Gem5 Pipeline Visualizer example

Usage

1. Requirements

• Ubuntu 22.04
• A working RISC-V toolchain, preferably version 2023.10.18
• Our custom version of gem5
• The pipeline visualizer

2. Write your RISC-V program

Use the editor of your choice (nano, Visual Studio Code, gedit, ...) to write a RISC-V assembly file and save it. An example.s is provided in this repo

3. Compile it

Compile your example:

<path_to_your_riscv_toolchain>/riscv32-unknown-elf-gcc -march=rv32imf -mabi=ilp32 -mno-relax -nostartfiles <your_assembly_file> -o <output_executable_name>

If you are using the virtual machine or a LABINF PC, the riscv_compile script can be used:

riscv_compile <your_assembly_file>

It will produce an executable with the same name as the assembly file.

4. Run gem5

In order to run the simulation, you will need a Python configuration file. Use the gem5_config.py provided with this repository (remember to copy the common folder along with gem5_config.py if you are copying it in a different folder).

<path_to_gem5>/build/RISCV/gem5.opt
  --debug-flags=MinorGUI
  gem5_config.py
  --cpu-type MinorCPU
  --caches
  --l1d_size 8388608
  --l1i_size 8388608
  --cacheline_size 512
  --cpu-clock 1MHz
  --sys-clock 10GHz
  -c <output_executable_name> > <log_file_name>

Provide the executable file produced in the previous step and specify a log file. gem5_config.py must be used to specify pipeline parameters like functional unit latencies.

On the virtual machine and on LABINF PCs, the command gem5_run provides a shortcut to start the simulation. It takes as arguments the Python file containing the gem5 system configuration, the executable you want to run and the name of the log file where you want gem5's output to be. For example:

gem5_run gem5_config.py example example.log

5. Open the log file with the visualizer

Open the pipeline visualizer, click on File -> Open and select the log file generated in the previous step. On the virtual machine, the visualizer is available as a command gem5_pipeline_visualizer.