Related Experiment Videos
A standard timing benchmark for EGS4 Monte Carlo calculations
1Institute for National Measurement Standards, National Research Council of Canada, Ottawa.
Medical Physics
|March 1, 1992
Summary
A new Fortran 77 Monte Carlo code benchmarks electron and photon transport simulations. This standalone code offers flexible portability and efficient timing across diverse computing architectures, from PCs to supercomputers.
Area of Science:
- Computational physics
- Medical imaging simulation
- High-performance computing
Background:
- Accurate simulation of electron and photon transport is crucial for medical imaging, particularly in computed tomography (CT) slice analysis.
- Existing Monte Carlo systems like EGS4 require significant computational resources and setup.
- A need exists for a portable and efficient benchmarking tool for simulating energy deposition in 3D rectilinear geometries.
Purpose of the Study:
- To develop and evaluate a standalone Fortran 77 Monte Carlo source code for timing benchmark purposes.
- To assess the performance of this code across a wide range of 29 different computer systems.
- To demonstrate the code's capability in simulating electron beam energy deposition in 3D rectilinear geometries relevant to CT imaging.
Main Methods:
- Developed a Fortran 77 Monte Carlo code derived from the EGS4 Monte Carlo code system.
- Designed the code to simulate energy deposition from incident electron beams in 3D rectilinear geometry.
- Tested the standalone benchmark code on 29 distinct computer architectures, ranging from personal computers to a BBN TC2000 supercomputer.
Main Results:
- The Fortran 77 Monte Carlo code successfully simulated electron and photon transport, applicable to CT slice modeling.
- The benchmark code proved to be a standalone system, independent of the EGS4 installation.
- Execution times varied significantly, from approximately 5 hours on a PC/386/387 to mere seconds on a 512-processor BBN TC2000 supercomputer.
Conclusions:
- The developed Fortran 77 Monte Carlo code serves as an effective and portable tool for timing benchmarks in computational physics.
- The code's ability to run on diverse architectures highlights its adaptability for performance evaluation in electron and photon transport simulations.
- This benchmark provides valuable insights into the computational demands of simulating medical imaging processes across different hardware platforms.