Topic/Type: 2.2 Kinetic methods, Particle-In-Cell and Vlasov, Poster
P. Abreu1, 2, R. A. Fonseca1, 3, J. M. Pereira4, L. O. Silva1
1 IPFN/IST, Lisbon, Portugal
2 UNIDCOM/IADE, Lisbon, Portugal
3 DCTI/ISCTE, Lisbon, Portugal
4 IST/INESC-ID, Lisbon, Portugal
Modern graphic processors (GPU) can easily achieve performances of one TFLOP in single precision calculations and are thus an interesting source of cheap computing power. NVIDIA\'s CUDA architecture enables non-graphic applications to easily use that computing power. We have thus implemented a full relativistic PIC code on CUDA enabled devices using the CUDA API. We present some results from 2D simulations and a comparison against a similar state-of-the-art PIC code (Osiris 2.0). We have also implemented a real-time visualization feature so that the user can interact on the screen with the simulation, watching the virtual particles\' movement, enabling real-time diagnostic (EM field intensity, charge density, current), rotating, translating and zooming. We also present an evaluation of the time penalty for displaying and interacting with the simulation.