DOCSLIB.ORG

Extended Precision Floating Point Arithmetic

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Extended Precision Floating Point Arithmetic Extended Precision Floating Point Numbers for Ill-Conditioned Problems Daniel Davis, Advisor: Dr. Scott Sarra Department of Mathematics, Marshall University Floating Point Number Systems Precision Extended Precision Patriot Missile Failure Floating point representation is based on scientific An increasing number of problems exist for which Patriot missile defense modules have been used by • The precision, p, of a floating point number notation, where a nonzero real decimal number, x, IEEE double is insufficient. These include modeling the U.S. Army since the mid-1960s. On February 21, system is the number of bits in the significand. is expressed as x = ±S × 10E, where 1 ≤ S < 10. of dynamical systems such as our solar system or the 1991, a Patriot protecting an Army barracks in Dha- This means that any normalized floating point The values of S and E are known as the significand • climate, and numerical cryptography. Several arbi- ran, Afghanistan failed to intercept a SCUD missile, number with precision p can be written as: and exponent, respectively. When discussing float- trary precision libraries have been developed, but leading to the death of 28 Americans. This failure E ing points, we are interested in the computer repre- x = ±(1.b1b2...bp−2bp−1)2 × 2 are too slow to be practical for many complex ap- was caused by floating point rounding error. The sentation of numbers, so we must consider base 2, or • The smallest x such that x > 1 is then: plications. David Bailey’s QD library may be used system measured time in tenths of seconds, using binary, rather than base 10. Thus, a non-zero num- −(p−1) in applications where two or four times double pre- only 24-bit registers. The algorithms employed by (1.00...01)2 = 1 + 2 ber, x, is written in the form, x = ±S × 2E, where cision is sufficient. Though much faster than ar- the Patriot were implemented in a way that allowed The gap between this number and 1 is called 1 ≤ S < 2. It follows, then, that the binary expan- • bitrary precision, the fact that these algorithms are the rounding error for the time to compound over machine epsilon, which we can write as: sion of the significand is given by S = (b0.b1b2b3...)2, implemented in software still forces them to be much time. When the Dharan system failed, it had been −(p−1) with b0 = 1. Consider x = 7.0. Then, in a floating m = (0.00...01)2 = 2 slower than double calculations. The table below left on for about 100 hours. 2 point system, we might have x = 1.75 × 2 . • Let xc be a floating point representation of a real displays a time comparison among double, double- number, x. Then the absolute error of xc is given double, and quad-double calculations for the basic Conclusion/Future Work |xc−x| IEEE Double Standard by |xc − x|. Its relative error is given by |x| . arithmetic operators, with the data normalized so • The maximum value of the rounding error for a that the times for doubles represent one time unit. Extended precision offers a solution for many prob- The ANSI/IEEE std 754 defines the double floating m lems in numerical mathematics that suffer from the binary floating point representation is equal to 2 . Type Add. Sub. Mult. Div. point format used in virtually all microprocessors. rounding error in IEEE double. However, that solu- Format Precision Machine Epsilon Double 1 1 1 1 tion comes at the cost of efficiency. Modern Graph- • Double values use a 64-bit word, with one bit −23 −7 Double-Double 21.45 11.87 13.83 15.53 ics Processing Units (GPUs) offer thousands of cores reserved for the sign, 11 for the exponent, and 52 Single p = 24 m = 2 ≈ 1.2 ∗ 10 −52 −16 Quad-Double 70.22 74.10 120.21 129.11 which can be highly advantageous for parallelizable for the significand. Double p = 53 m = 2 ≈ 2.2 ∗ 10 Table 2: Times Required for Arithmetic Operations with Data algorithms. A parallel implementation of the QD • The exponent field is stored using biased Table 1: Precision of Floating Point Formats Normalized so that Double Precision is 1 Unit For Each Operator library known as GQD was published in 2010, but representation, meaning that the value of the abandoned shortly after. In the future, we hope to exponent B is stored as E + B, where B is the implement the QD algorithms in OpenCL for exe- exponent bias, which is 1023 for double precision. Solutions of Chaotic Lorenz Equations cution across GPUs and other highly parallel plat- The 52-bit binary significand allows for about 16 • Chaotic systems are those which are highly sensitive to changes in initial conditions.. We seek time-step forms. decimal places of precision. independent solutions to the chaotic Lorenz equations: References 0 x = σ(y − x), x(0) = x0, 0 Sample Conversion y = rx − y − xz, y(0) = y0, (1) 0 [1] Yozo Hida, Xiaoye S Li, and David H Bailey. z = −bz + xy, z(0) = z0. Library for double-double and quad-double arithmetic. Let us convert x = 4.5 into an IEEE double. The We performed the experiment with step sizes ∆t = 1 × 10−6 and ∆t = 1 × 10−7. Figure 1 shows NERSC Division, Lawrence Berkeley National sign is positive, so the sign bit is 0. We use division 1 2 the resulting x-values of the solutions calculated in double precision, with the blue line representing the Laboratory, 2007. to find that x/22 = 1.125, giving us an exponent of solutions found using ∆t , and the green line using ∆t . The solutions visually diverge around t = 42. [2] Scott A Sarra and Clyde Meador. 2. Using biased representation, this value is stored 1 2 Figure 2 represents the same calculation performed in double-double precision. Here, the solutions do not On the numerical solution of chaotic dynamical systems as 1025, which has a binary string of 10000000001. using extend precision floating point arithmetic and very visually diverge for t ≤ 50. The decimal portion of .125 is converted to binary high order numerical methods. Nonlinear Analysis, 16(3):340–352, 2011. using the Repeated Multiply-by-2 method, append- ing the value left of the decimal point to the frac- tional part at each stage, until we reach the exact Acknowledgements value or the limit of the length of the significand. Thus, the double precision value is given by: Thank you to Dr. Sarra and Dr. Niese for your Sign Exponent Mantissa Figure 1: Solutions in double precision. Figure 2: Solutions in double-double precision. support and for answering so many questions during 0 10000000001 0010000000 ... this project..
Load more

Recommended publications
  • Arithmetic Algorithms for Extended Precision Using Floating-Point Expansions Mioara Joldes, Olivier Marty, Jean-Michel Muller, Valentina Popescu
    Arithmetic algorithms for extended precision using floating-point expansions Mioara Joldes, Olivier Marty, Jean-Michel Muller, Valentina Popescu To cite this version: Mioara Joldes, Olivier Marty, Jean-Michel Muller, Valentina Popescu. Arithmetic algorithms for extended precision using floating-point expansions. IEEE Transactions on Computers, Institute of Electrical and Electronics Engineers, 2016, 65 (4), pp.1197 - 1210. 10.1109/TC.2015.2441714. hal- 01111551v2 HAL Id: hal-01111551 https://hal.archives-ouvertes.fr/hal-01111551v2 Submitted on 2 Jun 2015 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. IEEE TRANSACTIONS ON COMPUTERS, VOL. , 201X 1 Arithmetic algorithms for extended precision using floating-point expansions Mioara Joldes¸, Olivier Marty, Jean-Michel Muller and Valentina Popescu Abstract—Many numerical problems require a higher computing precision than the one offered by standard floating-point (FP) formats. One common way of extending the precision is to represent numbers in a multiple component format. By using the so- called floating-point expansions, real numbers are represented as the unevaluated sum of standard machine precision FP numbers. This representation offers the simplicity of using directly available, hardware implemented and highly optimized, FP operations.
    [Show full text]
  • Floating Points
    Jin-Soo Kim ([email protected]) Systems Software & Architecture Lab. Seoul National University Floating Points Fall 2018 ▪ How to represent fractional values with finite number of bits? • 0.1 • 0.612 • 3.14159265358979323846264338327950288... ▪ Wide ranges of numbers • 1 Light-Year = 9,460,730,472,580.8 km • The radius of a hydrogen atom: 0.000000000025 m 4190.308: Computer Architecture | Fall 2018 | Jin-Soo Kim ([email protected]) 2 ▪ Representation • Bits to right of “binary point” represent fractional powers of 2 • Represents rational number: 2i i i–1 k 2 bk 2 k=− j 4 • • • 2 1 bi bi–1 • • • b2 b1 b0 . b–1 b–2 b–3 • • • b–j 1/2 1/4 • • • 1/8 2–j 4190.308: Computer Architecture | Fall 2018 | Jin-Soo Kim ([email protected]) 3 ▪ Examples: Value Representation 5-3/4 101.112 2-7/8 10.1112 63/64 0.1111112 ▪ Observations • Divide by 2 by shifting right • Multiply by 2 by shifting left • Numbers of form 0.111111..2 just below 1.0 – 1/2 + 1/4 + 1/8 + … + 1/2i + … → 1.0 – Use notation 1.0 – 4190.308: Computer Architecture | Fall 2018 | Jin-Soo Kim ([email protected]) 4 ▪ Representable numbers • Can only exactly represent numbers of the form x / 2k • Other numbers have repeating bit representations Value Representation 1/3 0.0101010101[01]…2 1/5 0.001100110011[0011]…2 1/10 0.0001100110011[0011]…2 4190.308: Computer Architecture | Fall 2018 | Jin-Soo Kim ([email protected]) 5 Fixed Points ▪ p.q Fixed-point representation • Use the rightmost q bits of an integer as representing a fraction • Example: 17.14 fixed-point representation
    [Show full text]
  • Hacking in C 2020 the C Programming Language Thom Wiggers
    Hacking in C 2020 The C programming language Thom Wiggers 1 Table of Contents Introduction Undefined behaviour Abstracting away from bytes in memory Integer representations 2 Table of Contents Introduction Undefined behaviour Abstracting away from bytes in memory Integer representations 3 – Another predecessor is B. • Not one of the first programming languages: ALGOL for example is older. • Closely tied to the development of the Unix operating system • Unix and Linux are mostly written in C • Compilers are widely available for many, many, many platforms • Still in development: latest release of standard is C18. Popular versions are C99 and C11. • Many compilers implement extensions, leading to versions such as gnu18, gnu11. • Default version in GCC gnu11 The C programming language • Invented by Dennis Ritchie in 1972–1973 4 – Another predecessor is B. • Closely tied to the development of the Unix operating system • Unix and Linux are mostly written in C • Compilers are widely available for many, many, many platforms • Still in development: latest release of standard is C18. Popular versions are C99 and C11. • Many compilers implement extensions, leading to versions such as gnu18, gnu11. • Default version in GCC gnu11 The C programming language • Invented by Dennis Ritchie in 1972–1973 • Not one of the first programming languages: ALGOL for example is older. 4 • Closely tied to the development of the Unix operating system • Unix and Linux are mostly written in C • Compilers are widely available for many, many, many platforms • Still in development: latest release of standard is C18. Popular versions are C99 and C11. • Many compilers implement extensions, leading to versions such as gnu18, gnu11.
    [Show full text]
  • Floating Point Formats
    Telemetry Standard RCC Document 106-07, Appendix O, September 2007 APPENDIX O New FLOATING POINT FORMATS Paragraph Title Page 1.0 Introduction..................................................................................................... O-1 2.0 IEEE 32 Bit Single Precision Floating Point.................................................. O-1 3.0 IEEE 64 Bit Double Precision Floating Point ................................................ O-2 4.0 MIL STD 1750A 32 Bit Single Precision Floating Point............................... O-2 5.0 MIL STD 1750A 48 Bit Double Precision Floating Point ............................. O-2 6.0 DEC 32 Bit Single Precision Floating Point................................................... O-3 7.0 DEC 64 Bit Double Precision Floating Point ................................................. O-3 8.0 IBM 32 Bit Single Precision Floating Point ................................................... O-3 9.0 IBM 64 Bit Double Precision Floating Point.................................................. O-4 10.0 TI (Texas Instruments) 32 Bit Single Precision Floating Point...................... O-4 11.0 TI (Texas Instruments) 40 Bit Extended Precision Floating Point................. O-4 LIST OF TABLES Table O-1. Floating Point Formats.................................................................................... O-1 Telemetry Standard RCC Document 106-07, Appendix O, September 2007 This page intentionally left blank. ii Telemetry Standard RCC Document 106-07, Appendix O, September 2007 APPENDIX O FLOATING POINT
    [Show full text]
  • Scilab Textbook Companion for Numerical Methods for Engineers by S
    Scilab Textbook Companion for Numerical Methods For Engineers by S. C. Chapra And R. P. Canale1 Created by Meghana Sundaresan Numerical Methods Chemical Engineering VNIT College Teacher Kailas L. Wasewar Cross-Checked by July 31, 2019 1Funded by a grant from the National Mission on Education through ICT, http://spoken-tutorial.org/NMEICT-Intro. This Textbook Companion and Scilab codes written in it can be downloaded from the "Textbook Companion Project" section at the website http://scilab.in Book Description Title: Numerical Methods For Engineers Author: S. C. Chapra And R. P. Canale Publisher: McGraw Hill, New York Edition: 5 Year: 2006 ISBN: 0071244298 1 Scilab numbering policy used in this document and the relation to the above book. Exa Example (Solved example) Eqn Equation (Particular equation of the above book) AP Appendix to Example(Scilab Code that is an Appednix to a particular Example of the above book) For example, Exa 3.51 means solved example 3.51 of this book. Sec 2.3 means a scilab code whose theory is explained in Section 2.3 of the book. 2 Contents List of Scilab Codes4 1 Mathematical Modelling and Engineering Problem Solving6 2 Programming and Software8 3 Approximations and Round off Errors 10 4 Truncation Errors and the Taylor Series 15 5 Bracketing Methods 25 6 Open Methods 34 7 Roots of Polynomials 44 9 Gauss Elimination 51 10 LU Decomposition and matrix inverse 61 11 Special Matrices and gauss seidel 65 13 One dimensional unconstrained optimization 69 14 Multidimensional Unconstrainted Optimization 72 3 15 Constrained
    [Show full text]
  • Introduction & Floating Point NMMC §1.4.1, Atkinson §1.2 1 Floating
    Introduction & Floating Point NMMC x1.4.1, Atkinson x1.2 1 Floating-point representation, IEEE 754. In binary, 11:01 = 1 × 21 + 1 × 20 + 0 × 2−1 + 1 × 2−2: Double-precision floating-point format has 64 bits: ±; a1; : : : ; a11; b1; : : : ; b52 The associated number is a1···a11−1023 # = ±1:b1 ··· b52 × 2 : If all the a are zero, then the number is `subnormal' and the representation is −1022 # = ±0:b1 ··· b52 × 2 If the a are all 1 and the b are all 0 then it's ±∞ (± 1/0). If the a are all 1 and the b are not all 0 then it's NaN (±0=0). Machine-epsilon is the difference between 1 and the smallest representable number larger than 1, i.e. = 2−52 ≈ 2 × 10−16. The smallest and largest positive numbers are ≈ 10±308. Numbers larger than the max are set to 1. 2 Rounding & Arithmetic. Round-toward-nearest rounds a number towards the nearest floating-point num- ber. For nonzero numbers x within the range of max/min, the relative errors due to rounding are jround(x) − xj ≤ 2−53 ≈ 10−16 = u jxj which is half of machine epsilon. Floating-point arithmetic: add, subtract, multiply, divide. IEEE 754: If you start with 2 exactly- represented floating-point numbers x and y and compute any arithmetic operation `op', the magnitude of the relative error in the result is less than half machine precision. I.e. there is some δ with jδj ≤u such that fl(x op y) = (x op y)(1 + δ): The exception is underflow/overflow, where the result is either larger or smaller than can be represented.
    [Show full text]
  • 1.2 Round-Off Errors and Computer Arithmetic
    1.2 Round-off Errors and Computer Arithmetic 1 • In a computer model, a memory storage unit – word is used to store a number. • A word has only a finite number of bits. • These facts imply: 1. Only a small set of real numbers (rational numbers) can be accurately represented on computers. 2. (Rounding) errors are inevitable when computer memory is used to represent real, infinite precision numbers. 3. Small rounding errors can be amplified with careless treatment. So, do not be surprised that (9.4)10= (1001.0110)2 can not be represented exactly on computers. • Round-off error: error that is produced when a computer is used to perform real number calculations. 2 Binary numbers and decimal numbers • Binary number system: A method of representing numbers that has 2 as its base and uses only the digits 0 and 1. Each successive digit represents a power of 2. (… . … ) where 0 1, for each = 2,1,0, 1, 2 … 3210 −1−2−3 2 • Binary≤ to decimal:≤ ⋯ − − (… . … ) (… 2 + 2 + 2 + 2 = 3210 −1−2−3 2 + 2 3 + 22 + 1 2 …0) 3 2 1 0 −1 −2 −3 −1 −2 −3 10 3 Binary machine numbers • IEEE (Institute for Electrical and Electronic Engineers) – Standards for binary and decimal floating point numbers • For example, “double” type in the “C” programming language uses a 64-bit (binary digit) representation – 1 sign bit (s), – 11 exponent bits – characteristic (c), – 52 binary fraction bits – mantissa (f) 1. 0 2 1 = 2047 11 4 ≤ ≤ − This 64-bit binary number gives a decimal floating-point number (Normalized IEEE floating point number): 1 2 (1 + ) where 1023 is called exponent − 1023bias.
    [Show full text]
  • Floating Point Representation (Unsigned) Fixed-Point Representation
    Floating point representation (Unsigned) Fixed-point representation The numbers are stored with a fixed number of bits for the integer part and a fixed number of bits for the fractional part. Suppose we have 8 bits to store a real number, where 5 bits store the integer part and 3 bits store the fractional part: 1 0 1 1 1.0 1 1 $ 2& 2% 2$ 2# 2" 2'# 2'$ 2'% Smallest number: Largest number: (Unsigned) Fixed-point representation Suppose we have 64 bits to store a real number, where 32 bits store the integer part and 32 bits store the fractional part: "# "% + /+ !"# … !%!#!&. (#(%(" … ("% % = * !+ 2 + * (+ 2 +,& +,# "# "& & /# % /"% = !"#× 2 +!"&× 2 + ⋯ + !&× 2 +(#× 2 +(%× 2 + ⋯ + ("%× 2 0 ∞ (Unsigned) Fixed-point representation Range: difference between the largest and smallest numbers possible. More bits for the integer part ⟶ increase range Precision: smallest possible difference between any two numbers More bits for the fractional part ⟶ increase precision "#"$"%. '$'#'( # OR "$"%. '$'#'(') # Wherever we put the binary point, there is a trade-off between the amount of range and precision. It can be hard to decide how much you need of each! Scientific Notation In scientific notation, a number can be expressed in the form ! = ± $ × 10( where $ is a coefficient in the range 1 ≤ $ < 10 and + is the exponent. 1165.7 = 0.0004728 = Floating-point numbers A floating-point number can represent numbers of different order of magnitude (very large and very small) with the same number of fixed bits. In general, in the binary system, a floating number can be expressed as ! = ± $ × 2' $ is the significand, normally a fractional value in the range [1.0,2.0) .
    [Show full text]
  • A Variable Precision Hardware Acceleration for Scientific Computing Andrea Bocco
    A variable precision hardware acceleration for scientific computing Andrea Bocco To cite this version: Andrea Bocco. A variable precision hardware acceleration for scientific computing. Discrete Mathe- matics [cs.DM]. Université de Lyon, 2020. English. NNT : 2020LYSEI065. tel-03102749 HAL Id: tel-03102749 https://tel.archives-ouvertes.fr/tel-03102749 Submitted on 7 Jan 2021 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. N°d’ordre NNT : 2020LYSEI065 THÈSE de DOCTORAT DE L’UNIVERSITÉ DE LYON Opérée au sein de : CEA Grenoble Ecole Doctorale InfoMaths EDA N17° 512 (Informatique Mathématique) Spécialité de doctorat :Informatique Soutenue publiquement le 29/07/2020, par : Andrea Bocco A variable precision hardware acceleration for scientific computing Devant le jury composé de : Frédéric Pétrot Président et Rapporteur Professeur des Universités, TIMA, Grenoble, France Marc Dumas Rapporteur Professeur des Universités, École Normale Supérieure de Lyon, France Nathalie Revol Examinatrice Docteure, École Normale Supérieure de Lyon, France Fabrizio Ferrandi Examinateur Professeur associé, Politecnico di Milano, Italie Florent de Dinechin Directeur de thèse Professeur des Universités, INSA Lyon, France Yves Durand Co-directeur de thèse Docteur, CEA Grenoble, France Cette thèse est accessible à l'adresse : http://theses.insa-lyon.fr/publication/2020LYSEI065/these.pdf © [A.
    [Show full text]
  • Effectiveness of Floating-Point Precision on the Numerical Approximation by Spectral Methods
    Mathematical and Computational Applications Article Effectiveness of Floating-Point Precision on the Numerical Approximation by Spectral Methods José A. O. Matos 1,2,† and Paulo B. Vasconcelos 1,2,∗,† 1 Center of Mathematics, University of Porto, R. Dr. Roberto Frias, 4200-464 Porto, Portugal; [email protected] 2 Faculty of Economics, University of Porto, R. Dr. Roberto Frias, 4200-464 Porto, Portugal * Correspondence: [email protected] † These authors contributed equally to this work. Abstract: With the fast advances in computational sciences, there is a need for more accurate compu- tations, especially in large-scale solutions of differential problems and long-term simulations. Amid the many numerical approaches to solving differential problems, including both local and global methods, spectral methods can offer greater accuracy. The downside is that spectral methods often require high-order polynomial approximations, which brings numerical instability issues to the prob- lem resolution. In particular, large condition numbers associated with the large operational matrices, prevent stable algorithms from working within machine precision. Software-based solutions that implement arbitrary precision arithmetic are available and should be explored to obtain higher accu- racy when needed, even with the higher computing time cost associated. In this work, experimental results on the computation of approximate solutions of differential problems via spectral methods are detailed with recourse to quadruple precision arithmetic. Variable precision arithmetic was used in Tau Toolbox, a mathematical software package to solve integro-differential problems via the spectral Tau method. Citation: Matos, J.A.O.; Vasconcelos, Keywords: floating-point arithmetic; variable precision arithmetic; IEEE 754-2008 standard; quadru- P.B.
    [Show full text]
  • MIPS Floating Point
    CS352H: Computer Systems Architecture Lecture 6: MIPS Floating Point September 17, 2009 University of Texas at Austin CS352H - Computer Systems Architecture Fall 2009 Don Fussell Floating Point Representation for dynamically rescalable numbers Including very small and very large numbers, non-integers Like scientific notation –2.34 × 1056 +0.002 × 10–4 normalized +987.02 × 109 not normalized In binary yyyy ±1.xxxxxxx2 × 2 Types float and double in C University of Texas at Austin CS352H - Computer Systems Architecture Fall 2009 Don Fussell 2 Floating Point Standard Defined by IEEE Std 754-1985 Developed in response to divergence of representations Portability issues for scientific code Now almost universally adopted Two representations Single precision (32-bit) Double precision (64-bit) University of Texas at Austin CS352H - Computer Systems Architecture Fall 2009 Don Fussell 3 IEEE Floating-Point Format single: 8 bits single: 23 bits double: 11 bits double: 52 bits S Exponent Fraction x = (!1)S "(1+Fraction)"2(Exponent!Bias) S: sign bit (0 ⇒ non-negative, 1 ⇒ negative) Normalize significand: 1.0 ≤ |significand| < 2.0 Always has a leading pre-binary-point 1 bit, so no need to represent it explicitly (hidden bit) Significand is Fraction with the “1.” restored Exponent: excess representation: actual exponent + Bias Ensures exponent is unsigned Single: Bias = 127; Double: Bias = 1203 University of Texas at Austin CS352H - Computer Systems Architecture Fall 2009 Don Fussell 4 Single-Precision Range Exponents 00000000 and 11111111 reserved Smallest
    [Show full text]
  • Absolute and Relative Error All Computations on a Computer Are Approximate by Nature, Due to the Limited Precision on the Computer
    4 2 0 −1 −2 • 5.75 = 4 + 1 + 0.5 + 0.25 = 1 × 2 + 1 × 2 + 1 × 2 + 1 × 2 = 101.11(2) 4 0 −1 • 17.5 = 16 + 1 + 0.5 = 1 × 2 + 1 × 2 + 1 × 2 = 10001.1(2) 2 0 −1 −2 • 5.75 = 4 + 1 + 0.5 + 0.25 = 1 × 2 + 1 × 2 + 1 × 2 + 1 × 2 = 101.11(2) Note that certain numbers are finite (terminating) decimals, actually are peri- odic in binary, e.g. 0.4(10) = 0.01100110011 ...(2) = 0.00110011(2) Lecture of: Machine numbers (a.k.a. binary floating point numbers) 24 Jan 2013 The numbers stored on the computer are, essentially, “binary numbers” in sci- entific notation x = ±a × 2b. Here, a is called the mantissa and b the exponent. We also follow the convention that 1 ≤ a < 2; the idea is that, for any number x, we can always divide it by an appropriate power of 2, such that the result will be within [1, 2). For example: 2 2 x = 5(10) = 1.25(10) × 2 = 1.01(2) × 2 Thus, a machine number is stored as: b x = ±1.a1a2 ··· ak-1ak × 2 • In single precision we store k = 23 binary digits, and the exponent b ranges between −126 ≤ b ≤ 127. The largest number we can thus represent is (2 − 2−23) × 2127 ≈ 3.4 × 1038. • In double precision we store k = 52 binary digits, and the exponent b ranges between −1022 ≤ b ≤ 1023. The largest number we can thus represent is (2 − 2−52) × 21023 ≈ 1.8 × 10308.
    [Show full text]