Overview

Authors: Justin Mendes and Shazil Razzaq

Tests the functionality of your RV32UI processor by executing test_pd with Verilator and verifying the results with a mock RV32UI processor created in Python.

1. Runs all of the *.x tests in the benchmarks (rv32-benchmarks).

2. Verifies all of the resulting *.trace files in ../sim/verilator/test_pd/ for behaviour

TLDR;

1. Put test.py and parse.py in /projects/pd4/verif/scripts

2. In /project/pd4/verif/scripts enter python3 test.py <PATH to /rv32-benchmarks>

If your output looks like this, you're processor is likely good to go!

Otherwise, read on for debugging instructions.

Setup/Configuration

1. Place test.py and parse.py in the ./verif/scripts directory of the project you want to test (e.g. ece320/projects/pd4/verif/scripts).

2. Have local access to /rv32-benchmarks found in the ece320 GitLab.

For an end-to-end test follow instructions for test.py.

For testing individual programs follow instructions for parse.py

test.py

Runs a full end-to-end test of your processor.

Default Behaviour: Executes and Verifies all tests.

Assumptions Made:

1. Your /rv32-benchmarks directory has subfolders: /individual-instructions and /simple-programs.

2. *.x files in /individual-instructions begin with: rv32ui-p-

3. Your results - *.trace files, are in directory ../sim/verilator/test_pd

Format

python3 test.py [-h/--help] [-e/--execute] [-v/--verify] [any flags/options for parse.py...] <PATH: to "/rv32-benchmarks" directory>

Must be ran in /projects/pd#/verif/scripts/ directory

Arguments

PATH: Required.

Path to "/rv32-benchmarks" directory.

Flags

Optional flags to help with debugging:

1. -h/--help:

Get more information how to run the command and what each of the arguments and options/flags do.

2. -e/--execute:

Add this flag to EXECUTE the tests only (runs all .x tests at the specified path).

3. -v/--verify:

Add this flag to VERIFY the trace files of the tests only (runs parse.py on all trace files in "../sim/verilator/test_pd/*.trace").

Other Flags (Forwarded)

Flags forwarded to parse.py:

1. -s/--skip
2. -i/--instructions
3. -in/--instructions-no-mem
4. -r/--regfile
5. -m <MEM_DEPTH>

Examples

python3 test.py ../../../../../rv32-benchmarks -in -s

python3 test.py -e ../../../../../rv32-benchmarks

python3 test.py -i -r ../../../../../rv32-benchmarks --verify

python3 test.py -v --skip -m 1024000 /home/ece320/rv32-benchmarks -i

python3 test.py --execute ../../../../../rv32-benchmarks

python3 test.py ../../../../../rv32-benchmarks --execute

parse.py

Compares the results of your processor's *.trace output file under a *.x testbench to a mock Python RV32UI processor to verify correctness of your processor under the given *.x test.

Assumptions Made:

1. Your dmemory is initialized with the same data as imemory.

2. Your reg_file (i.e. each register x0-x31) is initialized with all 0s.

Format

python3 parse.py [-h/--help] [-s/--skip] [-i/--instructions] [-in/--instructions-no-num] [-r/--regfile] [-m <MEM_DEPTH> / --mem <MEM_DEPTH>] <TRACE: path to .trace file> <HEX: path to corresponding .x file>

Arguments

TRACE: Required.

Path to .trace file from executing a *.x testbench: make run VERILATOR=1 TEST=test_pd MEM_PATH=*.x.

HEX: Required.

Path to .x file used to generate the *.trace file: make run VERILATOR=1 TEST=test_pd MEM_PATH={HEX}.

The contents of this file will be initially loaded into dmemory.

Flags

1. -s/--skip: Default False

CONTINUE execution on errors.

2. -i/--instructions: Default: False

Print all instructions in RISC-V format (instructions enumerated).

3. -in/--instructions-no-num: Default: False

Print all instructions in RISC-V format (instructions NOT enumerated).

4. -r/--regfile: Default: False

Print the state of the Register File at each [R] in the .trace file.

5. -m <MEM_DEPTH>/--mem <MEM_DEPTH>: Default: -m 1048576

Number of bytes (8-bit values) in data memory (memory).

Examples

python3 parse.py ../sim/verilator/test_pd/rv32ui-p-sltiu.trace ../../../../../rv32-benchmarks/individual-instructions/rv32ui-p-sltiu.x -in -s

python3 parse.py -i -r ../data/rv32ui-p-add.trace ../../../../../rv32-benchmarks/individual-instructions/rv32ui-p-add.x

python3 parse.py --skip -m 1024000 ../tests/out/BubbleSort.trace -i /rv32-benchmarks/simple-programs/BubbleSort.x

Future Versions

v1.x

1. Add support for flags -pd1 -pd2 -pd3 -pd4 -pd5 to allow user to isolate which project they would like to test up to.