3.8 ERROR UNIT
The error unit was devoted to check for errors resulting from the division operation since it is the most critical operation in the chip in addition to the fact that the error checking for all the operations in the chip will cause the area to increase dramatically. This units outputs an error whenever the two conditions present during the division operation :
1. The divisor is equal to zero ( DIVISION BY ZERO !!!!).
2. The divisor is less than 0.5 or -0.5. This cndition will cause an error in the result since the number format for division is 2.6, so division by nubers less than 0.5 or -0.5 will cause the answer to be greater than 2 which can not be represented in the format choosen.
This unit is simply a combinational circuit that checks if the divisor is zero by simply using an eight input nor gate. To check for the first condition, the most significant three bits of divisor where checked. the output of the two subunits were then anded with the muldiv signal so that the error is outputed when the operation is division only.
|
Divisor |
Binary value |
|
less than 0.5 |
(000) xxxxx |
|
less than -0.5 |
(111) xxxxx |
Table 5: the bits checked for the divisor range