Hi, On Mon, 16 May 2016 03:05:13 Thor Lancelot Simon wrote: > > But given what you wrote below, wouldn't doing so coerce all streams to > 44100/16? It seems like it might be worth the extra work to avoid > resampling when all inputs/outputs are at the same, non 44.1/16 rate. > > Would this project benefit from the ability to use the floating point > unit, or vector extensions requiring the FP registers, in the kernel? > > Thor
Avoiding resampling may be possible if all streams were the same, I try and implement this. There would be benefits in using FPU registers as this would distribute the work between cpu and fpu I'll also look into this. Best regards, Nat.
