Differences between PGS ad TGS solvers

I was wondering if someone can elaborate on the differences between the PGS and TGS solvers available with Isaac Sim.
From my limited experience I see that with PGS I can get higher real-time factor from the simulator. How do they compare in terms of accuracy, efficiency? Do they have some limitations?
Thanks in advance

After having created the issue, I managed to find some documentation on the PhysX engine here.
Since this might be a topic of interest to many I just quote here what the documentation says (more details at the link) and answer myself in doing so:

Temporal gauss-seidel offers several advantages over the PGS-style solver:

  • Dramatically improved convergence
  • Improved handling of high-mass ratios
  • Minimizes energy introduced when correcting penetrations (depending on solver convergence)
  • Improved joint drive accuracy

TGS is generally a little slower than PGS. This is partially due to the increased complexity of the constraint solver and also partially due to TGS solving friction constraints every iteration, whereas PGS solves friction constraints only in the final 3 position iterations by default.



This is really good information, thanks for sharing @AndrePatri. I am curious what the future roadmap is for the Isaac team in terms of continuing support of both or focusing on one.

Our current plan is to continue supporting both PGS and TGS.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.