Quiz submission record for quiz3-2-1 at Wed Jul  7 09:26:18 2004:

Your Answer for Question 1:
MIPS instructions are represented by field types of I, J and R. These field types has different bit patterns that presented opcode, registers, shtamount, functions and immediates. Since they are presented in bit, it is possible for someone to alter the bits since one can just change 1 bit, then MIPS will interpret it very differently. Someone can try to access restricted memory address which will mess up the operating system and the memory management. That is when the computer crashes. 
One way to minimize the threat is to specify the opcodes for each instruction. We can also limit how a person can change the bit patterns. Or strictly following the rules of making instructions. The main trade off will be less elbow room for programmers or users to work with instructions that have too many restrictions.

Your Answer for Question 2:
First, subnormal is related to underflow threshold. Underflow threshold is defined when a calculation result in non-zero number whose magnitude would be less than the underflow threshold, than it is considered to have underflow. The result will then sometimes be "subnormal number", but it maybe round down or round up to the threshold value.
So, we can define subnormal as the the magnitude of a non-zero number that is less than the underflow threshold.
It is important because when result is too small and might not be able to be represented. Underflow is differnent than overflow issues. So, when we want to calculate smaller than predefined limites, underflow is useful. In another word, we can represent a very small value that fall into the gap between zero and the underflow thresholds. So, in this small range, we can present subnormal numbers.


Your Answer for Question 3:
Programmer used to base their design according to the hardware specification. So, for single and double precisions, the missing guard bits has created many problems in terms of precision and accuracies. The format of writing (0.5-x)+0.5 is a way to compensate the missing guard bits to avoid the errors. This method were used in the older machines such as S/360 7090,7094 and Cray's. So, the main idea here is they had to deal with the guard bit issues.

Your Answer for Question 4:
Con: 
1. standardization is hard to implemented, and it demands both time and regulation.
2. we must deal with issues such as balancing language and complier against other aspects.

Pro:
1. standarization makes program to work across systems and the method of representing floating points can be resolved.
2. it is more efficient and accurate when things are standardized. Unless more accuracy is required, then, there should be no need to alter the standard.


Your unique submission ID is quiz3-2-1-cs61c-es-1089217578-1913.