accelerate NIOS for DSP

I wrote functions that draws shapes on a matrix.

it seems that in the original settings of the NIOS processor, it has problems working fast with this functions.

I guess it is the floating point calcs (please correct me if I'm wrong).

I'm trying to accelerate my NIOS processor in order to calculate on floating points faster.

are there any settings\techniuqes which can accelerate the NIOS.

I know that multiple DMAs can help, but I dont know how to set them and why would they help.

it would be best if I can make it behave like a GPU.

10x a lot!!!

Assaf