PA-MIPS-Processor

Processor Description

How to compile?

make clean; 
make verilate 2>&1 |tee build.log; 

How to run?

./obj_dir/Vcore_tb 100

How to open the generated waveform?

gtkwave trace.fst

########################################## TESTS ##########################################

The verification of this project consisted on elaborating basic tests in order to ensure that the next features have been implemented properly:

GLOBAL
---------------------
- Processor boots from address 0x1000
- Exception handler is located on adress 0x2000

INSTRUCTION SET
---------------------
- R-type instructions work as expected.
- M-type instructions work as expected.
- Branches change the PC value properly.
- MUL instructions have 5 cycle latency

CACHE
----------------------
- Data Cache and Instruction Cache work as expected.
- Data Cache store buffer works as expected.
- Memory access has 10 cycles latency

BYPASS
----------------------
- Bypasses work as expected

EXCEPTIONS
----------------------
- Next exceptions are supported:
    - iTLB miss
    - fetch bus error  
    - illegal instruction
    - ALU overflow
    - MUL overflow
    - dTlb_miss        
    - cache bus error  
    - cache address fault

VIRTUAL MEMORY
----------------------
- iTLB and dTLB work as expected.
- Supervisor mode changes when entering on exception handler and when returning with IRET
  instruction.


Reorder Buffer
----------------------
- RoB works as expected.

Basic testing

The basic tests have been described on: PA-MIPS-Processor/tests/note

In order to run a test the next commands should be performed:

Choose the test: cd PA-MIPS-Processor/ cp tests/<selected_test> data_input_file.hex

Compile:

make clean; 
make verilate 2>&1 |tee build.log; 

Execute:

./obj_dir/Vcore_tb 100

Matrix Multiply test

Matrices A and B are generated randomly. Then they are used to compute matrix C (c = a x b). The python scripts can be easily modified to change the the randomization range:

- Generate matrices A and B:
    This createstwo hex files containing the values for matrices A and B: 
        PA-MIPS-Processor/tests/matrix_multiply/data_in_MxM_A.hex
        PA-MIPS-Processor/tests/matrix_multiply/data_in_MxM_B.hex
        PA-MIPS-Processor/tests/matrix_multiply/data_in_MxM_C.hex

    Command is:
        python matrixGenerator.py

- Compute matrix C golden data:
    This creates an hex file containing the values for matrix C: 
        PA-MIPS-Processor/tests/matrix_multiply/data_golden_MxM_C.hex
    
    Command is:
        python mxmValidator.py

- Run matrix multiply test on verilator:

    1. First, we have to define MATRIX_MULTIPLY_TEST (uncomment from rtl/inc/soc.vh)
    2. Then, we can execute the test;

        cd PA-MIPS-Processor/
        make clean; 
        make verilate 2>&1 |tee build.log;

    3. The resulting matrix C is written on:  
          
        PA-MIPS-Processor/matrix_multiply/verilator_matrix_C.hex

    4. We can just compare the computation of matrix C from python and verilator:
        
        cd PA-MIPS-Processor/tests/matrix_multiply
        diff verilator_matrix_C.hex data_golden_MxM_C.hex

//NOTE: In matrixGenerator you can indicate the maximum random number that the //matrixGenerator can assing to matrix a and b (randint(0, X)), if the number is to big //the mxmValidator will complain because there will be results bigger than 32 bits.

Waveform

Can be openned by running:

gtkwave trace.fst