
the thing is, cuda is used. let it be openCL, i don´t know, but
rendering a small map the gpu is on 100%
but rendering a big map it is below of 30% @ clocks somewhere between turbo and idle!
we also could talk about multi gpu support ( i am not talking about SLI or C-fire stuff!)
just using the cores which are available.
the whole software with some exceptions on market can´t use multi core and this is sad truth!
oh here, some old threads about openCL effort by my side

http://forum.maniaplanet.com/viewtopic.php?f=32&t=4243
http://forum.maniaplanet.com/viewtopic.php?f=28&t=4997
br
