Processor Simulator

A detailed processor simulator that models the backend of a superscalar out-of-order processor with various functional units. Computer Assignment Report can be found at CA2-Report.pdf

Project Structure

Core Components

• processor.hpp/cpp: Main processor class implementing the pipeline stages
• procsim.hpp/cpp: Processor simulator interface and configuration
• procsim_driver.cpp: Main driver program
• dispatch_unit.hpp/cpp: Instruction dispatch unit implementation
• functional_unit.hpp/cpp: Functional unit implementation
• register_file.hpp/cpp: Register file implementation
• rob.hpp/cpp: Reorder buffer implementation
• schedule_queue.hpp/cpp: Schedule queue implementation
• update_state.hpp/cpp: State update logic implementation

Scripts

1. run_all.sh

   • Runs all traces with default configuration
   • Configuration: R=2, J=3, K=2, L=1, F=4
   • Compares outputs with golden outputs
   • Usage: ./run_all.sh

2. run_experiment.sh

   • Tests multiple parameter configurations
   • Parameter ranges:
     • R: 2, 4, 8 (Result Busses)
     • J: 1, 2, 3 (k0)
     • K: 1, 2, 3 (k1)
     • L: 1, 2, 3 (k2)
     • F: 2, 4, 8 (fetch width)
   • Organizes results in two ways:
     • By configuration: results/experiments/R2_J3_K2_L1_F4/
     • By trace: results/gcc/, results/gobmk/, etc.
   • Usage: ./run_experiment.sh

Build System

• Makefile: Build configuration
  • Build: make build
  • Clean: make clean
  • Run: make run
  • Run Debug: make run-debug
  • Run Log: make run-log
  • Run Output: make run-output
  • Run a specific trace make run TRACE=<trace_name>

Configuration Parameters

• -r: Number of Result Busses
• -j: Number of k0 Functional Units
• -k: Number of k1 Functional Units
• -l: Number of k2 Functional Units
• -f: Fetch Width
• -v: Enable Logging
• -d: Enable Debug
• -o: Enable Output

Traces

The simulator supports the following traces:

• gcc
• gobmk
• hmmer
• mcf

Output Organization

1. Configuration-based organization:

results/experiments/
└── R2_J3_K2_L1_F4/
    ├── gcc.output
    ├── gobmk.output
    ├── hmmer.output
    └── mcf.output

2. Trace-based organization:

results/
├── gcc/
│   ├── R2_J1_K1_L1_F2.output
│   ├── R2_J1_K1_L1_F4.output
│   └── ...
├── gobmk/
└── ...

Building and Running

1. Build the simulator:

make

2. Run all traces with default configuration and check against golden:

./run_all.sh

3. Run experiments with different configurations:

./run_experiment.sh

4. Run a specific trace:

./procsim -r2 -f4 -j3 -k2 -l1 < traces/gcc.100k.trace > results/gcc.output

Dependencies

• C++ compiler with C++11 support
• Make
• Bash shell

Notes

• The simulator implements a superscalar out-of-order processor
• Supports multiple functional units and pipeline stages
• Includes reorder buffer for maintaining instruction order
• Implements register renaming and dynamic scheduling
• Provides detailed performance statistics and pipeline state information