Linear Acceleration of HDL Simulation Using Naive Parallel Processing

This webinar will address how multiprocessor simulation accelerates performance. The results show that when all CPU cores perform equally and the tasks are_ equally balanced and heavy enough_, simulation performance gain will be proportional to the number of used CPUs.

The push for multiprocessor simulation drives the wish for better performance. While software providers talk about “time to bug” or “time to full coverage”, engineers are interested in cycles per second. They see “oceans” of available CPU cores and the inherent parallel structure of HDL verification tasks as compelling reasons to ask for multiprocessor simulation options.

The basic ModelSim and Questa were used in this presentation to run basic Verilog test benches. The examples run on a multicore CPU and demand no “exotic” undocumented beta switches or licenses.

What you will learn:

• When multiprocessor simulation is useful
• How to avoid common pitfalls of parallel processing
• How to optimize the usage of available resources for parallel
  processing

Who should attend:

• Anyone who has heavy enough simulation to run
• Anyone who is into parallel processing
• Anyone concerned with verification of large systems at HDL level
• HDL verification managers
• Project managers/technical team leaders