Vienna Symphonic Library Forum
Forum Statistics

194,481 users have contributed to 42,922 threads and 257,973 posts.

In the past 24 hours, we have 2 new thread(s), 9 new post(s) and 80 new user(s).

  1. Home
  2. Sample Libraries
  3. Vienna Instruments Libraries
  4. Mac: RAM's Relation to CPU Overload

  • Mac: RAM's Relation to CPU Overload

    P
    P
    This from the compendious file of things I should know but don't....

    I'm running a dual 1.8, OS 10.4.4, Logic 7.1.1, with a song that uses both EXS and VI instances.

    With 4.5 GB of memory, I can consistently cause a Core Audio system overload at a certain tutti. If I mute one, maybe two tracks, I can get through the section and play on safely (until the next tutti). Needless to say, it offline bounces fine -- it's just the live play that stops.

    Question: will an upgrade of 2 GB (for a total of 6.5 GB) help? It looks like a CPU issue, but could more RAM relieve that?

    Any other possible tweaks? Thanks.
  • J
    J
    Check in the Activity Monitor for your Ram usage. If the VSL Server app is already using about 2-2.3 GB of ram, then adding more memory will not change anything.

    Jerome
  • P
    P
    Thanks, Jerome. My Activity Monitor reads that a suprising 1 GB is still free overall. VSL-Server is only using 1.27 GB real and 1.56 GB Virtual. But the free memory drops consistently as I work on the song (without loading any new samples). I know I've seen the same song as low as 400 MB free only after a couple hours' work.

    So I'm still uncertain. When I re-assign two Space Designer busses to something far less hoggy like GoldVerb, I still get overloads. The two audio moniitors in Logic don't change, and the only thing that allows play-through is still the muting one or two tracks.
  • J
    J
    Plowman,

    First of all, welcome to my world of coreaudio overloads. In fact I'm just stepping back for a moment from this very same problem. Low and behold somebody else is suffering the same thing. Every time I get to a certain density of a composition up pops this gremlin.

    The Activity Monitor is a mystery to me in most regards, but I'm not convinced that the entire problem is RAM. It is certainly RAM-related though. And it is a problem that preceded the VI. My observation is that it seems to be also a sonic density issue that occurs in tandem with RAM issues.

    For example, I've been working on a mock-up of a classical piece that requires a full orchestra and necessitates RAM optimization on each track as the composition is gone through. The Memory Free window in the Performance page of the VI says I have over a 1.5 GB available. I do have some instruments in EXS, but not many. Still almost (not always though) every time I get to the big, dramatic moment the coreaudio error occurs.

    Mind you, I do have 6.5 GB of RAM in my dual 2Gig G5. I think the extra two gigs of RAM helps a little but not a lot.

    At this point I'm not convinced that we can solve this issue and that I may not be able to avoid going to multiple machines some time soon. I don't know if it will be additional Mac Minis or whatever Apple comes out with in the next month or so - or a couple PC's. I know it will be a real change of work habits and will take some more technological information for me to get it all together.

    But having to bounce to audio before the composition is far enough along means that I have to more often than not have to go back to VI/MIDI again for final grooming of several tracks all seems so time consuming. For the most part I'm organized enough to keep all this in my head but it is difficult to sustain the memory of the original emotional content.

    Basically, I need a way to work faster. When dealing with multipart harmony in a section of like-instruments it usually ends up sounding so organ-like (i.e. VSL's demo of Pictures at an Exhibition) unless each instrument has its own articulations and separate MIDI controllers. And of course you end up creating a lot of VIs. Then the RAM and coreaudio problems occurs and then the endless circle of optimizing, bouncing and unbouncing, fixing, re-optimizing and re-bouncing until either schedule or end product is compromised.

    Yeah, multiple cpu's are beginning to sound enticing.

    Best regards,
    Jack
  • J
    J
    last edited
    last edited

    @Plowman said:

    This from the compendious file of things I should know but don't....

    I'm running a dual 1.8, OS 10.4.4, Logic 7.1.1, with a song that uses both EXS and VI instances.

    With 4.5 GB of memory, I can consistently cause a Core Audio system overload at a certain tutti. If I mute one, maybe two tracks, I can get through the section and play on safely (until the next tutti). Needless to say, it offline bounces fine -- it's just the live play that stops.

    Question: will an upgrade of 2 GB (for a total of 6.5 GB) help? It looks like a CPU issue, but could more RAM relieve that?

    Any other possible tweaks? Thanks.



    Sounds lke you may be reaching the maximum number of voices your EXS and VI instruments can deliver. This is more likely to be CPU related than RAM. One way round would be to select the tracks that use the most voices (thickest chords) during the tutti's and freeze these. you may find that by just freezing a couple of tracks the session will play out fine.

    One thing to consider if you are running a session close to the limits is how you mix it. If you do a real time bounce (or just record the mix outside the box) you may have the EXS dropping samples earlly to make room (CPU wise) for the most recently triggered samples and you will not be achieving full polyphony. If you do an offline bounce all voices will be rendered for their full length.

    Hope you track down the bottlenecks.

    Julian
  • P
    P
    These detailed, well-thought reponses are of comfort and aid. I'm quite grateful.

    "You may be reaching the maximum number of voices your EXS and VI instruments can deliver." Right. I have two patches in particular from SAM Brass (horns and trombones) that are four-layered. At the tipping point, muting just one of those usually gets me past the crescendo.

    Curiously, I need to mute three VI winds (flute, oboe, and clarinet) to get past the same sticking point. In other words, three VI tracks seem to be demanding less power than a four-layered EXS instrument. And the topper is, they're legato patches.

    By the way, I moved my whole Woodwind I from an external Firewire LaCie to an internal Hitachi. It made no difference.

    Jack, your articulate review makes me think we live in parallel universes, from the loathsome organ-like tuttis to the hand-wrenching over a new Mac Mini.

    My problem also preceded VI. In fact, VI had no effect, which is great praise considering the tremendous power and realism we gain by it.

    "I think the extra two gigs of RAM helps a little but not a lot." That's exactly the conclusion I'm forming. There may be some tangential benefit to 2 more GB, but I feel like our "core" issue remains.

    "For the most part I'm organized enough to keep all this in my head but it is difficult to sustain the memory of the original emotional content."

    That is precise truth. After all the stat sheets and system requirements, that is one number you will never see.

    In the past, I've spent months developing Giga templates to access from Logic, for the express purpose of sidestepping the core overloads. But now, after VI, many of my Giga sounds are unacceptable. Even some of the VSL stuff from First Edition! Add to that the inability to bounce offline once Giga is in the mix, and I happily wash my hands of it -- only to re-inherit the core overloads.

    Perhaps a single-computer solution (or seamless network) for a *reasonable* orchestra is just over the river. Maybe Mac Mini's are the bridge.

    And if people are reporting outrageous performance from the mini's, imagine the power of the new MacPros (desktop Macs). Macrumors reports today, "...the Mac Pro will be announced at WWDC and shipping in August."
  • J
    J
    Julian,

    Yes, I believe you are right that it's mostly a voice limit issue. Thanks for reminding me of that. I was mentally wrapped up in the RAM thing and was overlooking voices. The fact that I use a breath controller and velocity crossfader most of the time probably adds to the voicing density problem. However, it works well for most instruments musically and has become an important part of my work flow.
    It seems that you, Plowman and I have quite similar rigs. But anyway you look at it and no matter what the careful observation of the problem may show it's still the same problem - insufficient resources on a single cpu to playback a large number of instrument articulations. It especially gets wobbly if you add in other software like Kontakt, etc., etc., etc.

    Plowman,

    I'm starting to feel pretty strongly about going to multiple cpu's. I don't know what form this will take yet. I guess this is the start of my investigation. I probably will wait until January before I implement any new system. This gives me the chance to hear from the Apple Developers Conference next week, AES in October, MacWeek and NAMM in the first part of Jaunary. By that time there will be further annoucements of new Macs (hoping even for a new Mac Mini) and developments with VSL and other library developers. This will give me the time and the mental resources to figure out the best way to go. I'm not counting out going the PC route either. If PCs present the best workflow I will go that way.

    Is it SvK who has the Mac Minis? Does he use Plogue Bidule? I don't even know if each of the slave computers need an audio interface. It will be a real learning experience for me.

    I feel for you on the Giga issues. I didn't get into this whole thing until Giga had mostly run its course. Glad I never had to go there.
  • J
    J
    If you use Intel Mac Minis, you don't need an audio interface if the s/pidf stereo output is enough for you.

    If you want to get more discreet channels, then getting an audio interface will be necessary.

    Jerome
  • J
    J
    Thanks Jerome,

    That's one more important piece of info I will need. As much as I'd like to stay Mac there are some other considerations that will probably make me to go the PC route.

    I probably should not hijack this thread with this topic. I'll start a new one as soon as I can conceive all the things I need to ask.

    Best regards,

    Jack
  • P
    P
    Jack, I salute your intrepid approach to PC networking. I've read svonkampen's thread with great interest (but limited undertanding) for several weeks.

    I only have one PC left, a Carillon that's four years old. I've played VI's directly without latency, but Plogue Bidule introduced delays of nearly a half second (with only one VI loaded within it). Fixable, maybe, but then I'm left with VI's on a PC computer with no effects processing of its own.

    And that's the stumbling block, really. If we're willing to slam a summed stereo output from a slave into our Mac and overlay a uniform verb on it, our world suddenly gets a lot easier. It's the differentiation of audio channels that starts the dominos falling. It seems to require either networking skills that I've yet to develop or cable sprawl from audio breakout boxes that I reflexively resist.

    If I'm reading svonkampen's posts correctly, multiple ins into Logic are possible via networking. But then the clock issue came up in one thread, and frankly, I got scared away. The sad truth is, you only really internalize this stuff if you buy it and work with it.

    I'll follow your progress. Six months could change the computer landscape quite a bit.
  • J
    J
    This is only half OT--

    There is a banner up at WDCC that says "Mac OSX Leopard: Introducing Vista 2.0". Interesting mix of metaphors here.

    To my understanding (however limited) this will be the harbinger of more tightly integrated Mac/PC networking or a whole new Pandora's Box of nightmares.

    And while many of us are in line to spend some serious cash to get what we need over the next year or so to make our systems work, I have a feeling that the news coming from Monday's news on Leopard (and Vista) will have some significant impact (I hope) on how all of this networking business for heavy VI users might work --- one of these days.

    I know how close-mouthed Apple is, but I can't help but to wonder just how much teaser code Apple has doled out to developers at this time with a vow of silence....? I turn, it makes me wonder what VSL is working on for 2007, and if it will still take a cityscape of computers to run the VI Cube...

    Moreover, because of the mention of the word 'Vista' in this context, such things as MIR take on an entirely different timbre with Mac users.

    Might we be at the beginning of the this-does-not-suck era? [[:|]]
  • J
    J
    The "solution" right now is to dedicate computers to orchestra sections or (even better but more expensive) half sections.

    That's what we're doing here. On an Intel Mac Mini with 2GB of ram, we're able to load 4-5 instances with 9-16 patches loaded, including two legato patches per instance. I'm still in testing phase, but for now it works well.

    We connect the Intel Mac Mini stereo optical output to an interface which converts four, 2-channel, S/PDIF to one, 8-channel, ADAT. Then we connect the ADAT to a MOTU 2408.

    This gives us two stereo pair per orchestra section back to Logic - not bad at all!

    Jerome
  • J
    J
    JWL,

    The "Vista 2.0" thing is just Apple being cocky, along the line of the "I'm a Mac, and I'm a PC" adds. The idea, of course, is that Vista isn't even released yet (i.e., version 0.9b, or something), but Apple's already got Vista to version 2. You know, geek humour.

    I'll be running one PC P4 2.8, 1 MacBook core duo 1.83, and 1 G5 dual 1.8 when I get back to the UK. We'll see how it goes, but I'm pretty optimistic. So far, the reports on Intel Macs with VI have been pretty great... I know I'll have to run XP on the MacBook for the time being because of the "external sequencer makes for popping" issue, but I'm sure VSL is on their way to a fix for slaved Macs, if it's not fixed already. That reminds me that I'm going to need yet another Vienna Key... yuk.

    J.
  • J
    J
    OK, spent a relatively sleepless night on this network subject. I won't be surprised if there is a few more just like last night.

    Not anywhere near any conclusions yet. However, I did notice on Apple's website today that there benchmarks for the new Mac Pro's were based on Logic 7.2.2. Hmm, I don't think I've seen any mention of this version anywhere. It would be wonderful to think that soon-to-be-released versions of Logic coupled with the Mac Pro might be able to finally address the full 16GB of RAM. Then again it might be too much to hope for so soon. Obviously there are some chapters that are going to be added soon to this whole story of VI networking.

    I wonder when VSL will do their own benchmarking on the Vienna Instrument with a Mac Pro. Comments moderators?

    I'm currently under the impression that some applications I plan to get require PC's only. Otherwise I would stay in the Mac world. The current state of VI networks seems so midieval. Plowman has me somewhat concerned over Plogue Bidule and what I know so far (which is still definitely very little) of the monitoring of slave audio seems Byzantine.

    Still, if people can have networks of a master G5 and nine PC slaves, as reported in this month's VI Magazine, it must work.

    Glad I don't have to do this all by tomorrow.
  • S
    S
    Jack,

    7.2.2 IS the current version of Logic....just download the update.

    SvK
  • J
    J
    last edited
    last edited

    @svonkampen said:

    Jack,

    7.2.2 IS the current version of Logic....just download the update.

    SvK


    You sure?

    Julian
  • P
    P
    I didn't mean to discourage anyone from Plogue Bidule. Keep in mind, my PC is only a Pentium 4 1.7 GHz with 1.5 GB of memory. I can't explain why a VI standalone would play so well and Plogue would cause latency. But on equipment this old, no fair conclusion can be drawn.

    I will say this in light of today's announcement: if there's anything in a user (like me) who resists complication (and does so increasingly as he ages), then all that could be done within VI and Mac / Logic / EXS can and should be pursued. The Mac endgame looks quite compelling.

    I'm personally energized to see through the short term with hand-me-down Macs and/or Mac Minis. Sure, a 16 GB MacPro on Leopard may not be viable until well into '07. But it's worth the wait, because so much music can already be made between now and then. And "then" is an extremely powerful single computer workstation.

    This is the closest we've ever been to a single computer orchestra with convolution reverb. I suspect MIR will require another computer, but I can live with Space Designer for a while. Times are good in this regard. Expensive, but good.

    P.S. I haven't seen the whole presentation, because Apple's site is getting slammed. But in the background, I saw a graphic that said that Logic runs 1.6 times faster on a MacPro (compared to a G5 Quad, I'd assume).
  • J
    J
    SvK,

    I am still aware of only 7.2. Where would one go to find 7.2.2?


    Plowman,

    Yeah, the new Mac Pro is compelling. I'm releived that I don't really have to move on a network at this very moment.

    Now it seems that perhaps my current G5 along with a fully loaded Mac Pro will handle VSL VI quite well sometime after the first of the year. This is of course hoping that the Leopard OS and Logic come forward with a 64 bit system that can access all the memory and VSL does their homework to create a 64 bit VI. Then I can get a PC or two to handle any virtual instrument/application that needs that platform.

    I'm sure that VSL realizes what this new computer can do for their system.
  • J
    J
    last edited
    last edited

    @Jack Weaver said:

    SvK,

    I am still aware of only 7.2. Where would one go to find 7.2.2?




    I've seen 7.2.1, and I've read forums where members were *asking* if 7.2.2 would be announced soon, but 7.2.1 is the latest version I've seen so far. I could be half blind, though.
  • P
    P
    "Logic Pro 7.2.2 Results Up to 1.8x the plug-ins of the Power Mac G5 Quad."

    That's from the Apple site, but I see no public release of 7.2.2 anywhere. This version A. is likely still in-house for diagnostics, and B. will offer compatibility with new hardware but few new features or fixes for the average end-user.

    I think the next Logic will hinge on 64-bit optimizing, and that will take some time. Natives will start beating their sampled drums around NAMM '07 if by then Logic hasn't gone on record with a version that maximizes the WWDC's premiered hardware.

    Remember, full 64 bit OS is news to us, not the Logic code writers.