i'm a big fan of nebula and want to use as many instances as possible on my mac quadcore 3Ghz 13GB Ram in logic 9. Is Cuda currently a valid solution for running many instances of nebula simultaneously? What card do i need at least for it to make sense. Does cuda use the ram of my mac? Would i have to decide between using all instances native or cuda or is it possible to mix all the available solutions?
Hello musicgreator. CUDA toolkit is used to use GPU instead CPU, emulation presets are loaded into graphic card RAM instead PC RAM through AA CUDA bridge. Right now only a part of the Nebula code was ported into CUDA so vectorial engine is still processed into CPU. CUDA is a good solution for run reverbs or other long kernels emulation presets like high end EQs. Due the CUDA is a bridging technology between CPU and GPU is not able to run in very low latency mode now.
Due AU protocol not allow to create a "S*****" copies Nebula like into VST I don't recommend to use CUDA or you will lost compatibility to load emulation presets that are not compatible with CUDA.
Enrique Silveti. Acustica Audio customer and technical support.
Thanks for the reply. Will there be a solution for multiple au instances for different purposes? Wouldn't it be possible from the programmers side, to offer multiple audio units with different names or numbers and the user would edit the xml for his needs. For example nebula1, nebula2... and the user would know, nebula 2 is cuda, nebula3 is server etc...