CS61c Midterm Spring 96
Question 1 (2 points):
Subtract 0x001f from 0xfff7, interpreting each as a 2's complement 16-bit integer. Express the answer in
decimal and hexadecimal. Show your work.
Question 2 (2 points):
Two lab partners Harry and David are arguing. Harry says "All integers greater than 0 exactly divisible by 6
have exactly 2 1's in their binary representation." David disagrees. He says, "No, but all such numbers have an
even number of 1's in their representation."
Do you agree with Harry or David, or neither? (Why).
Question 3 (4 points):
The instruction sllv $8, $9, $10 uses the value in register 10 as the shift amount. Actually, it uses only the
least significant 5 bits. Why doesn't it use ALL the bits?
Another way of doing the same SLLV would be to do the mysterious:
lw $8, shifter # first instruction
and $8, mask # where mask is 0xfffff83f
andi $10, 0x1f
sll $10, $10, 6
or $8, $8, $10 # fifth instruction
sw $8, shifter # last instruction
shifter:
sll $8, $9, 0
In particular,
What part of the word is masked out by the first AND instruction?
What is in register $8 after the second instruction?
What is in register $8 after the fifth instruction?
Why do you suppose this code sequence is a bad idea?
Question 4 (3 points):
Recall that in the homework for week 4 you translated C code for conversion of decimal numbers into a
simplified version of Roman numerals. The program is reproduced in its entirety at the end of this test.
At the end of the C program there are these two statements:
*currentRoman = 0;
printf ("Roman numeral equivalent = %s\n", roman);
What does that first statement do, and why?
Assuming that your input was 123, and the statement *currentRoman = 0; was omitted, what would be
printed?
CS61c Midterm Spring 96 1
