sequence that represents the reciprocal of the divisor, most significant bit first. As this add and shift sequence is being formed, it is, at the same time, being utilized to manipulate the dividend, thereby forming the quotient, most significant bit first. In effect, the dividend is being multiplied by the reciprocal of the divisor so as to form a product of the dividend and reciprocal of the divisor, most significant bit first. This product is actually the quotient of the dividend and divisor. The binary product of two numbers is formed, most significant bit first, by manipulating the multiplicand according to an add and shift sequence determined by use of the multiplier.

Abstract translation: 由公共模块构成的三个模块化阵列通过逐次逼近形成二进制商的第一种方式连接在一起，或者形成二进制产品的第二种方法。 三个模块阵列中的任何一个可用于加或减两个二进制数。 为了划分，使用一个数组来产生代表除数倒数的移位和加法序列，最高有效位。 由于正在形成这个加法和移位序列，同时利用它来操纵分红，从而形成商，最重要的是位。 实际上，红利乘以除数的倒数，从而形成除数的乘数和倒数的乘积，最重要的是位。 该产品实际上是股息和除数的商。 通过根据通过使用乘法器确定的加法和移位序列来操纵被乘数，形成两个数字的二进制积，最高有效位。

公开(公告)号：US4011439A

公开(公告)日：1977-03-08

申请号：US639514

申请日：1975-12-15

Applicant: Walter Scott Bennett

Inventor： Walter Scott Bennett

IPC: G06F7/52 ,  G06F7/535

CPC classification number: G06F7/535 ,  G06F2207/5356

Abstract: A plurality of modular arrays, each structured from a common module, are connected together so as to form a binary quotient by successive approximations. For divisors that fall into that group of numbers that have reciprocals with a reasonably short period, the forming of a quotient with such a divisor and any dividend can be greatly accelerated after the add and shift sequence for the first period of the divisor reciprocal is obtained. A unity array, divisor array, dividend array, and quotient array may all be of equal length, but must be longer than the length of the periods of the reciprocals of the divisors utilized. The reciprocal of the divisor is effectively formed in the divisor array by generating a shift and add sequence that will produce a product that is a series of binary ones. After the first period of the divisor reciprocal is formed, the binary bits of the reciprocal start to repeat for the second period, and so on. By using the formed shift and add sequence that effectively represents the reciprocal of the divisor for a single period, to manipulate the dividend, the dividend is effectively multiplied by the reciprocal of the divisor, producing a product, most significant bit first, that is the quotient of the dividend and divisor. After obtaining the shift and add sequence representative of the first period of the divisor reciprocal, the quotient has been formed, most significant bit first, to a precision equal to the number of bits in the first period of the divisor reciprocal. The precision of the quotient can now be doubled by adding the formed quotient with itself after the quotient addend is shifted to the right the number of bit positions to which the quotient is precise. At the next step, the quotient precision can be quadrupled, and so on.

Abstract translation: 由公共模块构成的多个模块化阵列连接在一起，以便通过逐次逼近形成二进制商。 对于属于具有相当短的期间的数量的数字的除数，在获得除数互惠的第一个周期的加法和移位序列之后，可以大大加速形成具有这种除数的任意商和任何股息 。 unity数组，除数数组，除数数组和商数组可能都是相等的长度，但必须长于所使用的除数的倒数的周期长度。 除数的倒数通过产生一个产生一系列二进制数的乘积的移位和相加序列在除数阵列中有效地形成。 形成除数互逆的第一个周期后，倒数的二进制位在第二个周期开始重复，依此类推。 通过使用形成的转移和增加序列，有效地代表除了一个时期的除数的倒数，操纵股息，股息被乘数乘以除数的倒数，产生一个乘积，最高有效位第一，即 股息和除数的商。 在获得代表互易的第一个周期的移位和相加序列之后，已经形成了最高有效位的第一位，其精度等于除数倒数的第一个周期中的位数。 商商的精度现在可以通过在商加法向右移动商精确的位位数之后将其自身相加而形成商加倍。 在下一步，商精度可以是四倍，依此类推。