|
Title: Incomplete AI code Post by: charliemb on June 07, 2013, 06:23:22 pm Notice that I'm not jumping ahead and call this a bug. ...perhaps just incomplete coding on the part of the developer. ;)
So now I'm running mega-fast with my XXHighEnd PC and so of course now I want to try AI since I have an OS DAC. I find that AI works, but I also see that I only get one core working on AI interpolation. All other interpolation methods will use all virtual cores (12). But AI, only one. :( Why penalize those that want to use AI? (It turns out that AI sounds the best on at least some tracks and I'd like to use it more often) Would it be possible to turn on hyperthreading for AI? Please? :yes: Title: Re: Incomplete AI code Post by: PeterSt on June 07, 2013, 07:09:06 pm Small question Charlie :
Is this with FLAC files ? If WAV/AIF(F) that will be the reason. So, notice that loading WAV files (Arc Prediction) already goes more slow than loading FLAC files. This is because there's nothing to "multithread" in that case. Btw, I am not sure at all that AI can't or can do it - so maybe it cannot anyway. Also notice that -generally- using the Playback Drive also will hold back. This still is multithread "behavior", but it holds back. And since all the stupid combinations it well can be that AI holds back "per track" (like WAV does). Sorry for the counter questions, but maybe that is better than me looking for the unknown, which won't be today anymore anyway ... Kind regards, Peter Title: Re: Incomplete AI code Post by: charliemb on June 07, 2013, 11:39:00 pm I just tried various resolutions of flac and aif. I have very little in WAV.
With either flac or wav, only one yellow square lights up using AI, whereas I get 12 squares lighting up with ArcP or depreciated linear, or of course with no method selected. What's interesting is that even though only on square lights up with flac, it seems that high resolution flac loads a little faster than low resolution aif. I believe that I am using Playback drive, if that can make a difference. Playback drive is 7200 RPM SATA III. Title: Re: Incomplete AI code Post by: PeterSt on November 05, 2013, 06:16:04 pm Charlie,
I just tried to get the AI going multithreaded, but along the way I already saw that I made this so on purpose. However, I persisted and ... nope, all stalls. There's an internal multithreaded process going on somewhere and this actually fails (that is, I see no multiple cores getting active). And if *I* make it multithreaded at the higher level things start to fight and with zero % cpu after 10 minutes you better stop the loading. :yes: So I can't do much about this at the moment. But at least it works by (my) design. Peter |