Quiz submission record for quiz3-2-1 at Wed Jul  7 02:06:49 2004:

Your Answer for Question 1:
I am not sure how one would specifically exploit security holes, i assume it would be something like loading an integer, then using it as an address and modifying whatever was at the address.

To prevent this, one could type-tag data(000 for int, 001 for float etc.)
The tradeoff is that there are then fewer bits available to represent the data.


Your Answer for Question 2:
I do not recall reading about "subnormals," but for floating points, which are stored as 1.xxxx * 2^yyyy
a non-normal floating point would be of the format zzz.xxxx *2^yyyy or 0.00xxx * 2^yyyyy
The convention for storing floating points does not account for non-normalized numbers and would return an incorrect number.

Your Answer for Question 3:
Older hardware lacked a guard bit for doubles.

Your Answer for Question 4:
For:
Not adhering to the standard opens up more possibilities, where a certain method may yield faster algorithms and thus optimize the system

Against:
If the system follows the standard, it can use older software which also utilizes the standards and be able to port its code with other systems using the standard


Your unique submission ID is quiz3-2-1-cs61c-am-1089191209-783.