Hi!
I am fairly new to CUDA and would be happy if somebody here could give me some advice on the following topic:
I currently trying to optimize parameters of a pretty complex Simulink model of a Fuel Cell hybrid car using evolutionary algorithms. Ideally i would like to evaluate the fitness of all populations members parallel on my GPU. I used Matlab’s RealTime Workshop to convert the Simulink mdl into C code.
Does anybody know if and how it is possible to convert this C code into CUDA? And if so, can the converted code be used as one kernel for the fintess evaluation of the population? Is CUDA and my GPU capable of evaluating complex system models? The input and output data is pretty small in terms of memory space, but the computation is expensive (about 20sec on my CPU). The population size is >1000.
Does anybody have some similar experience?
Thank you!
Holger