hello fellows,

at the moment I´m working on a rather big sound-synthesis-module, which, for the sake of complexity, drives my machine towards its limitation. basically everything´s clean and beautiful as long as i run just a single instance of the module (about 45% cpu-consumption), but for some reasons it would be interesting or even necessary to run up to 8 instances at the same time.
Unfortunately it is no real option to change or reduce the module any further, so now I´m wondering if there could be any advantage in using another programming-language for the core of the sound-synthesis.
I´m not deep enough into this whole CPU-usage-efficiency-whatever-thing, I just imagine, that maybe CSound for example "runs a little closer to the hardware" and could therefore be a bit less grabby. Probably this is too naive and/or just stupid, so I would be thankful for any thought of you that could diminish my confusion.

longstoryshortquestion:
Is there any chance for a complex additive synthesis in CSound to be less CPU-consuming than in Pd?

I haven't used Csound in a long time. Both are written in C, though, so they should more or less be on the same level to the hardware as each other. But one of the advantages it has CPU-wise is that it has a ton of opcodes (which is Csound's version of "objects" or "externals"). A lot of those opcodes are higher level than what is offered in Pd-vanilla (and, to a lesser extent, Pd-extended), which basically means that it does in one opcode what it takes for several objects in Pd to do.

This isn't so much "runs closer to the hardware" as it is "less overhead". As an example, when a block of audio samples is passed from object to object, there is an initialization phase that happens before the audio is processed. When you take several objects and essentially re-write them as a single object, you only have to run that initialization once, reducing some redundancy.

At any rate, I've heard that the object-oriented approach of SuperCollider makes things like additive synthesis a breeze.