[SOLVED] EECS 151/251A Lab 10 : MIPS Pipelining

1 Questions
1. Show or list all of the data dependencies in the following code fragment. Make
sure that you clearly indicate which instructions and registers are involved in each
dependency:
1 add $8 , $5 , $5
2 sub $8 , $8 , $10
3 lw $6 , 4 ( $8 )
4 add $4 , $6 , $8
1
Computer Architecture and Design, Lab 10 2
2. The following code has pipeline hazard(s) in it.
1 haz : add $5 , $0 , $0
2 lw $10 , 1 0 0 0 ( $20 )
3 addiu $20 , $20 , 4
4 addu $5 , $5 , $10
5 bne $20 , $0 , haz
Assuming you have a processor with the following specifications:
Five pipeline stages as seen in the lecture notes (IF, ID, AL, ME, WB).
Branch comparing done in the third stage.
Cannot fetch instruction until branch comparison is done.
Forwarding logic exists.
Register write and register read can happen in the same clock cycle.
Describe the dependencies that need to be resolved, and show how the code will
actually be executed (incorporating any stalls) so as to resolve the identified problems.
Computer Architecture and Design, Lab 10 3
2 Objective
The objective of this lab is to build and test the pipelined processor to test your program,
you should use the provided testbench. Your code should run the provided testbench
without errors. Ensure that your program is clearly commented. Once all tests in the
testbench pass make sure to demo to your TA.
3 Processor
In this lab, you will build the MIPS pipelined processor using Verilog. You will also simulate
some programs running on your pipelined processor. By the end of this lab, you should
thoroughly understand the internal operation of the MIPS pipelined processor.
The lab will involve several steps. First, you are given a template of the control module
in PipelinedControl.v that needs to be completed. You will also need modules from
previous labs as well as the provided PipelinedProc (found in PipelinedProc.v) modules.
Run the simulation in PipelinedProcTest.v to test your processor on real programs.
Additionally, you will need to complete the Hazard modules (given in Hazard.v).
4 Hazards
In this lab, you will need to design a module which handles both control and data hazards. Since a pipelined processor has several instructions in execution at once, consecutive
instructions could cause a hazard. Consider the following code segment:
1 lw $t0 , 0 ( $ t 2 )
2 add $a0 , $a0 , $ t 0
3 add $a2 , $a2 , $ t 0
4 beq $a0 , $0 , Lend
5 addi $a1 , $a2 , 4
The instruction on line 2 requires the instruction on line 1 to have completed before it can
be executed. Furthermore, the instruction on line 4 needs the instruction on line 2 to be
completed before it can be executed. These are both data hazards. The instruction on line
5 cannot start until the processor knows whether the branch will be taken or not. This is
a control hazard.
Both hazards will be handled in your hazard module. You can handle data hazards
through setting the Bubble signal to 1 for the correct number of clock cycles. In the above
code, 3 bubbles must be inserted between instructions 1 and 2, while only 2 are needed
between 3 and 4 (to handle the dependancy between instructions 2 and 4). Your hazard
module is given the list of registers both read and written by the current instruction
(registers will be 0 if not used). Your module will need to hold Bubble high for the
appropriate amount of time. You may find it useful to remember which register is being
written for the past 3 instructions, and see if the current instruction needs to read any of
Computer Architecture and Design, Lab 10 4
them. In such a case, Bubble will be 1, otherwise it will be 0. Remember, when you set
Bubble, the current instruction will not executed until Bubble is set to 0.
Control hazards can be handled by setting the PCWrite and IFWrite signals. When a
Branch instruction is decoded, PCWrite must be immediately set to 0, and held to 0 for
2 more clock cycles. For jump instructions, PCWrite must be immediately set to 0, and
held to 0 for 1 more clock cycle. IFWrite must be 0 for one additional clock cycle than
PCWrite, unless the instruction is a branch instruction, and the branch is not taken.
5 Deliverables
Once all tests have successfully completed and your design synthesizes, demo your progress
to the TA.
Please turn-in the following:
All of your verilog modules.
Filled out PDF Form.