Date post: | 02-Jun-2018 |
Category: |
Documents |
Upload: | dilshad-sabir |
View: | 213 times |
Download: | 0 times |
of 25
8/11/2019 cse431-06mipspipeline
1/25
CSE431 L06 Basic MIPS Pipelining.1 Irwin, PSU, 2005
CSE 431Computer ArchitectureFall 2005
Lecture 06: Basic MIPS Pipelining Review
Mary Jane Irwin ( www.cse.psu.edu/~mji)
www.cse.psu.edu/~cg431
[Adapted from Computer Organization and Design,
Patterson & Hennessy, 2005, UCB]
http://www.cse.psu.edu/~mjihttp://www.cse.psu.edu/~http://www.cse.psu.edu/~http://www.cse.psu.edu/~mji8/11/2019 cse431-06mipspipeline
2/25
CSE431 L06 Basic MIPS Pipelining.2 Irwin, PSU, 2005
Review: Single Cycle vs. Multiple Cycle Timing
Clk Cycle 1
Multiple Cycle Implementation:
IFetch Dec Exec Mem WB
Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7 Cycle 8 Cycle 9
Cycle 10
IFetch Dec Exec Mem
lw sw
IFetch
R-type
Clk
Single Cycle Implementation:
lw sw Waste
Cycle 1 Cycle 2
multicycle clock
slower than 1/5thofsingle cycle clockdue to stage registeroverhead
8/11/2019 cse431-06mipspipeline
3/25
CSE431 L06 Basic MIPS Pipelining.3 Irwin, PSU, 2005
How Can We Make It Even Faster?
Split the multiple instruction cycle into smaller and
smaller steps There is a point of diminishing returns where as much time is
spent loading the state registers as doing the work
Start fetching and executing the next instruction before
the current one has completed Pipelining(all?) modern processors are pipelined for
performance
Remember theperformance equation:
CPU time = CPI * CC * IC
Fetch (and execute) more than one instruction at a time
Superscalar processingstay tuned
8/11/2019 cse431-06mipspipeline
4/25
CSE431 L06 Basic MIPS Pipelining.4 Irwin, PSU, 2005
A Pipelined MIPS Processor
Start the nextinstruction before the current one hascompleted
improves throughput- total amount of work done in a given time
instruction latency(execution time, delay time, response time -time from the start of an instruction to its completion) is notreduced
Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5
IFetch Dec Exec Mem WBlw
Cycle 7Cycle 6 Cycle 8
sw IFetch Dec Exec Mem WB
R-type IFetch Dec Exec Mem WB
- clock cycle (pipeline stage time) is limited by the slowest stage
- for some instructions, some stages are wastedcycles
8/11/2019 cse431-06mipspipeline
5/25
CSE431 L06 Basic MIPS Pipelining.5 Irwin, PSU, 2005
Single Cycle, Multiple Cycle, vs. Pipeline
Multiple Cycle Implementation:
Clk
Cycle 1
IFetch Dec Exec Mem WB
Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7 Cycle 8 Cycle 9
Cycle 10
IFetch Dec Exec Mem
lw sw
IFetch
R-type
lw IFetch Dec Exec Mem WB
Pipeline Implementation:
IFetch Dec Exec Mem WBsw
IFetch Dec Exec Mem WBR-type
Clk
Single Cycle Implementation:
lw sw Waste
Cycle 1 Cycle 2
8/11/2019 cse431-06mipspipeline
6/25
CSE431 L06 Basic MIPS Pipelining.6 Irwin, PSU, 2005
MIPS Pipeline Datapath Modifications What do we need to add/modify in our MIPS datapath?
State registersbetween each pipeline stage to isolatethem
ReadAddress
Instruction
Memory
Add
PC
4
Write Data
Read Addr 1
Read Addr 2
Write Addr
Register
File
ReadData 1
ReadData 2
16 32
ALU
Shift
left 2
Add
Data
Memory
Address
Write Data
Read
DataIFetch/Dec
De
c/Exec
Exe
c/Mem
Mem/WB
IF:IFetch ID:Dec EX:Execute MEM:
MemAccess
WB:
WriteBack
System Clock
Sign
Extend
8/11/2019 cse431-06mipspipeline
7/25CSE431 L06 Basic MIPS Pipelining.7 Irwin, PSU, 2005
Pipelining the MIPS ISA
What makes it easy all instructions are the same length (32 bits)
- can fetch in the 1ststage and decode in the 2ndstage
few instruction formats (three) with symmetryacross formats
- can begin reading register file in 2ndstage
memory operations can occur only in loads and stores
- can use the execute stage to calculate memory addresses
each MIPS instruction writes at most one result (i.e.,changes the machine state) and does so near the end of thepipeline (MEM and WB)
What makes it hard structural hazards: what if we had only one memory?
control hazards: what about branches?
data hazards: what if an instructions input operands dependon the output of a previous instruction?
8/11/2019 cse431-06mipspipeline
8/25CSE431 L06 Basic MIPS Pipelining.8 Irwin, PSU, 2005
Graphically Representing MIPS Pipeline
Can help with answering questions like:
How many cycles does it take to execute this code?
What is the ALU doing during cycle 4?
Is there a hazard, why does it occur, and how can it be fixed?
ALUIM Reg DM Reg
8/11/2019 cse431-06mipspipeline
9/25CSE431 L06 Basic MIPS Pipelining.9 Irwin, PSU, 2005
Why Pipeline? For Performance!
I
n
s
t
r.
O
r
d
e
r
Time (clock cycles)
Inst 0
Inst 1
Inst 2
Inst 4
Inst 3
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
Once thepipeline is full,one instruction
is completedevery cycle, so
CPI = 1
Time to fill the pipeline
8/11/2019 cse431-06mipspipeline
10/25CSE431 L06 Basic MIPS Pipelining.10 Irwin, PSU, 2005
Can Pipelining Get Us Into Trouble?
Yes: Pipeline Hazards
structural hazards: attempt to use the same resource by twodifferent instructions at the same time
data hazards: attempt to use data before it is ready
- An instructions source operand(s) are produced by a priorinstruction still in the pipeline
control hazards: attempt to make a decision about programcontrol flow before the condition has been evaluated and thenew PC target address calculated
- branch instructions
Can always resolve hazards by waiting
pipeline control must detect the hazard
and take action to resolve hazards
8/11/2019 cse431-06mipspipeline
11/25CSE431 L06 Basic MIPS Pipelining.11 Irwin, PSU, 2005
I
n
s
t
r.
O
r
d
e
r
Time (clock cycles)
lw
Inst 1
Inst 2
Inst 4
Inst 3
ALU
Mem Reg Mem Reg
ALU
Mem Reg Mem Reg
ALU
Mem Reg Mem Reg
ALU
Mem Reg Mem Reg
ALU
Mem Reg Mem Reg
A Single Memory Would Be a Structural Hazard
Reading data frommemory
Reading instructionfrom memory
Fix with separate instr and data memories (I$ and D$)
8/11/2019 cse431-06mipspipeline
12/25CSE431 L06 Basic MIPS Pipelining.13 Irwin, PSU, 2005
How About Register File Access?
I
n
s
t
r.
O
r
d
e
r
Time (clock cycles)
Inst 1
Inst 2
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
Fix register fileaccess hazard bydoing reads in thesecond half of the
cycle and writes in
the first half
add $1,
add $2,$1,
clock edge that controlsregister writing
clock edge that controlsloading of pipeline state
registers
8/11/2019 cse431-06mipspipeline
13/25CSE431 L06 Basic MIPS Pipelining.15 Irwin, PSU, 2005
Register Usage Can Cause Data Hazards
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
Dependencies backward in time cause hazards
add $1,
sub $4,$1,$5
and $6,$1,$7
xor $4,$1,$5
or $8,$1,$9
Read before writedata hazard
8/11/2019 cse431-06mipspipeline
14/25
CSE431 L06 Basic MIPS Pipelining.16 Irwin, PSU, 2005
Loads Can Cause Data Hazards
I
n
s
t
r.
O
r
d
e
r
lw $1,4($2)
sub $4,$1,$5
and $6,$1,$7
xor $4,$1,$5
or $8,$1,$9
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALUIM Reg DM Reg
Dependencies backward in time cause hazards
Load-usedata hazard
8/11/2019 cse431-06mipspipeline
15/25
CSE431 L06 Basic MIPS Pipelining.17 Irwin, PSU, 2005
stall
stall
One Way to Fix a Data Hazard
I
n
s
t
r.
O
r
d
e
r
add $1,ALU
IM Reg DM Reg
sub $4,$1,$5
and $6,$1,$7
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
Can fix data
hazard bywaitingstallbut impacts CPI
8/11/2019 cse431-06mipspipeline
16/25
CSE431 L06 Basic MIPS Pipelining.19 Irwin, PSU, 2005
Another Way to Fix a Data Hazard
A
LU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
Fix data hazards
by forwardingresults as soon asthey are availableto where they are
needed
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
I
n
s
t
r.
O
r
d
e
r
add $1,
sub $4,$1,$5
and $6,$1,$7
xor $4,$1,$5
or $8,$1,$9
8/11/2019 cse431-06mipspipeline
17/25
CSE431 L06 Basic MIPS Pipelining.21 Irwin, PSU, 2005
Forwarding with Load-use Data Hazards
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALUIM Reg DM Reg
Will still need one stall cycleeven with forwarding
I
n
s
t
r.
O
r
d
e
r
lw $1,4($2)
sub $4,$1,$5
and $6,$1,$7
xor $4,$1,$5
or $8,$1,$9
8/11/2019 cse431-06mipspipeline
18/25
CSE431 L06 Basic MIPS Pipelining.22 Irwin, PSU, 2005
Branch Instructions Cause Control Hazards
I
n
s
t
r.
O
r
d
e
r
lw
Inst 4
Inst 3
beqALUIM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
ALU
IM Reg DM Reg
Dependencies backward in time cause hazards
8/11/2019 cse431-06mipspipeline
19/25
CSE431 L06 Basic MIPS Pipelining.23 Irwin, PSU, 2005
stall
stall
stall
One Way to Fix a Control Hazard
In
s
t
r.
Or
d
e
r
beq
A
LUIM Reg DM Reg
lw
AL
UIM Reg DM Reg
ALUInst 3
IM Reg DM
Fix branch
hazard bywaiting
stallbutaffects CPI
8/11/2019 cse431-06mipspipeline
20/25
CSE431 L06 Basic MIPS Pipelining.25 Irwin, PSU, 2005
Corrected Datapath to Save RegWrite Addr Need to preserve the destination register address in
the pipeline state registers
ReadAddress
Instruction
Memory
Add
PC
4
Write Data
Read Addr 1
Read Addr 2
Write Addr
Register
File
ReadData 1
ReadData 2
16 32
ALU
Shiftleft 2
Add
Data
Memory
Address
Write Data
ReadData
IF/ID
Sign
Extend
ID/EX EX/MEM
MEM/WB
8/11/2019 cse431-06mipspipeline
21/25
CSE431 L06 Basic MIPS Pipelining.26 Irwin, PSU, 2005
MIPS Pipeline Control Path ModificationsAll control signals can be determined during Decode
and held in thestate registersbetween pipeline stages
ReadAddress
Instruction
Memory
Add
PC
4
Write Data
Read Addr 1
Read Addr 2
Write Addr
Register
File
ReadData 1
Read
Data 2
16 32
ALU
Shift
left 2
Add
Data
Memory
Address
Write Data
ReadData
IF/ID
Sign
Extend
ID/EX
EX/MEM
MEM/WB
Control
8/11/2019 cse431-06mipspipeline
22/25
CSE431 L06 Basic MIPS Pipelining.27 Irwin, PSU, 2005
Other Pipeline Structures Are Possible
What about the (slow) multiply operation?
Make the clock twice as slow or
let it take two cycles (since it doesnt use the DM stage)
ALU
IM Reg DM Reg
MUL
ALU
IM Reg DM1 RegDM2
What if the data memory access is twice as slow asthe instruction memory?
make the clock twice as slow or
let data memory access take two cycles (and keep the sameclock rate)
8/11/2019 cse431-06mipspipeline
23/25
CSE431 L06 Basic MIPS Pipelining.28 Irwin, PSU, 2005
Sample Pipeline Alternatives
ARM7
StrongARM-1
XScale
A
LUIM1 IM2 DM1
Reg
DM2
IM Reg EX
PC updateIM access
decodereg
access
ALU opDM accessshift/rotatecommit result
(write back)AL
UIM Reg DM Reg
Reg SHFT
PC updateBTB access
start IM access
IM access
decodereg 1 access
shift/rotatereg 2 access
ALU op
start DM accessexception
DM writereg write
8/11/2019 cse431-06mipspipeline
24/25
CSE431 L06 Basic MIPS Pipelining.29 Irwin, PSU, 2005
Summary
All modern day processors use pipelining
Pipelining doesnt help latencyof single task, it helpsthroughputof entire workload
Potential speedup: a CPI of 1 and fast a CC
Pipeline rate limited by slowestpipeline stage
Unbalanced pipe stages makes for inefficiencies
The time to fill pipeline and time to drain it can impactspeedup for deep pipelines and short code runs
Must detect and resolve hazards
Stalling negatively affects CPI (makes CPI less than the idealof 1)
8/11/2019 cse431-06mipspipeline
25/25
CSE431 L06 Basic MIPS Pipelining 30 Irwin PSU 2005
Next Lecture and Reminders
Next lecture
Overcoming data hazards
- Reading assignmentPH, Chapter 6.4-6.5
Reminders
HW2 due September 29th
SimpleScalar tutorials scheduled
- Thursday, Sept 22, 5:30-6:30 pm in 218 IST
Evening midterm exam scheduled
- Tuesday, October 18th, 20:15 to 22:15, Location 113 IST- You should have let me know by now if you have a conflict