Science Accessible via GPU-accelerated Planetary Dynamics Integrators

by Brett Gladman

Salle Denisse (Observatoire de Paris, Salle Denisse)

Salle Denisse

Observatoire de Paris, Salle Denisse

77 avenue Denfert-Rochereau, 75014 Paris.

The hardware characterstics of Graphical Processing Units (GPUs) have advanced to the point that two order of magnitude performance improvements can be obtained in problems that are of general utility. 
I will describe high-level design of the code GLISSE (GPU Longterm Integrator in Solar System Evolution), and use it to illustrate scientific exploration of the stability of orbits (1) between Uranus and Neptune, and (2) in Kuiper Belt resonances. 
For these applications GLISSE is 300-1000 times faster than a CPU core, and thus the hardware cost makes it vastly more financially effective  to use GPU based hardware rather than clusters of nodes.
We are extending the integrator's capabilities to handle small bodies encountering planets, handling such relatively rare encouters on CPU  cores while the GPU deals with huge numbers of test particles.  
I will discuss the application of this GLISSER code (the R standing for 'regularized') the the orbital evolution of the Kuiper Belt region under the existence of additional planets in the early Solar System.  

Organized by

Karine Briand, Florent Deleflie, Sébastien Lambert