• Register
  • Help
Results 1 to 9 of 9

Topic: latency

Share/Bookmark
  1. #1

    latency

    I would go back to a very controversy subject: latency in Gigastudio 160.
    I\'ve just done some comprehensive measurements and I would like to share with everybody.
    I am running GST on a dedicated P3/933, 512MB, dedicated Maxtor HD, Frontier Design WaveCenter/PCI soundcard.

    I\'ve made all the optimization adjustments to the system.

    First of all, as a reference, KORG TRITON has a 4 ms latency, but with no jitter (variance of latency). It\'s always 4 ms.

    STEP-1:
    -------

    soundcard buffer = 128 samples,
    cache=8192, or 4096, or 0 (it doesn\'t matter)
    latency is between 15 ms and 19 ms, that means jitter = 4 ms

    Step-2:
    ------

    soundcard buffer = 64 samples
    cache = 8192, 4096 or 0
    then latency is between 9 and 10 ms (jitter is 1 ms).
    The last one is not bad, but even my CPU cannot handle 4 NFXs at a high polyphony without cracking. Even at peak polyphony=70 it cracks sometimes (when I play the gigaharp quite hard).

    ============================================
    The first one is more stable, I got no cracks at all at high polyphony (peak - 120), with 4 NFXs plus EQs, but again: LATENCY!! The timing is terrible!
    OK, I can adjust the timing within CUbase (setting a negative delay of the track), but the problem is the variance of the latency (jitter).

    So I’m wondering if everybody is getting the same kind of results or is it something wrong with my system?

    Any input is appreciated.

    mihmar


  2. #2

    Re: latency

    hi mihmar,
    I\'ve got hard problems like u. I\'ve read your message now, and if you say me how I will also make tests to get clear numbers.

    And what do you mean by setting a negative delay of the track by to adjust the timing within cubase? Perhaps could help a bit?
    So please tell me what I have to do (how to test and how to get delay timing?)


  3. #3

    Re: latency

    This is very inconsistant with normal behavior and you both must have a system issue and you should call tech support (with your registration number in hand).

    Tens of thousands of GIGA users are getting well below 10ms, with or without NFX, and this is not a partial measurement - but is all the way from external MIDI In through to Audio out. This is also true when you have
    high polyphony going and while streaming. Also, it is technically impossible for the NFX to add any delay whatsoever to the original signal, as all NFX are of RMW (read modify write) type and no buffering is added whatsoever. The only way then can add dely would be if you are removing all of the dry signal and purposely adding a deley in the effect algorithm.

    Please call NemeSys technical support, and be sure to have your CD Key number. If your sole agenda is to get your system working, then this user forum is not intended to be a tech support forum for registered users.

  4. #4

    Re: latency

    This is not normal. From my experience GS is very tight. Having a \'clean\' system is important however.

    Even when i first tried Gigasampler on my old PI-166 with a soundblaster gold latency was ok.

  5. #5

    Re: latency

    Just taking a far-out guess: maybe your soundcard is sharing an IRQ with something else (perhaps the video card)? In that case your soundcard *will* suffer a delay in actually getting to playing out its audio streams.

    If this is the case and you can\'t reassign the IRQ inside windows, move the card to a different PCI slot.


  6. #6

    Re: latency

    Gentlemen, may I step into this one?
    It seems to me that Mihmarīs topic was not the latency itself, but rather the jitter (the variance of latency). I can confirm this - but the tests that I have done show that not only GigaStudio has this jitter. I could see the same jitter on my other soundcards and softsamplers as well - not a bit smaller than Gigastudioīs jitter.(EWS88MT for Giga, EWS64XL sampler, AWE64 sampler) Ok, I admit, I could see something like 3 ms jitter, but even 3 ms is a lot and can ruin the song - especialy when youīre working with loops.
    So, my 2 pence: itīs not a GigaStudio problem. Itīs rather a PC problem, or better - a Windows problem.
    Just like Mihmar says, I took real care not to have garbage apps running along, and my rig is very properly configured.
    I donīt see a solution in sight. Whereīs BeOS???? Where the hell is it????
    Best wishes,
    Endicott

  7. #7

    Re: latency

    Sorry for having told an untrue thing. I said I tested ALL my softsamplers etc - but I didnīt test them all. Now I tested one more - and see here - itīs tight like an audiotrack. Itīs simply perfect. I canīt detect no trace, no shade of a jitter. Iīm talking about Emagicīs EXS sampler. So gentlemen, it seems we should mix the cards again :-)
    (I didnīt test Halion yet - but since itīs a VST instrument, it might be tight as well).
    Iīm very grateful to Mihmar for this thread. Without it, I would have never checked this EXS.
    Best wishes,
    Endicott

  8. #8

    Re: latency

    thank you guys for your response.

    Endicot,

    Did try the Logic sampler playing it in REAL TIME (from an external sequencer)? Because the pre-recorded tracks have the advantage of being compensated latency-wise by the host software (I mean like Cubase, Logic), so their samplers seem to be a little tighter...but again, was it real time?

    mihmar

    p.s. in the meanwhile I\'ve upgraded to a faster PC and I\'ve squeezed the Frontier cards\'s buffer to 64 samples. It\'s reasonably nice.

  9. #9

    Re: latency

    Hey Mihmar,
    I didnīt test for latency. I tested for jitter. My test was like this: Iīve built a wav file \"per hand\", in SForge. It contained 5 very short impulses, at equal distance ( that is, 4 beats in a 4/4 bar + the next first beat). With this file, Iīve built a patch in all different samplers: in GigaS, in EWS64, in AWE64gold and finaly in EXS. Now, I started Logic audio, programmed some 100 bars which contained only one note/bar. Everything quantized at the same note, same volume, same length (one bar)etc. I started with a somewhat slower tempo, so the fifth beat was to be heard right before the new note, you see. Then I raised slowly the tempo, until the 5th beat and the new note started to overlap. Then I recorded everything on an audio track, and when ready I measured the jitter...coz one bar was perfectly overlaped, next bar I had both impulses etc. A random overlaping, between 0 samples and some 150 samples (that is, almost 3.5 ms).
    Now, when I played the same thing using the EXS, the overlaping was equal. I could not detect ANY JITTER AT ALL. They even sounded exactly the same. I never knew what they mean, when they say \"one-sample precision\". Now I know. In the meantime, I did the same test on Halion - as I expected, it was just as precise. No jitter at all.
    Once again, the latency was not my point of interest. And yes, it was \"realtime\" :-)
    As far as I can guess, itīs the midi itself which brings the jitter. Everytime when midi comunication was involved, the jitter was there. I donīt know how Logic talks with its internal instruments, but it seems not to be by means of midi. Thatīs the only explanation that I can come up with.
    I admit, Iīm a little frustrated coz I donīt exactly like neither EXS, nor Halion - but when it will come to precision, like percussion tracks or loops, Iīll have to use them. Thereīs no alternative for me.
    Best wishes,
    Endicott

Go Back to forum

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •