This preview shows page 1. Sign up to view the full content.
Unformatted text preview: are to the fourth? Ex plain. (Hint: look at the hex repres entations of the
res ults . How does the floating point format handle negative numbers ?)
3. Mathematic ally , would y ou ex pec t the s ame res ults in c omputations 6 and 7? Do y ou obs erve this res ult ex perimentally ?
Ex plain. (Hint: try s tepping through eac h c omputation)
4. W hat happens if y ou try c omputation 6 in double-prec is ion (64-bit) floating-point mode? W hy ?
5. W hy is there notic eable error in c omputation 8, but not 9? (Hint: think of multiply ing floating point numbers lik e multiply ing
two numbers in s c ientific notation, how do y ou do it?)
6. The root c aus e of the Ariane 5 roc k et failure was is olated to the c onvers ion of a floating point number, whic h s tored the
horiz ontal c omponent of the roc k et's veloc ity , to a 16-bit s igned integer. W hat is the mos t lik ely c aus e of the failure? (Hint:
this was n't s ome s mall rounding error, the c omputed veloc ity was way off, c aus ing the s y s tem to go hay wire)...
View Full Document
This document was uploaded on 03/22/2014.
- Fall '14