Page 1 of 1

Bullet acceleration with FPGA

Posted: Mon Apr 23, 2018 8:23 pm
by ptoupas
Hello,

I am currently working on my diploma thesis, aiming to accelerate some sectors of the bullet physics library using
hardware(more specifically using an FPGA). I am focussing only on the computational part of the library and not actually
changing any of the graphic algorithms. I would like to know if there are any realistic scenarios-examples of
this library where you think that it might be usefull to implement such a modification to decrease the computational time.
Which do you think that might be some sectors/functions/algorithms in the library which in a realistic scenario would
take a long time to execute in software (but with the use of an FPGA their execution time could be significant better)?
At the moment i am working with a scene that has 8000 rigid bodies colliding but i dont know if this is realistic or not.How many Rigid Bodies does an actual scene(a gaming scene for example) has on average?

Re: Bullet acceleration with FPGA

Posted: Tue Jan 15, 2019 3:37 am
by teolazza
Simulation environments can use it. They often need as many frames per second as possible in order to accelerate the speed of the simulation, so scaling is important.