Hello , I posted in the Nebula forum and would like to pose this question in regards to Nebula Man.
I have the luxury to spend time testing some equipment for audio mastering. Said that, I am interested in pushing pure convolution (Ltimed)to its limits (beyond what current CPU's could handle through Nebula3). One example would be a Nebula3 Reverb, processing seconds in Timed mode across multiple Kernels). Giancarlo, I know that the NebulaReverb.XML can be edited for 100msec but can this be extended to multiple seconds?
This Article raises an interesting issue not only for Nebula and NAT but also for NebulaMan. 1. Does NebulaMan detect the proper thread count in a multi-CPU setup and is it capable to take advantage of all CPU threads to its fullest? 2.Is there a means to set the thread count manually in NebulaMan in case the automatic detection has failed?
4. What is the limit of threads NebulaMan can use(4 CPU - 4 core - 64 threads)?
5. I also would like to put in a word for 192khz as well as higher sample-rates although i read that you are about to implement that capability.
NebulaMan is only a host which hosts Nebula. Pure processing done by NebulaMan (without Nebula itself) is very minor. NebulaMan uses/controls only few threads (GUI, preview and render) and I think there is not much more to do here.
The main multithreading 'magic' is happening inside Nebula (ACQUA library) itself. Nebula is the one, which actually puts a strain onto CPU(s) and distribute its processing tasks among multiple threads etc. Giancarlo is the first person who can tell, if there is any room for your suggested improvements