Lecture 4

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 00000000000000000000 +1 0 01111111 00000000000000000000000 ;1 1 01111111 00000000000000000000000 2;126 0 00000001 00000000000000000000000 2127(2 ; 2;23) 0 11111110 11111111111111111111111 +1 0 11111111 00000000000000000000000 ;1 1 11111111 00000000000000000000000 NaN ? 11111111 Anything but all zeros Denormal ? 00000000 Anything but all zeros { Under ow threshold: 2;126 1:17 10;38 { Over ow threshold: 2127(2 ; 2;23) 2128 3:40 { NaN: not a number { Denormal: un-normalized number (d0 = 0) { u = 2;24 5:96 10;8 2 1038 Floating Point Standards IEEE binary double-precision standard s η − 1023 f 52 bit fraction 1 bit sign 11 bit biased exponent { Represented number: (;1)s2 ;1023(1 + f ) { Under ow threshold: 2;1022 2:23 10;308 { Over ow threshold: 21023(2;2;52) 21024 1:80 10308 { u = 2;53 1:11 10;16 3 Stability Assume that an operation (+ ; = fl(a b) = a b(1 + ) { The relative error is jfl(a b) ; a bj ja bj p ) satis jj u es u (3a) (3b) Analyze the roundo error in computing a dot...
View Full Document

This document was uploaded on 03/16/2014 for the course CSCI 6800 at Rensselaer Polytechnic Institute.

Ask a homework question - tutors are online