Double precision approximate quotient network

Claim 2

Apparatus for determining a 2n bit partial product of a first 2 n bit number and a second n bit number, said n bit number having a value equal to one of (2 n-1 -2), (2 n-1 -1), (2 n-1 ) and (2 n-1 +1), said apparatus comprising: means for right shifting said first number by one bit to derive a fourth number having a value of either one-half of said first number or one-half of said first number minus 1; means for determining a third number comprising the n most significant bits of said first number; means for multiplying said third number by 2 to achieve a value of twice said third number; and means responsive to said shifting means and said determining means and said multiplying means and to the value of said second number for (a) subtracting twice said third number from said fourth number if the value of said second number is 2 n-1 -2, (b) subtracting said third number from said fourth number if the value of said second number is 2 n-1 -1, (c) selecting said fourth number if said the value of said second number is 2 n-1 , and (d) adding said third number to said fourth number if the value of said second number is 2 n-1 +1.

Claim 3

Apparatus according to claim 1 wherein said multiplying means and said responsive means comprises means for determining the 2's complement of said third number, means for determining twice the 2's complement of said third number, select means responsive to the value of said second number (a) for selecting twice the 2's complement of said third number when the value of said second number is 2 n-1 -2, (b) for selecting said 2's complement of said third number when the value of said second number is 2 n-1 -1, (c) for selecting zero when the value of said second number is 2 n-1 , and (d) for selecting said third number when the value of said second number is 2 n-1 +1, and add means for adding said fourth number to the number selected by said select means.

Claim 4

Apparatus for determining a 2n bit approximate quotient of a 2n bit dividend and a 2n bit divisor comprising: means for deriving a first number equal to 2 times 2 n-1 (2 n-1 -1) divided by the n most significant bits of said divisor; first multiplier means for multiplying said divisor by said first number to derive a result; first complement means for determining the 2's complement of the result from said first multiplier means; second multiplier means for multiplying said dividend by said first number to derive a result; second complement means for determining the 2's complement of the n most significant bits of the result from said second multiplier means; determine means for determining 2 times the 2's complement of the n most significant bits of the result from said second multiplier means; select means responsive to the n most significant bits of the 2's complement from said first complement means for selecting a value equal to one of (a) the 2's complement from said second complement means, (b) the 2's complement from said determine means, (c) zero, or (d) the n most significant bits of the result from said second multiplier means; shift means for right shifting the result from said second multiplier means one place; third multiplier means for multiplying the n most significant bits of the result from said second multiplier means by the n least significant bits of the 2's complement from said first complement means to derive a result; and adder means for adding the result from said third multiplier means and the result of right shifting by said shift means and the value selected by said select means to determine a 2n bit approximate quotient.

Claim 5

Apparatus for determining a 2n bit approximate quotient of a 2n bit dividend and a 2n bit divisor comprising: means for deriving a first number equal to 2 times 2 n-1 (2 n-1 -1) divided by the n most significant bits of said divisor; first multiplier means for multiplying said divisor by said first number to derive a result; first complement means for determining the 2's complement of the result from said first multiplier means; second multiplier means for multiplying said dividend by said first number to derive a result; second complement means for determining the 2's complement of the n most significant bits of the result from said second multiplier means; determine means for determining 2 times the 2's complement of the n most significant bits of the result from said second multiplier means; select means responsive to the n most significant bits of the 2's complement from said first complement means for selecting a value equal to one of (a) the 2's complement from said second complement means, (b) the 2's complement from said determine means, (c) zero, or (d) the n most significant bits of the result from said second multiplier means; shift means for right shifting the result from said second multiplier means one place; third multiplier means for multiplying the n most significant bits of the result from said second multiplier means by the n least significant bits of the 2's complement from said first complement means to derive a result; and adder means for adding the result from said third multiplier means and the result of right shifting by said shift means and the value selected by said select means to determine a 2n bit approximate quotient.

Claim 6

Apparatus according to claim 3 further including means for right shifting said approximate quotient and said divisor by predetermined amounts and for left shifting said dividend a predetermined amount, fourth multiplier means for multiplying the n most significant bits of said shifted approximate quotient by the n least significant bits of said shifted divisor to achieve a result; fifth multiplier means for multiplying the n least significant bits of said shifted approximate quotient by said shifted divisor to achieve a result; and arithmetic means for determining the difference between the shifted dividend and the sum of the n least significant bits of the results of said fourth and fifth multiplier means to determine a correction value.

Claim 7

Apparatus according to claim 4 further including means for adjusting the value of said approximate quotient by said correction value to achieve a true quotient.

Claim 8

Apparatus according to claim 3 wherein said adder means further adds binary 0101 to said approximate quotient.

Claim 11

The method according to claim 7 wherein the third number is multiplied by 2 and the operations are performed by determining the 2's complement of said third number, multiplying 2's complement of said third number by 2, selecting twice the 2's complement of said third number when the value of said second number is 2 n-1 -3, selecting said 2's complement of said third number when the value of said second number is 2 n-1 -1, and selecting said third number when the value of said second number is 2 n-1 +1, and adding said fourth number to the selected number.