Cpu Usage Vs Audio Quality, audio degradation in native systems Options

[bjkiwi]

has anyone noticed the sonic integrity of their audio tracks degrade as the cpu is made 2 work harder? Obviously if the cpu max's out it hits a brick wall and dies but is there a difference between, say, only using 25% of your cpu (maybe a couple of compressors) and 75-80% (bringing in those reverbs!)? Like, the audio quality is inversely proportional to the cpu usage.
I've often wondered whether there's a trade off in audio quality (reading and processing the audio off the hard drive) to get more plug-in's going.
Sometimes I think I've heard the sound 'soften' or lose it's edge or transparency and depth as the cpu meter winds up.
I'm interested if anyone has pondered the same question, noticed any difference, or knows the answer.
I'm running Cubase 5.0r2 on a G3 400 with waves plug-ins. (I know it's old and slow but I've got it humming quite nice...a bit like old motorbikes, they just need some attention and understanding, but I digress.)
thanks..Bjkiwi

[Teiwaz]

It's true that TDM doesn't degrade audio in the slightest, you just hit a brick wall and can't enable any more channels of I/O or plug-ins, when you run out of DSP on the PCI cards. Unfortunately, with TDM, I have discovered that I run out of DSP steam way faster than I do on the ASIO/VST system...


The situation is rather different on ASIO/VST/RTAS; experiencing a sort of "fizziness" and a softening of dynamic range when approaching the max CPU load on the host Mac's processor.


I find that in this situation I just bounce my virtual instruments (well, maybe just a few - the most CPU hungry ones like Battery, Absynth or Stylus) to hard disk and then run them as audio tracks. On a dual 1Ghz Mac this gives up to, in some cases, a good 60% of DSP power back to the CPU!


And then, a great deal of the compressor/reverb/eq/etc tasks are taken care of by the UAD-1 card's powered plug-ins - this configuration takes a great deal of stress off the Mac's processor too...

Of course, I do realise that I am kind of lucky to be the owner of a fast mac with a UAD-1 card, and that slower machines are more tricky when running large amounts of audio/instruments.

When bouncing an instrument to disk, I always solo the instrument that's being bounced, so that I have plenty of CPU reserve to ensure optimum audio quality during the bounce.

Also, I never actually record audio on the ASIO/VST mac. I record audio separately on a TDM based mac, so I can be sure of optimum record quality at all times. I feel that if I were forced to overdub & record additional audio tracks in a song that was already running hot on the CPU on the ASIO/VST system, that this would compromise the quality of the audio source actually being recorded.

I feel fortunate enough to be running dual systems...this provides flexibility and assurance of audio quality (both systems run Logic and are synchronized together, and I never have any major system problems (touch wood!)