moebius7

Q: Sample drop out driving me insane

OK, this problem is now officially driving me insane. Before I continue, I must explain that this issue has "semi-plagued" me for several years, and spans different computers and hard drives, so I'm now led to conclude the problem is a software one, though I'm not expert enough to be sure. Otherwise I would have solved this and not be asking for help. I have never got to the bottom of this issue.

 

So, I work in Logic , and overwhelmingly use Kontakt, but I also use EXS24, and the same thing happens there, so it's not Kontakt. I have suffered sample dropouts, while constructing cues and tracks. And it doesn't seem to matter how much is loaded. It can happen with one instrument or many. I'll be playing away ( I mean physically playing the keyboard ) and suddenly the sound drops out. This usually happens on notes which are held for long periods, but in actual playback, it can happen to anything. A crash cymbal, starts to play, but immediately gets choked. Then again, there are periods when the whole thing behaves itself, but trouble is never far away.

 

The worst and most infuriating examples of this are when I'm hold down long notes for the purposes of drones. Randomly the sound will "choke" and I actually hear the hard drive mechanism protest ! Then the sound stops. This causes a huge waste of time for me, as I have to start all over again. It's important ,while playing this stuff to have it behave, as you have to keep the vibe going. Imagine repeatedly having to hold down notes, and pray that the thing won't drop out randomly before the end. It's ridiculous.

 

Weirdly, there are also occasions when a sample ( or samples) will cut out at exactly the same point every time. Then , mysteriously, that clears up. It's baffling, intensely annoying and I'm really at the end of my rope with it.

 

I must stress that when I bounce a mix, offline, this never, ever happens. The mixes are flawless. What the **** is that ??

 

So, in terms of technology this is what I'm using right now. A late 2012 iMac i7, 32 GB RAM, and still on Mountain Lion 10.8.5 ( don't ask ). I'm currently streaming the audio from a LaCie TBT2 12 terabyte RAID drive, configured as JBOD. It's Thunderbolt 2, and the iMac is TB1. This is not the problem , of course ( is it ?! ) . It's all backward-compatible.

 

Previously to that I was working with the LaCie 6 terabyte version, JBOD. Same problem. Further back, different iMac and drives… doesn't matter. Same problem. I'm on LPX 10.0.7. As I'm emphasising, I feel that none of this technical data means anything, because I've suffered this issue across many different systems, in the past 7 years.

 

I used to have Mac "farm" and ran that over ethernet successfully for a long time. I never had sample drop out issues at all, that I can remember ! I should point out that I do not use a sound card these days. I simply bounce to disk, "internally". All my clients are extremely happy with my results. The end mix is right up there, in quality. I'm happy in that respect .

 

Right now, I just completed a Discovery Channel series and this issue plagued me all the way through like never before. I cannot understand why , but this problem seems worse than ever on this new 12 TB drive. What should have been a pleasant creative assignment, turned into a total train wreck , all because of this utterly baffling problem. I've worked  on a lot of TV and games ,etc., and this has blighted me all the way through. I know, I know, I should have solved it sooner. What can I tell you? The jobs get done. I have no technical staff and work alone. Honestly , I'm beginning to wonder if LaCie is the problem. In the past i've had more than one drive go down, and they had a bad reputation. but these days, they are much more solid.

 

At some point , ASAP, I want to buy the latest iMac, install 64 GB RAM, and go over to a SSD equivalent drive system. Expensive, but if that solves it, I'm happy.

 

Can anyone help , please?? I've looked on the internet, but not yet found a solution .  Sorry for the overlong post. Many thanks to anyone who can solve this.

Posted on Oct 15, 2016 3:21 PM

Close

Q: Sample drop out driving me insane

  • All replies
  • Helpful answers

Page 1 Next
  • by Pancenter,

    Pancenter Pancenter Oct 15, 2016 5:17 PM in response to moebius7
    Level 6 (10,029 points)
    Audio
    Oct 15, 2016 5:17 PM in response to moebius7

    I'm heading out to a gig... but let me ask this, have you always used a RAID Array?

  • by moebius7,

    moebius7 moebius7 Oct 15, 2016 5:31 PM in response to Pancenter
    Level 1 (109 points)
    Audio
    Oct 15, 2016 5:31 PM in response to Pancenter

    Hi PC, thanks for the post. Answer is no. It would be true to say that the issue became much more noticeable after switching to such drives. My impression was JBOD would basically render multiple drives as  "one unit". I look forward to your further input…..

     

    Have a good gig !

  • by Pancenter,

    Pancenter Pancenter Oct 16, 2016 12:30 PM in response to moebius7
    Level 6 (10,029 points)
    Audio
    Oct 16, 2016 12:30 PM in response to moebius7

    The fact that the "Mac farm" worked perfectly leads me to believe the problems is basically related to JBOD... Just a Bunch of Disks" I've never thought that system was all that efficient as it depends on software, yet another layer of cr@p between the drives and Logic. Also, the software that manages JBOD is probably not designed for high speed real-time performance.

  • by moebius7,

    moebius7 moebius7 Oct 16, 2016 1:13 PM in response to Pancenter
    Level 1 (109 points)
    Audio
    Oct 16, 2016 1:13 PM in response to Pancenter

    I have to take what you suggest seriously, and therefore look into that as a possible / probable culprit. If that's the case I'm now using hardware that has no use or satisfactory alternative, since changing it to RAID will drastically reduce the disk capacity. I suppose I could use both the 12TB and the 6TB , both as RAID, and spread the samples over both drives. Ultimately, I think SSD must be considered as the only serious option , then.

     

    Before these these LaCie's I was using a "straight" 4TB Iomega. That was about five years ago,  and my memory isn't reliable, but I cannot recall it causing such problems. Have you actually heard of JBOD being less than efficient ?

     

    Many thanks for your opinion.

  • by moebius7,

    moebius7 moebius7 Oct 16, 2016 1:17 PM in response to Pancenter
    Level 1 (109 points)
    Audio
    Oct 16, 2016 1:17 PM in response to Pancenter

    Also, I'm still wondering why doing a bounce offline is always flawless. What is happening in a bounce that is different from just hitting play ?  In fact , I would have thought doing an offline bounce, where the drive is surely having to work harder, would be much more challenging for the drive to "keep up".

  • by Pancenter,

    Pancenter Pancenter Oct 16, 2016 6:36 PM in response to moebius7
    Level 6 (10,029 points)
    Audio
    Oct 16, 2016 6:36 PM in response to moebius7

    An offline bounce let's the computer use a single core for the bounce and it's not in real time.

    I'm convinced it's related to the JBOD drive setup, you would be better off with them Thunderbolt daisy chained... fastest drive first inline from the computer.

     

    In Logic's Preferences/Audio When you click-hold on processing threads... what's the highest number available?

  • by moebius7,

    moebius7 moebius7 Oct 16, 2016 7:04 PM in response to Pancenter
    Level 1 (109 points)
    Audio
    Oct 16, 2016 7:04 PM in response to Pancenter

    Excuse my ignorance, but that would appear to present more of a challenge to efficiency than having all the the cores available and in real time. Not really following that.

     

    As for the daisy chaining, are you suggesting to still have them configured as JBOD ?

     

    The highest available number is 8, which is what I normally have it set to. However, the 8th core is always peaking out and averaging far higher readings, even on modest cues. Sometimes I'll get cpu overloading, unless I select an audio track, in which case this seems to cause the 8th core to settle down. I've heard that the latest Logic ( which I cannot run on Mountain Lion ) is much more stable with cpu issues…..

     

    Thanks again.

  • by Pancenter,

    Pancenter Pancenter Oct 16, 2016 7:38 PM in response to moebius7
    Level 6 (10,029 points)
    Audio
    Oct 16, 2016 7:38 PM in response to moebius7

    You have to understand Logic's architecture, maybe this will help.

     

    From: Logic Pro/Express: Tips for balancing multi-core performance - Apple Support

    It's an old article but most, if not all of it is still valid.

     

    Logic breaks tasks down into threads that can be assigned to a single core for execution. It's not possible to split a single thread across multiple cores, although a single core may run multiple threads at the same time. This is one reason that you may sometimes see uneven load distributions across the available cores on your computer. If one thread is very processor-intensive, the core running that thread may show much more activity than the cores running less demanding tasks. This is the main reason that you may get "Core Audio Overload" warning messages even though some of the cores on your system may not show significant activity.

    -------------------------------------------------------------------------------- --

    Basically, a single channel strip can only use a single core. however, several channel strips can use the same core.

    So, your bounce goes through the master bus (a single core) and an offline bounce is not in real-time. Using a single core is actually an advantage due to Logic's architecture.

     

    As an experiment, set the maximum cores to 4, close Logic, reboot Logic and give it a try.

  • by moebius7,

    moebius7 moebius7 Oct 16, 2016 7:43 PM in response to Pancenter
    Level 1 (109 points)
    Audio
    Oct 16, 2016 7:43 PM in response to Pancenter

    Yes, sorry, when I said core, I should have said thread. Thanks for the info. Yes, it's bouncing faster than real time, which made me wonder why it was more reliable. In fact 100% reliable. In the past i tried all the options for cores numbers, but I'll try that, thanks. I did try automatic, too, btw. Are you perhaps suggesting this might improve the drive problem, which is what I'm solely interested in, frankly.

     

    If I may ask again. You suggested daisy chaining, but would that still be as JBOD's?

     

    Many thanks for your input so far.

  • by Pancenter,Helpful

    Pancenter Pancenter Oct 17, 2016 7:16 AM in response to moebius7
    Level 6 (10,029 points)
    Audio
    Oct 17, 2016 7:16 AM in response to moebius7

    moebius7 wrote:

     

    Yes, sorry, when I said core, I should have said thread. Thanks for the info. Yes, it's bouncing faster than real time, which made me wonder why it was more reliable. In fact 100% reliable. In the past i tried all the options for cores numbers, but I'll try that, thanks. I did try automatic, too, btw. Are you perhaps suggesting this might improve the drive problem, which is what I'm solely interested in, frankly.

     

    If I may ask again. You suggested daisy chaining, but would that still be as JBOD's?

     

    Many thanks for your input so far.

     

    Yes, it will bounce faster... the computer is not doing any real-time calculations, just bouncing the mix through a single channel strip. Using 4-cores will only use the processors hardware cores, the virtual cores will not be used.

     

    No, personally I wouldn't use JBOD, it's another layer of software "managing" the transaction... I've never found that to be a good idea. In your first post you wrote  "I actually hear the hard drive mechanism protest !" that tells me either the JBOD software is possibly not able to handle the streaming requirement... or... the drives have entered a low power state and are waking up.. so many new drives have built in power management, can't help you with that one though.

     

    p.s.  It's worth a shot to try the 4 cores.

  • by oingo boingo,Helpful

    oingo boingo oingo boingo Oct 17, 2016 7:16 AM in response to moebius7
    Level 1 (127 points)
    Audio
    Oct 17, 2016 7:16 AM in response to moebius7

    I'll attempt a simple summary:

     

    You have a lot of disks and your computer and the access busses might be overloaded in accessing all the content, so they cannot load fast enough. In a RAID array, any write access will have to copied to more than one disk, slowing things down further.

    To investigate: Measure the disk data transfer rate, for example in Logic by using the custom view or use Apple OS Utility Monitor.

     

    hd meter.png

     

    Possible solution: get one really fast disk (I have an SSD, it's awesome) and do everything on there. See whether that helps. And definitely avoid JBOD. Turn RAID off to see whether that improves things. Etc. 

     

    As for Bounce in "offline" mode: basically, your computer takes as much time as it needs for all processing to be completed, rather than attempting to bounce at the same speed the project would play in playback mode. That way slow disk access is not a problem. It is a very good sign that everything works fine when you bounce, because this means that if you unblock the data bottleneck, things will work fine.

  • by Pancenter,

    Pancenter Pancenter Oct 17, 2016 12:06 AM in response to oingo boingo
    Level 6 (10,029 points)
    Audio
    Oct 17, 2016 12:06 AM in response to oingo boingo

    It's not a real RAID array... there is no redundancy, no copies made. It's using different drives under software control seen as one big disk... to me, that spells problems. JBOD stands for 'just a bunch of disks".

    But yes, I think the bottleneck lies there.

  • by Pancenter,

    Pancenter Pancenter Oct 17, 2016 7:01 AM in response to moebius7
    Level 6 (10,029 points)
    Audio
    Oct 17, 2016 7:01 AM in response to moebius7

    Also, you may want to finish projects up before breaking the array.. projects are using defined paths.

  • by moebius7,

    moebius7 moebius7 Oct 17, 2016 7:09 AM in response to oingo boingo
    Level 1 (109 points)
    Audio
    Oct 17, 2016 7:09 AM in response to oingo boingo

    Thanks, but as Pancenter says I'm not using RAID at all. I purposely used JBOD, because RAID s no good to me at all. Why would I wish to steam sample and have copies of them all on the same device ? The older 6TB is my backup. Therefore, RAID has to turns off in order to us JBOD .

     

    Thanks re the bouncing info.

Page 1 Next