Sequential Implementation CS:APP Chapter 4 Computer Architecture

advertisement
CS:APP Chapter 4
Computer Architecture
Sequential
Implementation
Randal E. Bryant
Carnegie Mellon University
http://csapp.cs.cmu.edu
CS:APP
Y86 Instruction Set
Byte
0
nop
0
0
halt
1
0
rrmovl rA, rB
2
0 rA rB
irmovl V, rB
3
0
8 rB
V
rmmovl rA, D(rB)
4
0 rA rB
D
mrmovl D(rB), rA
OPl rA, rB
jXX Dest
call Dest
ret
pushl rA
popl rA
–2–
5
1
2
3
0 rA rB
4
5
addl
6
0
subl
6
1
andl
6
2
xorl
6
3
jmp
7
0
jle
7
1
jl
7
2
je
7
3
jne
7
4
jge
7
5
jg
7
6
D
6 fn rA rB
7 fn
8
9
A
B
0
Dest
Dest
0
0 rA 8
0 rA 8
CS:APP
Building Blocks
fun
Combinational Logic






–3–
Store bits
Addressable memories
Non-addressable registers
Loaded only as clock rises
=
A
L
U
Compute Boolean functions of
inputs
B
Continuously respond to input
changes
Operate on data and implement
control
Storage Elements

A
0
MUX
1
valA
srcA
A
valW
Register
file
valB
srcB
B
W
dstW
Clock
Clock
CS:APP
Hardware Control Language


Very simple hardware description language
Can only express limited aspects of hardware operation
 Parts we want to explore and modify
Data Types

bool: Boolean
 a, b, c, …

int: words
 A, B, C, …
 Does not specify word size---bytes, 32-bit words, …
Statements
–4–

bool a = bool-expr ;

int A = int-expr ;
CS:APP
HCL Operations

Classify by type of value returned
Boolean Expressions

Logic Operations
 a && b, a || b, !a

Word Comparisons
 A == B, A != B, A < B, A <= B, A >= B, A > B

Set Membership
 A in { B, C, D }
» Same as A == B || A == C || A == D
Word Expressions

Case expressions
 [ a : A; b : B; c : C ]
 Evaluate test expressions a, b, c, … in sequence
 Return word expression A, B, C, … for first successful test
–5–
CS:APP
SEQ Hardware
Structure
newPC
PC
valE, valM
Write back
valM
State

Program counter register (PC)
Condition code register (CC)
Register File

Memories


 Access same memory space
Data
Data
memory
memory
Memory
Addr, Data
valE
Bch
Execute
CC
CC
aluA, aluB
 Data: for reading/writing program
data
 Instruction: for reading
instructions
Instruction Flow

Read instruction at address
specified by PC
Process through stages

Update program counter

ALU
ALU
valA, valB
srcA, srcB
dstA, dstB
Decode
A
B
Register
RegisterM
file
file
E
icode ifun
rA , rB
valC
valP
,
Fetch
Instruction
Instruction
memory
memory
PC
PC
increment
increment
PC
–6–
CS:APP
newPC
SEQ Stages
PC
valE, valM
Write back
valM
Fetch

Read instruction from instruction
memory
Data
Data
memory
memory
Memory
Addr, Data
Decode

Read program registers
Execute

valE
Bch
Execute
CC
CC
aluA, aluB
Compute value or address
valA, valB
Memory

Read or write data
icode ifun
rA , rB
valC
B
valP
,
Fetch

A
Register
RegisterM
file
file
E
Write program registers
PC
srcA, srcB
dstA, dstB
Decode
Write Back

ALU
ALU
Instruction
Instruction
memory
memory
PC
PC
increment
increment
Update program counter
PC
–7–
CS:APP
Instruction Decoding
Optional
5
0 rA rB
Optional
D
icode
ifun
rA
rB
valC
Instruction Format



–8–
Instruction byte
icode:ifun
Optional register byte
rA:rB
Optional constant word valC
CS:APP
Executing Arith./Logical Operation
OPl rA, rB
Fetch

Memory
Read 2 bytes
Decode

Read operand registers
Execute
–9–
6 fn rA rB

Perform operation

Set condition codes

Do nothing
Write back

Update register
PC Update

Increment PC by 2
CS:APP
Stage Computation: Arith/Log. Ops
OPl rA, rB
icode:ifun  M1[PC]
Read instruction byte
rA:rB  M1[PC+1]
Read register byte
valP  PC+2
Compute next PC
valA  R[rA]
Read operand A
valB  R[rB]
Read operand B
valE  valB OP valA
Perform ALU operation
Set CC
Set condition code register
Memory
Write
R[rB]  valE
Write back result
back
PC update
PC  valP
Update PC
Fetch
Decode
Execute


– 10 –
Formulate instruction execution as sequence of simple
steps
Use same general form for all instructions
CS:APP
Executing rmmovl
rmmovl rA, D(rB) 4 0 rA rB
Fetch

Memory
Read 6 bytes
Decode

Read operand registers
Execute

– 11 –
D
Compute effective address

Write to memory
Write back

Do nothing
PC Update

Increment PC by 6
CS:APP
Stage Computation: rmmovl
rmmovl rA, D(rB)
Fetch
Decode
Execute
Memory
Write
back
PC update

– 12 –
icode:ifun  M1[PC]
Read instruction byte
rA:rB  M1[PC+1]
Read register byte
valC  M4[PC+2]
Read displacement D
valP  PC+6
Compute next PC
valA  R[rA]
Read operand A
valB  R[rB]
Read operand B
valE  valB + valC
Compute effective address
M4[valE]  valA
Write value to memory
PC  valP
Update PC
Use ALU for address computation
CS:APP
Executing popl
popl rA
Fetch

Memory
Read 2 bytes
Decode

Read stack pointer
Execute

b 0 rA 8
Increment stack pointer by 4

Write back


Update stack pointer
Write result to register
PC Update

– 13 –
Read from old stack pointer
Increment PC by 2
CS:APP
Stage Computation: popl
popl rA
icode:ifun  M1[PC]
Read instruction byte
rA:rB  M1[PC+1]
Read register byte
valP  PC+2
valA  R[%esp]
Compute next PC
valB  R [%esp]
Read stack pointer
valE  valB + 4
Increment stack pointer
Memory
Write
valM  M4[valA]
R[%esp]  valE
Read from stack
back
PC update
R[rA]  valM
Write back result
PC  valP
Update PC
Fetch
Decode
Execute


Read stack pointer
Update stack pointer
Use ALU to increment stack pointer
Must update two registers
 Popped value
– 14 –
 New stack pointer
CS:APP
Executing Jumps
jXX Dest
7 fn
fall thru:
XX XX
Not taken
target:
XX XX
Taken
Fetch


Memory
Read 5 bytes
Increment PC by 5
Decode

Do nothing
Execute

– 15 –
Dest
Determine whether to take
branch based on jump
condition and condition
codes

Do nothing
Write back

Do nothing
PC Update

Set PC to Dest if branch
taken or to incremented PC
if not branch
CS:APP
Stage Computation: Jumps
jXX Dest
Fetch
icode:ifun  M1[PC]
Read instruction byte
valC  M4[PC+1]
Read destination address
valP  PC+5
Fall through address
Bch  Cond(CC,ifun)
Take branch?
PC  Bch ? valC : valP
Update PC
Decode
Execute
Memory
Write
back
PC update


– 16 –
Compute both addresses
Choose based on setting of condition codes and branch
condition
CS:APP
Executing call
8 0
call Dest
return:
XX XX
target:
XX XX
Fetch
Memory

Read 5 bytes

Increment PC by 5
Decode

Read stack pointer
Execute

– 17 –
Dest
Decrement stack pointer by
4

Write incremented PC to
new value of stack pointer
Write back

Update stack pointer
PC Update

Set PC to Dest
CS:APP
Stage Computation: call
call Dest
icode:ifun  M1[PC]
Read instruction byte
valC  M4[PC+1]
Read destination address
valP  PC+5
Compute return point
valB  R[%esp]
Read stack pointer
valE  valB + –4
Decrement stack pointer
Memory
Write
M4[valE]  valP
R[%esp]  valE
Write return value on stack
back
PC update
PC  valC
Set PC to destination
Fetch
Decode
Execute


– 18 –
Update stack pointer
Use ALU to decrement stack pointer
Store incremented PC
CS:APP
Executing ret
9 0
ret
return:
XX XX
Fetch

Memory
Read 1 byte

Decode

Read stack pointer
Execute

Increment stack pointer by 4
Write back

Update stack pointer
PC Update

– 19 –
Read return address from
old stack pointer
Set PC to return address
CS:APP
Stage Computation: ret
ret
icode:ifun  M1[PC]
Read instruction byte
valA  R[%esp]
Read operand stack pointer
valB  R[%esp]
Read operand stack pointer
valE  valB + 4
Increment stack pointer
Memory
Write
valM  M4[valA]
R[%esp]  valE
Read return address
back
PC update
PC  valM
Set PC to return address
Fetch
Decode
Execute


– 20 –
Update stack pointer
Use ALU to increment stack pointer
Read return address from memory
CS:APP
Computation Steps
OPl rA, rB
Fetch
Decode
Execute
icode:ifun  M1[PC]
Read instruction byte
rA,rB
rA:rB  M1[PC+1]
Read register byte
valC
[Read constant word]
valP
valP  PC+2
Compute next PC
valA, srcA
valA  R[rA]
Read operand A
valB, srcB
valB  R[rB]
Read operand B
valE
valE  valB OP valA
Perform ALU operation
Cond code Set CC
Set condition code register
Memory
Write
valM
[Memory read/write]
back
PC update
dstM


– 21 –
icode,ifun
dstE
PC
R[rB]  valE
Write back ALU result
[Write back memory result]
PC  valP
Update PC
All instructions follow same general pattern
Differ in what gets computed on each step
CS:APP
Computation Steps
call Dest
icode,ifun
Fetch
Decode
Execute
Read instruction byte
[Read register byte]
valC
valC  M4[PC+1]
Read constant word
valP
valP  PC+5
Compute next PC
valA, srcA
[Read operand A]
valB, srcB
valB  R[%esp]
Read operand B
valE
valE  valB + –4
Perform ALU operation
Cond code
valM
back
PC update
dstM

– 22 –
rA,rB
Memory
Write

icode:ifun  M1[PC]
dstE
PC
[Set condition code reg.]
M4[valE]  valP
R[%esp]  valE
[Memory read/write]
[Write back ALU result]
Write back memory result
PC  valC
Update PC
All instructions follow same general pattern
Differ in what gets computed on each step
CS:APP
Computed Values
Fetch
Execute
icode
ifun
rA
Instruction code
Instruction function
Instr. Register A
rB
valC
valP
Instr. Register B
Instruction constant
Incremented PC


valE
Bch
ALU result
Branch flag
Memory

valM Value from
memory
Decode
– 23 –
srcA
srcB
dstE
dstM
valA
Register ID A
Register ID B
Destination Register E
Destination Register M
Register value A
valB
Register value B
CS:APP
newPC
SEQ Hardware
New
PC
PC
Key
valM
data out

Blue boxes:
predesigned hardware
blocks
read
write
Addr
 E.g., memories, ALU

Gray boxes:
control logic
Data
Data
memory
memory
Mem.
control
Memory
Execute
Bch
valE
CC
CC
ALU
ALU
ALU
A
Data
ALU
fun.
ALU
B
 Describe in HCL




– 24 –
White ovals:
labels for signals
Thick lines:
32-bit word values
Thin lines:
4-8 bit values
Dotted lines:
1-bit values
valA
Decode
A
valB
dstE dstM srcA
srcB
dstE dstM srcA
srcB
B
Register
Register M
file
file E
Write back
icode
Fetch
ifun
rA
rB
Instruction
Instruction
memory
memory
valC
valP
PC
PC
increment
increment
PC
CS:APP
Fetch Logic
icode
ifun
rA
rB
valC
valP
Need
valC
Instr
valid
Need
regids
Split
Split
PC
PC
increment
increment
Align
Align
Byte 0
Bytes 1-5
Instruction
Instruction
memory
memory
Predefined Blocks




– 25 –
PC
PC: Register containing PC
Instruction memory: Read 6 bytes (PC to PC+5)
Split: Divide instruction byte into icode and ifun
Align: Get fields for rA, rB, and valC
CS:APP
Fetch Logic
icode
ifun
rA
rB
valC
valP
Need
valC
Instr
valid
Need
regids
Split
Split
PC
PC
increment
increment
Align
Align
Byte 0
Bytes 1-5
Instruction
Instruction
memory
memory
PC
Control Logic



– 26 –
Instr. Valid: Is this instruction valid?
Need regids: Does this instruction have a register
bytes?
Need valC: Does this instruction have a constant word?
CS:APP
Fetch Control
Logic
nop
0
0
halt
1
0
rrmovl rA, rB
2
0 rA rB
irmovl V, rB
3
0
8 rB
V
rmmovl rA, D(rB)
4
0 rA rB
D
mrmovl D(rB), rA
5
0 rA rB
D
OPl rA, rB
6
fn rA rB
jXX Dest
7
fn
Dest
call Dest
8
0
Dest
ret
9
0
pushl rA
A
0 rA 8
popl rA
B
0 rA 8
bool need_regids =
icode in { IRRMOVL, IOPL, IPUSHL, IPOPL,
IIRMOVL, IRMMOVL, IMRMOVL };
bool instr_valid = icode in
{ INOP, IHALT, IRRMOVL, IIRMOVL, IRMMOVL, IMRMOVL,
IOPL, IJXX, ICALL, IRET, IPUSHL, IPOPL };
– 27 –
CS:APP
Decode Logic
valA
valB
A
B
valM valE
Register File



Read ports A, B
Write ports E, M
Addresses are register IDs or
8 (no access)
Register
Register
file
file
dstE
dstM
srcA
M
E
srcB
dstE dstM srcA srcB
Control Logic


– 28 –
srcA, srcB: read port
addresses
dstA, dstB: write port
addresses
icode
rA
rB
CS:APP
A Source
Decode
OPl rA, rB
valA  R[rA]
Read operand A
rmmovl rA, D(rB)
Decode
valA  R[rA]
Read operand A
popl rA
Decode
valA  R[%esp]
Read stack pointer
jXX Dest
Decode
No operand
call Dest
Decode
No operand
ret
Decode
valA  R[%esp]
Read stack pointer
int srcA = [
icode in { IRRMOVL, IRMMOVL, IOPL, IPUSHL
icode in { IPOPL, IRET } : RESP;
1 : RNONE; # Don't need register
];
– 29 –
} : rA;
CS:APP
E Destination
OPl rA, rB
Write-back R[rB]  valE
Write back result
rmmovl rA, D(rB)
Write-back
None
popl rA
Write-back R[%esp]  valE
Update stack pointer
jXX Dest
Write-back
None
call Dest
Write-back R[%esp]  valE
Update stack pointer
ret
Write-back R[%esp]  valE
Update stack pointer
int dstE = [
icode in { IRRMOVL, IIRMOVL, IOPL} : rB;
icode in { IPUSHL, IPOPL, ICALL, IRET } : RESP;
1 : RNONE; # Don't need register
];
– 30 –
CS:APP
Execute Logic
Units

ALU
 Implements 4 required functions
 Generates condition code values

Bch
CC
 Register with 3 condition code
bcond
bcond
bits

valE
bcond
ALU
fun.
ALU
ALU
CC
CC
 Computes branch flag
Set
CC
Control Logic




– 31 –
Set CC: Should condition code
register be loaded?
ALU A: Input A to ALU
ALU B: Input B to ALU
icode ifun
ALU
A
valC
ALU
B
valA
valB
ALU fun: What function should
ALU compute?
CS:APP
ALU A Input
Execute
OPl rA, rB
valE  valB OP valA
Perform ALU operation
rmmovl rA, D(rB)
Execute
valE  valB + valC
Compute effective address
popl rA
Execute
valE  valB + 4
Increment stack pointer
jXX Dest
Execute
No operation
call Dest
Execute
valE  valB + –4
Decrement stack pointer
ret
Execute
valE  valB + 4
Increment stack pointer
int aluA = [
icode in { IRRMOVL, IOPL } : valA;
icode in { IIRMOVL, IRMMOVL, IMRMOVL } : valC;
icode in { ICALL, IPUSHL } : -4;
icode in { IRET, IPOPL } : 4;
# Other instructions don't need ALU
– 32 –
];
CS:APP
ALU Operation
OPl rA, rB
Execute
valE  valB OP valA
Perform ALU operation
rmmovl rA, D(rB)
Execute
valE  valB + valC
Compute effective address
popl rA
Execute
valE  valB + 4
Increment stack pointer
jXX Dest
Execute
No operation
call Dest
Execute
valE  valB + –4
Decrement stack pointer
ret
Execute
– 33 –
valE  valB + 4
int alufun = [
icode == IOPL : ifun;
1 : ALUADD;
];
Increment stack pointer
CS:APP
Memory Logic
Memory

valM
Reads or writes memory
word
data out
Mem.
read
Mem.
write
Control Logic




– 34 –
Mem. read: should word be
read?
Mem. write: should word be
written?
Mem. addr.: Select address
Mem. data.: Select data
read
Data
Data
memory
memory
write
data in
Mem
addr
icode
valE
Mem
data
valA valP
CS:APP
Memory Address
OPl rA, rB
Memory
No operation
rmmovl rA, D(rB)
Memory
M4[valE]  valA
Write value to memory
popl rA
Memory
valM  M4[valA]
Read from stack
jXX Dest
Memory
No operation
call Dest
Memory
M4[valE]  valP
Write return value on stack
ret
Memory
valM  M4[valA]
Read return address
int mem_addr = [
icode in { IRMMOVL, IPUSHL, ICALL, IMRMOVL } : valE;
icode in { IPOPL, IRET } : valA;
# Other instructions don't need address
CS:APP
– 35 – ];
Memory Read
OPl rA, rB
Memory
No operation
rmmovl rA, D(rB)
Memory
M4[valE]  valA
Write value to memory
popl rA
Memory
valM  M4[valA]
Read from stack
jXX Dest
Memory
No operation
call Dest
Memory
M4[valE]  valP
Write return value on stack
ret
Memory
valM  M4[valA]
Read return address
bool mem_read = icode in { IMRMOVL, IPOPL, IRET };
– 36 –
CS:APP
PC Update Logic
PC
New PC

New
PC
Select next value of PC
icode
– 37 –
Bch
valC
valM
valP
CS:APP
PC
Update
OPl rA, rB
PC update
PC  valP
Update PC
rmmovl rA, D(rB)
PC update
PC  valP
Update PC
popl rA
PC update
PC  valP
Update PC
jXX Dest
PC update
PC  Bch ? valC : valP
Update PC
call Dest
PC update
PC  valC
Set PC to destination
ret
PC update
– 38 –
PC  valM
Set PC to return address
int new_pc = [
icode == ICALL : valC;
icode == IJXX && Bch : valC;
icode == IRET : valM;
1 : valP;
];
CS:APP
SEQ Operation
State
PC register
 Cond. Code register
 Data memory
 Register file
All updated as clock rises

Combinational
Logic
Read
Write
Data
Data
memory
memory
CC
CC
Read
Ports
Write
Ports
Register
Register
file
file
Combinational Logic


PC
0x00c

ALU
Control logic
Memory reads
 Instruction memory
 Register file
 Data memory
– 39 –
CS:APP
SEQ
Operation
#2
Combinational
Logic
CC
CC
100
100
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Clock
Cycle 1:
0x000:
irmovl $0x100,%ebx
# %ebx <-- 0x100
Cycle 2:
0x006:
irmovl $0x200,%edx
# %edx <-- 0x200
Cycle 3:
0x00c:
addl %edx,%ebx
# %ebx <-- 0x300 CC <-- 000
Cycle 4:
0x00e:
je dest
# Not taken
Write
Read

Data
Data
memory
memory
Write
Ports
Read
Ports
Register
Register
file
file

state set according to
second irmovl
instruction
combinational logic
starting to react to
state changes
%ebx
0x100
%ebx == 0x100
PC
0x00c
– 40 –
CS:APP
SEQ
Operation
#3
Combinational
Logic
CC
CC
100
100
000
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Clock
Cycle 1:
0x000:
irmovl $0x100,%ebx
# %ebx <-- 0x100
Cycle 2:
0x006:
irmovl $0x200,%edx
# %edx <-- 0x200
Cycle 3:
0x00c:
addl %edx,%ebx
# %ebx <-- 0x300 CC <-- 000
Cycle 4:
0x00e:
je dest
# Not taken
Write
Read

Data
Data
memory
memory
Write
Ports
Read
Ports
Register
Register
file
file

state set according to
second irmovl
instruction
combinational logic
generates results for
addl instruction
%ebx
0x100
%ebx == 0x100
0x00e
PC
0x00c
– 41 –
CS:APP
SEQ
Operation
#4
Cycle 1
Cycle 3
Cycle 4
Cycle 1:
0x000:
irmovl $0x100,%ebx
# %ebx <-- 0x100
Cycle 2:
0x006:
irmovl $0x200,%edx
# %edx <-- 0x200
Cycle 3:
0x00c:
addl %edx,%ebx
# %ebx <-- 0x300 CC <-- 000
Cycle 4:
0x00e:
je dest
# Not taken
Read
Write

Combinational
Logic
CC
CC
000
000
Cycle 2
Clock
Data
Data
memory
memory
Read
Ports
Write
Ports
Register
Register
file
file
%ebx
%ebx == 0x300
0x300

state set
according to addl
instruction
combinational
logic starting to
react to state
changes
PC
0x00e
– 42 –
CS:APP
SEQ
Operation
#5
Cycle 1
Cycle 3
Cycle 4
Cycle 1:
0x000:
irmovl $0x100,%ebx
# %ebx <-- 0x100
Cycle 2:
0x006:
irmovl $0x200,%edx
# %edx <-- 0x200
Cycle 3:
0x00c:
addl %edx,%ebx
# %ebx <-- 0x300 CC <-- 000
Cycle 4:
0x00e:
je dest
# Not taken
Read
Write

Combinational
Logic
CC
CC
000
000
Cycle 2
Clock
Data
Data
memory
memory
Read
Ports
Write
Ports
Register
Register
file
file
%ebx
%ebx == 0x300
0x300

state set
according to addl
instruction
combinational
logic generates
results for je
instruction
0x013
PC
0x00e
– 43 –
CS:APP
SEQ Summary
Implementation




Express every instruction as series of simple steps
Follow same general flow for each instruction type
Assemble registers, memories, predesigned combinational
blocks
Connect with control logic
Limitations




– 44 –
Too slow to be practical
In one cycle, must propagate through instruction memory,
register file, ALU, and data memory
Would need to run clock very slowly
Hardware units only active for fraction of clock cycle
CS:APP
Download