vishnuv0209 / be-sst

Coarse-grained system-level simulator developed at PSAAP II Center for Compressible Multiphase Turbulence at University of Florida

BE-SST

BE-SST is a parallel and scalable coarse-grained simulator that employs statistical models to run probabilistic, discrete-event simulation by combining Behavioral Emulation (BE), a coarse-grained modeling approach and Structural Simulation Toolkit (SST), a parallel discrete-event simulation framework from Sandia National Laboratories. BE-SST can be used for application design-space exploration (DSE) where alternate algorithms of an application can be studied on a given architecture; or architectural DSE can be performed for a given application. BE-SST provides a distributed parallel simulation library for Behavioral Emulation through simple interfaces and framework for development of coarse-grained BE models which can be easily extended to model new notional architectures. BE-SST also provides the capability to plug-n-play different subsystems (node, interconnect, memory, etc.), which is necessary to study abstract and notional machines.

Note: Both SST-core and SST-element-BE are provided to you through this repository. You DO NOT have to download the SST-core from the SST website.

Build and Run

This is a quick start guide for BE-SST. For more detailed and step-by-step instructions on how to build, install and run BE-SST, refer to DETAILED_BUILD_INSTALL.md.

Following dependencies are required to build and run BE-SST:

1. gcc 4.7 and newer
2. python 2 (anything in 2.7 branch)
   2.1 numpy package
   2.2 python-dev
3. GNU make 4.0 or newer
4. autotools (libtool, autoconf, automake)
5. OpenMPI 2.1.3 (strongly recommended)
6. Boost 1.56

Note: OpenMPI 2.1.3 and Boost 1.56 installation steps are listed DETAILED_BUILD_INSTALL.md

Example Build and Install Directories

These instructions will use the following conventions (the user can adjust these as they see fit):

• Download directory $HOME/scratch
  • This directory will contain downloaded source code packages for SST and its dependencies.
  • The following directories should be created on the users machine
    • $HOME/scratch
    • $HOME/scratch/src
• Download sst-core-devel and SST-elements-BE from the this GitHub repo inside the $HOME/scratch/src directory
• Installation directory $HOME/local
  • This directory will be the installation directory for SST and its dependencies.
  • The following directories should be created on the users machine:
    • $HOME/local
    • $HOME/local/packages

Basic Build and Install

1. Obtain sst-core-devel and sst-elements-BE from this github repo.
2. Place sst-core-devel and sst-elements-BE in $HOME/scratch/src directory.
3. Change directory to the SST-Core directory.

cd $HOME/scratch/src/sst-core-devel  

4. Run the autogen.sh script to setup the configure for SST-core

./autogen.sh  

5. Configure SST-Core.

./configure --prefix=$HOME/local/sst-core-devel  

6. Compile SST-Core.

make all  

7. Install SST-Core.

make install  

8. Change directory to sst-elements-BE directory

cd $HOME/scratch/src/sst-elements-BE  

9. Run the autogen.sh script to setup the configure for BE element.

./autogen.sh  

10. Configure BE element.

./configure --prefix=$HOME/local/sst-elements-BE --with-sst-core=$SST_CORE_HOME --with-boost=$BOOST_HOME  

11. Compile BE element.

make all  

12. Install BE element.

make install  

13. Verify SST Operation with a very simple sanity test (Note: Path to sst executable must be properly setup.)

sst --version  

14. Navigate to the tests folder

cd src/sst/elements/behavioralEmulation/tests/  

15. Running the test script for BE to check its functional correctness

./BE_TEST.sh  

16. Running a test configuration to obtain a simulated time

./BE_RUN.sh -c testConfig.py  

---

Refer to the following papers for more information on BE [1] and BE-SST [2]:

[1] Kumar N., Pascoe C., Hajas C., Lam H., Stitt G., George A. (2016) Behavioral Emulation for Scalable Design-Space Exploration of Algorithms and Architectures. In: Taufer M., Mohr B., Kunkel J. (eds) High Performance Computing. ISC High Performance 2016. Lecture Notes in Computer Science, vol 9945. Springer, Cham. https://link.springer.com/chapter/10.1007/978-3-319-46079-6_1
[2] Ajay Ramaswamy, Nalini Kumar, Aravind Neelakantan, Herman Lam, and Greg Stitt. 2018. Scalable Behavioral Emulation of Extreme-Scale Systems Using Structural Simulation Toolkit. In Proceedings of the 47th International Conference on Parallel Processing (ICPP 2018). Association for Computing Machinery, New York, NY, USA, Article 17, 1–11. DOI:https://doi.org/10.1145/3225058.3225124
If you consider using this work, please cite the above papers.

For any questions regarding the simulator, contact Aravind Neelakantan at [email protected]