Vienna Symphonic Library Forum
Forum Statistics

196,803 users have contributed to 43,032 threads and 258,448 posts.

In the past 24 hours, we have 2 new thread(s), 18 new post(s) and 91 new user(s).

  • Well, if it was a general OS issue, shouldn't it happen in all other sample-players as well? Haven't had the problem with EXS24 or Kontakt [H]

  • hi,

    unfortunately it is true that OS X uses to "page out" data which doesn´t seem to be relevant at this very moment for the OS. And because you cannot switch off or limit the so called "Swap File" in OS X it can happen that some of our samples get swapped into the virtual memory (= located on the internal HD) on first load.

    After you have played with the samples a little it gets obvious for the OS that the regarding data is needed for playing back the samples in the required timeframe (meaning in realtime) and that it is important data.
    Therefore the samples get a higher priority and will be kept in" Real Memory", which means they do not get swapped to "Virtual Memory" anymore.

    This is exactly what happens when you notice that the more you play with the samples the less clicks you get.

    This also happens to samples loaded in an Instance of an EXS player very frequently.

    I just did a Workshop in Berlin a couple of weeks ago and played back a Piano track from an EXS instance on a MacBook Pro (from the internal Drive) and on first playback I think every sample header was swapped to virtual memory.

    I didn´t only get clicks but after playing with crackles and clicks for some bars the machine stopped with an error message saying "Samplerate 44073 detected. Error while syncronizing Audio and Midi"

    Obviously the HD drive was too slow to stream all sample headers with the right samplerate.

    Apple knows about this behaviour and that this is a problem for RAM-intense realtime applications. As far as we know the problem is being investigated by Apple but we don´t know at the moment what can/will be done about it.

    I got the information from an Apple engineer that the more RAM you have installed the better, because when physical RAM is still free, the Swap File Technology starts to use this free ram for its page file. So the samples are just moved from the Real memory of the thread "VSL-Server" to another area in the physical RAM.

    Therefore having more RAM installed will reduce the clicks significantly since for the samples it doesn´t matter at which spot of the physical RAM they are stored.

    best
    chrisK

  • I can corrobate this statement.

    Maya

  • simon,
    if the memory/hd access of the exs and the vi is identical, then you are surely right. i just thought that a difference here might make the vi more vulnerable to changes in apple's virtual memory management system.

    chris,
    thanks for your reply!
    i think it is not surprising that you can get such issues also with the exs when you stream from the internal hd - but i never experienced this with the exs when streaming from a decent external disk. what people report here, is that they have these dropouts even when using a fast sata disk and when basically none of the ram is used.
    i for one have these dropouts when i load a single matrix - which uses e.g. 20 mb so that there are still more than 1.5 gb free memory available (after a fresh startup) - and play a simple (basically monophonic) melody line - streaming from a dedicated firewire 400 disk from which i can stream whole arrangements without any problems when using the exs.

    best

    kai

  • kai, besides i have to confirm chris' observations - the reason for not noticing the behaviour when streaming from an external disk ist not so much the speed of the disk itself, but the fact that OS X pages to the system volume which is on the same disk on a notebook, so simultaneous access to different regions of a disk can easily cause such a behaviour.
    christian

    edit: this is one reason why we recommend for all products to have the data on a fast, seperate hardrive

    and remember: only a CRAY can run an endless loop in just three seconds.
  • A merry thanks to all the knowledgeable people who answered this thread! This all makes a lot of sense now. Overall I am extremely pleased with my Vienna Instruments. I made the decision to backup, erase, and reinstall my entire computer to clear out the HD and memory of all the junk I no longer use. I must say, after a fresh reinstall VI is working better than before. Thanks again!

    -Brian

  • last edited
    last edited

    @Another User said:

    I just did a Workshop in Berlin a couple of weeks ago and played back a Piano track from an EXS instance on a MacBook Pro (from the internal Drive) and on first playback I think every sample header was swapped to virtual memory.

    I didn´t only get clicks but after playing with crackles and clicks for some bars the machine stopped with an error message saying "Samplerate 44073 detected. Error while syncronizing Audio and Midi"


    This is very unusual, and usually it goes away after the first play. It is pretty customary for me to have a substantial number of EXS24 instances in my sessions without any issues.

    However a single instance of VSL VI is enough to create clicks and pops. While I don't dispute that OS X has odd ways of handling data, this is obviously an issue that others have resolved. EXS24 and Kontakt simply don't exhibit the same behavior as VSL VI, and I sincerely hope that VSL doesn't hide behind the excuse of "It's not us...it's them" without making a good effort at imrpoving the code of the VSL VI engine.

  • last edited
    last edited

    @cm said:

    kai, besides i have to confirm chris' observations - the reason for not noticing the behaviour when streaming from an external disk ist not so much the speed of the disk itself, but the fact that OS X pages to the system volume which is on the same disk on a notebook, so simultaneous access to different regions of a disk can easily cause such a behaviour.
    christian

    edit: this is one reason why we recommend for all products to have the data on a fast, seperate hardrive


    I have all samples on a RAID 0 set of fast WD harddrives, I still get clicks and pops for the first few minutes. I don't get that with Ivory, Akoustik, Kontakt 2 or other disk streaming based players.

  • last edited
    last edited

    @Simon Ravn said:


    I have all samples on a RAID 0 set of fast WD harddrives, I still get clicks and pops for the first few minutes. I don't get that with Ivory, Akoustik, Kontakt 2 or other disk streaming based players.


    Does VI appear to perform better with increasing time of usage or not? If not, its maybe not a problem caused by the OSX RAM Management...Could you please give us some more details?

    Best
    Maya

  • Same problem here. When played for the first time, most patches create pop / click. When playing the same notes a second time, everything's fine.

    As other people here, this never happens with any other sampler / plug-in (EXS24, Kontakt, Ivory, Trilogy, etc.).

    Jerome

  • In my experience this does not stop happening after repeated playback of the sequence. Clicks and Pops appear at first play, and the appear after playing it 20 times. All of my VSL SE sounds are housed on an internal SATA drive which is dedicated to samples only and nothing else.

    I suspect that as more users buy VSL VI products, more people will chime in with these issues. Once again, I really hope that you guys are taking a closer look at the code trying to determine what might be happening and hopefully releasing an update asap.

  • Just to add something else to the equation....could the Vienna Key be a possible culprit? I don't know if the VSL VI periodically checks on the ViennaKey, or even runs some of the code through it. In that case, could an overburdened hub be creating clicks and pops as USB traffic is preventing VSL VI from functioning in real-time?

    Just wondering what the connection is since some of the other users report good performance.

  • In my experience the clicks DO go away with repeating playback of the samples. And I don't think the LCC dongle is the problem, since it is used for other plugins and programs as well. And at least Steinberg e.g. claims that the productions take no CPU cycles of importance. I think the problem is somewhere in the playback engine.

  • Well....much to my resisting the idea....yesterday I plunked down a few hundred $$ to double my computer's RAM (I have nothing against buying more RAM....I just hate dumping money on a system I know I'll be replacing in a few months).

    I don't know if this will solve the clicks and pops but I'll keep everyone posted as I'm sure everyone is so damn curious about what happens next.

    If low RAM is the culprit (I don't really consider 3 gigs low) then hopefully my issues should go away.

  • I am getting some pops and clicks too with my VI. But do you hear them whem you record the music on a cd?

  • Same for me, more clicks and pops than before, very strange, and this is not the case when I load an old VSL instrument.
    Please VSL team, could you show up a little bit more on this subject ? Anytime someone is speaking about this kind of problems you seems absent [[;)]]

  • last edited
    last edited
    Dear customers,

    about Mac OSX RAM Management:

    as explained before by chriskard, physical memory is paged to the so called Swap-file (in 4096k blocks), whenever physical memory is being over subscribed. The swap file is located somewhere on the system HD (its estimated size is shown as VM = virtual memory). Continuous swapping will affect the system performance and the performance of all applications - errors might occur "without cause".

    The situation will become worse when it happens that sample data is swapped! The access time of an internal HD is way too slow in order to play back audio samples smoothly (that is why the beginning of the samples is loaded into RAM - - the rest of the sample is streamed from HD). Result: samples won´t play back properly anymore = clicks and pops will occur.
    (There is a mechanism called memory protection for data that should not be swapped, but to my knowledge this protection can be overruled by the OS).

    Since excessive swapping makes an impact on the performance it is necessary to install more RAM to avoid it.

    The actual number of page out´s is displayed either in activity monitor OR in the last line of the header in Terminal when entering the UNIX command <top>. The UNIX command <vm_stat> provides more detailed information about the swapping activity of OS X.

    Moreover OS X RAM Management tries to allocate free memory ideally, it will analyze which data is used frequently and give important data a higher priority (as chriskard already mentioned). This behaviour explains why VI and other applications appear to perform better with increasing time of usage and it also explains why OS X will require almost all RAM, no matter how much is installed.
    Apparently Mac OS X RAM management is not optimized for RAM intense realtime operations which need to handle a large number of very short files.

    @chriskard said:

    I got the information from an Apple engineer that the more RAM you have installed the better, because when physical RAM is still free, the Swap File Technology starts to use this free ram for its page file. So the samples are just moved from the Real memory of the thread "VSL-Server" to another area in the physical RAM.

    Yes!

    Some customers said that problems with clicks and pops would not occur with EXS. But most EXS programs aren´t as large as Vienna Instruments matrices or presets can be! Therefore significantly less RAM is needed and its rather unlikely to run into swapping problems. We did some tests with EXS where we forced the system to swap data from EXS and it showed the same behaviour like VI does when physical memory is overwritten (clicks and pops).

    The only solution_at the moment_is to add more RAM (if possible). Hopefully the situation will change with upcoming operating systems and improved memory protection- and we are also investigating in order to find a solution (which might however reduce the possibilities in VI to load many samples).

    It is also recommendable to use VI as efficiently as possible and to create your own customized matrices and presets out of the patches you really need for your arrangement and to remove all unused (!) playing techniques. The factory presets are extremely "RAM intense" and therefore they are not always the best choice - but they are great examples for showing the keyswitch-possibilities and functionalities of VI programs.

    Besides all of this we mentioned that VI does not perform as good in Digital Performer 5.12 as it does in Logic Pro 7.2.3 or Cubase 4.

    Best regards,

    Maya VINSON

  • Thanks Maya for all you valuable informations "-)

    Two questions:

    1 Why didn't you provide the perf legato patch p & f separated ? sometime we want to play piano, sometime louder, why can't we load only the desired p or f
    (as discuted in this post Herb said the computers can handle that, the fact that you can load it as separate on the VSL was only a matter of slow computer)
    http://community.vsl.co.at/viewtopic.php?t=11000

    2 What are the pops craks plips heard on some sound files even if there is only one instrument loaded, plenty of ram and a super fast computer ?
    http://community.vsl.co.at/viewtopic.php?t=11252
    it sounds like a looped click every 1.7 secondes.

  • Maya,

    what do you mean, VI doesn't perform as well in DP 5 than in Logic? What kind of performance boost is there in loading VI in Logic instead of DP? Is there a big difference?

    Thanks,
    Jerome

  • last edited
    last edited

    @chriskard said:

    I got the information from an Apple engineer that the more RAM you have installed the better, because when physical RAM is still free, the Swap File Technology starts to use this free ram for its page file. So the samples are just moved from the Real memory of the thread "VSL-Server" to another area in the physical RAM.

    Yes!

    Some customers said that problems with clicks and pops would not occur with EXS. But most EXS programs aren´t as large as Vienna Instruments matrices or presets can be! Therefore significantly less RAM is needed and its rather unlikely to run into swapping problems. We did some tests with EXS where we forced the system to swap data from EXS and it showed the same behaviour like VI does when physical memory is overwritten (clicks and pops).

    The only solution_at the moment_is to add more RAM (if possible). Hopefully the situation will change with upcoming operating systems and improved memory protection- and we are also investigating in order to find a solution (which might however reduce the possibilities in VI to load many samples).

    It is also recommendable to use VI as efficiently as possible and to create your own customized matrices and presets out of the patches you really need for your arrangement and to remove all unused (!) playing techniques. The factory presets are extremely "RAM intense" and therefore they are not always the best choice - but they are great examples for showing the keyswitch-possibilities and functionalities of VI programs.

    Besides all of this we mentioned that VI does not perform as good in Digital Performer 5.12 as it does in Logic Pro 7.2.3 or Cubase 4.

    Best regards,

    Maya VINSON

    maya,
    thanks for your explanations.
    what you point out may be the reason for the problems that some people experience (who work with large arrangements and really stress their system). however, it cannot be the reasons for the problems i (and as it seems also others) have.
    to sum it up once more:
    i experience these dropouts loading a single matrix (20mb!!!) and having according to activity monitor 1.5 gb free! ram. i have checked this in the meantime and activity monitor shows 0 poge-outs while these dropouts occur!
    it definitively has nothing to do with the size of the instruments either since i can run _much_ larger exs instruments than the simplest vi matrices which already cause problems.
    since the vienna instruments apparently work fine for most users it must be a more specific problem (maybe some software-incompatibility, ...). i would be happy to provide you with information to figure out what cuases this. but this requires that you take our input seriously and don´t go on presenting possible causes that evidently contradict our experience.
    best regards
    kai