# Principle of Computer Composition -- floating point addition and subtraction quiz

`````` One .  Single topic selection （ common 5 topic ,100 branch ）
1. ( Single topic selection ) During floating-point addition and subtraction , Before the mantissa is summed, the order must be checked , On order time （  ）.
A.  Must be aligned from small order to large order , Because the hardware is easy to implement
B.
Must be aligned from large order to small order , Because the hardware is easy to implement

C.
Must be aligned from small order to large order , Because the error caused by this is small

D.
Must be aligned from large order to small order , Because the error caused by this is small

20 branch
2. ( Single topic selection ) For floating-point addition and subtraction , In the following statement about the summation result of mantissa , The wrong is （   ）.
A.  If the sum of mantissa is 1.M In the form of , No need for normalization .
B.
If the sum of mantissa is 1φ.M In the form of , Indicates that the result overflows , Need to normalize to the left .

C.
If the sum of mantissa is 0.M In the form of , Need normalization .

D.  If you need to normalize to the right , Just move right 1 position .
20 branch
3. ( Single topic selection )IEEE754 Floating point operation , use “ Nearest rounding ” Law , hypothesis 23 Potential mantissa M The last two are 01, about , The following statement is wrong （   ）.
A.
If the extra bit is 101, When rounding , Enter into 1, The last two digits of the mantissa become 10

B.
If the extra bit is 011, When rounding , Give up , The last two digits of the mantissa are still 01

C.
If the extra bit is 100, When rounding , Give up , The last two digits of the mantissa are still 01

D.  If the extra bit is 111, When rounding , Enter into 1, The last two digits of the mantissa become 10
20 branch
4. ( Single topic selection ) about IEEE754 The standard 32 Bit addition and subtraction , About overflow handling , The following statement is wrong (  ).
A.
The overflow of floating-point number depends on whether its order code overflows .

B.
Order code e>127 when , Operation result overflow .

C.
Order code e≤-126 when , Operation result underflow .

D.  Mantissa overflow determines order code overflow .
20 branch
5. ( Single topic selection ) If the order code or mantissa overflow , The error in the following system processing methods is （    ）.
A.
When the order code overflows , This floating point number is generally regarded as +∞ or -∞.

B.
Order code underflow , This floating point number is generally regarded as O.

C.
When mantissa overflow , Need to normalize to the right .

D.  When mantissa underflow , This floating point number is generally regarded as O.