HACKWESTX Fall 2024 Challenge Problems
The Structural Simulation Toolkit (SST) is a parallel discrete event simulation infrastructure designed to support multi-reslution simulations. In this case, SST is most often utilized to simulate a variety of different computer architecture devices/components. Unlike other simulation infrastructures such as QEMU and Gem5, SST utilizes MPI (message passing interface) and C++ threading such that simulations can be scaled out to a large number of host systems. This permits us to build full-system simulations with high fidelity components for each discrete architectural unit.
One of the inherent downsides to building very large simulation experiments is the size and scope of the output data can be very large. SST provides distinct statistics output from each unique SST component in a variety of forms (CSV being the most common). For this challenge problem, we seek to ingest these large, structured CSV files into a web-based, graphical visualization and analysis infrastructure written in Python. We are providing SQLite3 database files that includes sample statistics output from sample simulations.
The difficulty in ingesting these data files into SST is that the particular components and included data will not be known until the filese are actually ingested. Further, we seek to cross correlate events and data values across disparate component data units in the visualization tool.
Be creative! Dynamic visualizations will be given priority over basic, static graphs. Additional points will be given to those that can build, run and demonstrate SST, especially if you can stream data directly into the visualization tool.
There are a number of basic requirements for this challenge, including:
- You must ingest and build the visualizations from the provided files. Note that you CANNOT build visualization templates beforehand. They must be generated dynamically.
- You MUST have a file selection utility so you can select different SQLite3 database files
- You MUST be able to select different graph types and different statistics
from the target database input file. EG,
SELECT stat1,stat2 from databasefrom a user input - Any external plugins utilized must be MIT, BSD or Apache licensed. No GPL-3 code will be accepted
- Jupyter notebooks tools/libraries CAN be used!
The deliverables for this challenge include:
- A working demo of your integrated visualization infrastructure. We must see it work!
- Any source code associated with the solution (scripts, etc)
- Concise documentation to build, integrate and test the solution. This should also include any required external packages outside of what is outlined above.
See the top-level LICENSE file
- John Leidel - Chief Scientist - Tactical Computing Labs