1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

New System, New Problems - Sample Rate and Bit Resolution

Discussion in 'Recording' started by Jeemy, Feb 21, 2005.

  1. Jeemy

    Jeemy Well-Known Member

    Well I just upgraded to the RME Fireface 800 from my crappy Edirol FA301.

    Firstly anybody searching, I cannot recommend the Fireface highly enough. The difference, mainly noticeable in the higher frequencies, and the clarity of what I can hear in old mixes that I never could before, is simply stunning.

    I finally feel I am moving towards a semi-professional level.

    But I could really do with some help on what resolutions to use and how to work out what performance level is best on the basis of dithering, what I can actually hear, and getting the best out of the mixing process by hearing the full resolution of my plugins (Waves Diamond).

    I am running a dual 1.8gHz G5 with 1256Mb RAM. Cubase SL, the RME, and some other bits and pieces.

    I have read through with great interest all of the the threads here on dithering, resolution, etc, and understand as much as it is possible at my skill and knowledge level, but would like some definite answers to the following, and some general advice please.

    I originally understood from reading here, that although recording at 48kHz would not offer me any advantages over 44.1 due to the dithering process having to perform too complex sums, recording at 88.2kHz would be a definite advantage and when dithering to 44.1, the conversion by half would be a simple enough sum for the L1 or L2 that I would get the advantages. However I am now wondering whether the performance issues of 88.2kHz are too much.

    Basically when the machine is set to record with 128 samples buffer I am getting latency of 7.256ms in and out, and a reduction to 4.354 using the Lower Latency function in Cubase.

    This at 88.2.

    However with the machine set to this, 88.2 and 32 bit fp, the VST Performance already hovers at about 10% before a project is even loaded.

    Dropping the buffer to 1024 reduces this to virtually zero with a little flicker, and clicking off the Lower Latency option zeros and flattens this.

    I seem to be able to record 10 tracks simultaneously stably enough even at 88.2, 32bit fp, and 4ms latency, with CPU usage of 30% peak.

    When I set up a fake mix I can get about 16 usages of TrueVerb before it starts peaking out. But only if I drop the buffer to 1024, otherwise I get 6. I can get 20 if I uncheck the Lower Latency buffer, stable at 60-65% CPU usage.

    Dropping the sample rate to 44.1 still at 32 bit fp, and 1024 samples, I am at 30% with 20 instances of TrueVerb.

    So I guess my questions are these:

    1) Given the fact I seem to be able to record okay at 88.2, is it worth seeing if I can run at this, and just dropping the latency when I am mixing, as 20 TrueVerbs is masses, so I should be able to run 30-40 mixed plugins. Or will I not get any benefit anyway so it is pointless.

    2) Should I go at 32 floating point? The RME processes internally at 24 bit so I am unsure what benefits this gives me - I understand it reduces the risk of clipping, and slight clipping I was not even hearing before at 16 or 24 bit is now becoming very obvious to me listening back to old mixes given the increased fidelity of the RME. But I never record with VST effects, so will 24 bit be better? Even if it makes seemingly no odds to performance, just disc space?

    3) Given the fact I have virtually zero-latency monitoring with the RME, and never use VST effects while tracking, should I bother setting the audio buffer so low for smaller latency? Am I correct in saying that if I set it for much higher say 7.256ms that everything will hit the drive exactly 7.256 x 2 = 14.512ms later than I hear or play it? In which case, can I set the Adjust for Record Latency setting to 15 (assuming a little extra for the airspace and cable runs) and this will correct this so close that noone will ever notice? I have had problems with rappers ending up out of time with their backing tracks before and having to fix it manually. (Not that I record very many rappers).

    4) And a couple other questions. If my G5 has a digital output, is that any use to me at all? Presumably not cos audio goes in and out via Firewire only....

    5) And finally, if the RME goes on and on about how good its clocking facilities are, why can't I clock Cubase to it? Would I want to?

    Guys I hope you can help, I just wanna get this sorted and move onto the actual business of recording. I cabled it all up on the weekend, and am gonna record my band on Saturday for practice and to test the settings out, before I embarass myself doing an album for somebody and getting glitches.

    I just want confirmation that recording at 24-bit and 44.1kHz is not a decision made cos of hardware limits but for other reasons.

    Thanks in advance,

    Jamie
     
  2. Jeemy

    Jeemy Well-Known Member

    sorry for the early *bump* but can anybody advise before the weekend so I can do some test 10-tracks?

    thanks,

    jamie
     
  3. David French

    David French Well-Known Member

    I really don't want to go into the long debate on sample rate nd bit depth again, but from my study of several such debates, I can tell you that i've come to the conclusion that 24/44.1 is not a compromise. If you want to read the end-all debate o sampling rate, see here.
     
  4. Jeemy

    Jeemy Well-Known Member

    Deep.

    Thats all I needed to hear Dave I will worry no more about it.

    Definitely not 32-bit fp though?? :lol:

    Kidding. Will rest happy,

    J
     

Share This Page