Scribd
Upload
94 views12 pages

Integer Arithmetic - Division: Dr. Arunachalam V Associate Professor, SENSE

Integer division involves dividing a dividend by a divisor to obtain a quotient and remainder. It is an important operation that is typically more expensive than multiplication. There are different types of division such as full division to obtain both quotient and remainder, or dividing only the quotient or remainder. Division algorithms assume the divisor is normalized. The naive division algorithm computes the quotient digit-by-digit from the most significant to the least significant. Other algorithms like BasecaseDivRem are more efficient.

Uploaded by

Anil Kumar Gorantala
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
94 views12 pages

Integer Arithmetic - Division: Dr. Arunachalam V Associate Professor, SENSE

Integer division involves dividing a dividend by a divisor to obtain a quotient and remainder. It is an important operation that is typically more expensive than multiplication. There are different types of division such as full division to obtain both quotient and remainder, or dividing only the quotient or remainder. Division algorithms assume the divisor is normalized. The naive division algorithm computes the quotient digit-by-digit from the most significant to the least significant. Other algorithms like BasecaseDivRem are more efficient.

Uploaded by

Anil Kumar Gorantala
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Integer Arithmetic - Division

Dr. Arunachalam V
Associate Professor, SENSE
Introduction
• Dividend(D) is divided by the divisor (d) to get the quotient (Q) and the
reminder (R).
• = × +
• Division is the next operation to consider after multiplication.
• Optimizing division is almost as important as optimizing multiplication,
since division is usually more expensive, thus the speedup obtained on
division will be more significant.
• On the other hand, one usually performs more multiplications than
divisions.
• One strategy is to avoid divisions when possible, or replace them by
multiplications.
• An example is when the same divisor is used for several consecutive
operations; one can then precompute its inverse.
Types of division
• We distinguish several kinds of division:
• Full division computes both quotient and remainder,
• While in other cases only the quotient (for example, when dividing two
floating-point significands) or remainder (when multiplying two residues
modulo n) is needed.
• Exact division — when the remainder is known to be zero — and the
problem of dividing by a single word.
Normalized divisors
• In all division algorithms, we assume that divisors are normalized.
• We say that ≔ ∑ is normalized when its most significant word
satisfies ≥ ⁄2.
• This is a stricter condition (for β>2) than simply requiring that be
nonzero.
• If B is not normalized, we can compute A' = 2kA and B‘ = 2kB so that B' is
normalized, then divide A' by B‘ giving A‘ = Q'B‘ + R'.
• k is {0,1,2,3,….}
• The Q is Q' and R is R'/2k.
• If = 52388 ; = 281 B is not normalized.
• For k = 1 ; B‘ = 21 ×281 = 562 and now B is normalized.
• Also, A‘ = 21 ×52388 = 104776
• The Q is 186 and R is 244/21 = 122. [ 104776 = 186 × 562 + 244 ]
Naive Division

Verify the division algorithm using = 52388 ; = 281


= 186 ; = 122
n = 3, m = 3 i 5 4 3 2 1 0
= 104776 ≥ 562 × 10 : ∴ =0 ai 1 0 4 7 7 6
j = m – 1 = 2, bi 5 6 2
∗ = 1 × 10 + 0 /5 = 2 = 2, 9 =2

← 104776 − 2 × 10 × 562 = −7624 =2−1 =1

← −7624 +10 × 562 = 48576 ai 0 4 8 5 7 6


j = 1,
∗ = 4 × 10 + 8 /5 = 9 = 9, 9 = 9

← 48576 − 9 × 10 × 562 = −2004 =9−1=8

← −2004 +10 × 562 = 3616 ai 0 0 3 6 1 6


j = 0,
∗ = 3 × 10 + 6 /5 = 7 = 7, 9 = 7

← 3616 − 7 × 562 = −318 =7−1 =6

← −318 +1 × 562 = 244


Q = (186)10 and R = 244/2 = (122)10 ,
Try this using BasecaseDivRem algorithm
• A = 766 970 544 842 443 844 and B = 862 664 913, with β = 1000, which
gives quotient Q = 889 071 217 and remainder R = 778 334 723.
• = = 862 ≥ 1000; and the B is normalized.
A = 766 970 544 842 443 844 and B = 862 664 913, with β = 1000, which
gives quotient Q = 889 071 217 and remainder R = 778 334 723.

i 5 4 3 2 1 0 n = 3, m = 3
ai 766 970 544 842 443 844
bi 862 664 913
= 766 970 544 842 443 844 ≥ 862 644 913 000 000 000 ∶ ∴ =0

j = m – 1 = 2,
∗ = 766 × 1000 + 970 /862 = 889 = 889, 999 = 889
← 766 970 544 842 443 844 − 889 × 1000 × 862 664 913
← 766 970 544 842 443 844 −889 × 862 664 913 000 000 = 061 437 185 443 844
862 664 913 000 000 × 889
766 590 811
318 296 657 000 000
-766 909 107 657 000 000
766 970 544 842 443 844
061 437 185 443 844
i 4 3 2 1 0
ai 061 437 185 443 844
bi 862 664 913
j=1
∗ = 61 × 1000 + 437 /862 = 71 = 71, 999 =71

← 61 437 185 443 844 − 71 × 1000 × 862 664 913


← 61 437 185 443 844 − 71 × 862 664 913 000 = 187 976 620 844

862 664 913 000 × 71


061 047 064
202 144 823 000
-061 249 208 823 000
061 437 185 443 844
187 976 620 844
i 3 2 1 0
ai 187 976 620 844
bi 862 664 913
j=0
∗ = 187 × 1000 + 976 /862 = 218 = 218, 999 =218

← 187 976 620 844 − 218 × 1000 × 862 664 913


← 187 976 620 844 − 218 × 862 664 913 = −084 330 190

862 664 913 × 218 = 218 − 1 = 217


187 144 199
← −084 330 190 + 1 × 862 664 913
916 752 034 = 778 334 723
-188 060 951 034
Q= = 889 071 217
187 976 620 844
-84 330 190 R =A= 778 334 723
Reference
1. Chapter 1.4.1 & 1.4.2 of Richard P Brent and Paul Zimmerman, “Modern
Computer Arithmetic”, Cambridge University Press 2010.
Next Class
SQUARE ROOT AND GCD

You might also like