Download 06_Branching & Shifting

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Branching & Shifting
CS 235
Computer Organization
&
Assembly Language
Two’s Complement
Branching Conditions
When the numbers are signed (two’s complement),
the branching conditions introduced earlier are
appropriate.
bl
branch <
0
(N xor V) = 1
ble
branch <= 0
Z or (N xor V)=1
be
branch == 0
Z=1
bne
branch != 0
Z=0
bge
branch >= 0
(N xor V) = 0
bg
branch >
0
Z or (N xor V)=0
How are Conditions Set?
Signed numbers
• Z is set when all bits of the result are 0
• N is set when the msb is 1
• V is set
– 1) when the register is not long enough to hold the result
– 2) on subtraction, d = m – s, when the sign of d is the
same as the sign of s and the sign of m is different from s
For example, most_negative – any_positive_number
(-8) – (2) = 10, not representable in 4 bits
Conditions with Unsigned Arithmetic
• No N flag!!!
• Condition codes set differently
• On addition, a carry out of the msb indicates overflow
and the carry bit C is set
• On subtraction. We imagine an extra bit to the left of
the number. By forming the two’s complement of the
subtrahend, and adding, we would expect a carry out of
the most significant bit to be added to the imaginary bit
to indicate a positive result. So the C bit is set if there is
no carry out of the msb.
• Result: the test for unsigned overflow is a test of the
carry bit C.
4 Bit Examples
1) 12
1100
09
1001
21
10101
On addition, the carry bit set indicates
overflow.
4 Bit Examples
2)
12
1100
- 09
0111 = (10000 –1 - 1001) +1
03 10011
and the carry indicates a correct result
4 Bit Examples
3) 09
- 12
21
1001
0100 = (10000 –1 - 1100) +1
01101
The carry bit is not set, so the result is
incorrect
Unsigned Branching Conditions
When the numbers are unsigned, the branching
conditions are different.
blu
branch <
0
C=1
bleu
branch <= 0
Z or C =1
be
branch == 0
Z=1
bne
branch != 0
Z=0
bgeu
branch >= 0
C=0
bgu
branch >
0
Z and C =0
Condition Code Tests
based directly upon condition codes
bneg
bpos
bz
bnz
bvs
bvc
bcs
bcc
branch on negative
branch on positive
branch on zero
branch on not zero
branch overflow set
branch no overflow
branch carry set
branch carry clear
N=1
N=0
Z=1
Z=0
V=1
V=0
C=1
C=0
Shifting
• Three shifts
– Shift right logical
srl
rs1, rs2_or_im, rdest
– Shift right arithmetic
sra
rs1, rs2_or_im, rdest
– Shift left logical
sll
rs1, rs2_or_im, rdest
The rs1 is shifted the numbers of bits in the second operand
and the result is stored in the destination
Shift is a Multiply by 2
• A fast multiply
• Used instead of .mul
• Consider %o0 * 11
1110 = 10112 = 8 + 2 + 1
sll
%o0, 3, %o1 ! 8*x
sll
%o0, 1, %o2 ! 2*x
add%o0, %o1, %o3 ! x + 8*x
add%o2, %o3, %o3 ! 8*x + 2*x + x
! = 11*x
SPARC Boolean Opcodes
Recall Boolean Operations (02:38)
andcc regs1, reg_or_imm, regdest
andncc regs1, reg_or_imm, regdest
xorcc regs1, reg_or_imm, regdest
orcc
regs1, reg_or_imm, regdest
xnorcc regs1, reg_or_imm, regdest
orncc regs1, reg_or_imm, regdest