Vienna Symphonic Library Forum
Forum Statistics

199,046 users have contributed to 43,151 threads and 258,882 posts.

In the past 24 hours, we have 3 new thread(s), 10 new post(s) and 58 new user(s).

  • christian, as most of us were not at NAMM, can you elaborate on the timescale for VE, as that might make a difference to people's plans.

    DG


  • Oh and one more thing. One of the flaws in FXT is that the network periodically checks something (I've no idea what), which can cause the odd pop or click. Does the same thing happen with VE?

    DG


  • i have to admit i didn't run FXT for a looong time, but be aware it is not neccessarily FXT which checks across the network .. there are so many services which decide by themselves to *do something* ... uPnP, detection of shares, announcements, software updates (eg. java), ... endless list VE itself does nothing, except listening to incoming data and delivering the result.

    i'm not up to date with a release schedule so i'd like to say nothing, but as you know meanwhile we have fully working copies.

    christian


    and remember: only a CRAY can run an endless loop in just three seconds.
  • Thanks for the reply. I would however say that many people will have network problems, and probably much more severe than the one I've mentioned, so a thorough, exhaustive networking guide will be essential. My network was set up by a specialist, and still I have these minor glitches, so it is not necessarily easy to solve. A networking "tool" that configures the network perfectly fro VE would be fantastic, but not easy to do I guess. I think that it could be even more tricky on Mac, where networking seems not to be so easy.

    DG


  • last edited
    last edited

    @DG said:

    A networking "tool" that configures the network perfectly fro VE would be fantastic

    would be nice to have, but impossible. too many options for too many possible environments.

     

    basically it doesn't matter if its a mac or pc or any other operating system fortunately - ethernet is ethernet is ethernet.

    of course you will get some advise or help or recommendation, but in detail it is you or your technician who is responsible for a working environment (as for a working operating system as such)

     

    what will be not supported is a 100 mBit network or a network only using afp or netbeui (the latters should no longer exist) - TCP/IP is mandatory.

    either mac and pc have the tools built in to check if a network is healthy and a connection can be established.

     

    what i have not yet checked in detail is network over firewire, but this should be only needed at rare occasions.

    christian


    and remember: only a CRAY can run an endless loop in just three seconds.
  • Interesting, having had a quick read through on the (now working) vicontrol site it looks like the demo rig was between two macpro's.  Which begs the question about multiple hosts and how well it works when you have more than 1 slave. 

    More than 1 slave = more network traffic = potential timeouts on ethernet?

    Hopefully some official info will come out soon - maybe they're still debating the price, who knows - personally some info and no price would be great.  If it works as well as we all hope then price might well be pretty immaterial. (well almost)


  • apperently vista 64 on the macPros has not even been launched ;-) 

    if you can't load enough into one slave (because you're limited with RAM) use two. or three ...

    i think i did the math already: 24bit 44.1 stereo is ~260 kB/s that gives times 64 tracks ~ 16.5 MB/s which is close beyond 100 Mbit/s but far below 1Gbit/s ...

    assuming now theoretically _every_ package would collide we are still below 50% of the gigabit network capacity (caused by the repeated send of packages) we would still be on the safe side. such a collided package would be delayed for average 14 microseconds, which is more than a factor of 1000 below usual soundcard latency.

    so to talk about network timeouts isn't of any relevance - there are a lot of other reasons for dropouts way before ...

    christian


    and remember: only a CRAY can run an endless loop in just three seconds.
  • Excellent news cm.  Do you have any stats on how big a farm you've used internally with VE3? 3,4,10 slaves?  Also when will the wait on official info be over - we're dying out here....


  • Christian:

    I note your lack of enthusiasm for PPC Macs with VE 3. I currently own a dual 2.5GHz G5 machine and am planning to buy a Mac Pro 8 processor machine as my main computer and to use it for Logic, DP etc. and samples while retaining the G5 as a slave for more samples. Is this unlikely to work well with VE 3? Would I be better off selling the G5 and replacing it with something else? 

    I'd very much appreciate your thoughts,

    Stephen 


  • haha - its not a lack of enthusiasm, it is just that they do not perform (as we understand performance currently) - you can't open as much instances on a G5 as you can on a macPro ... is it the little vs. big endian thing or the processor pipeline, don't ask me why in detail ...

    the 8-cores ... if you like to hear my opinion ... they are only a minor step forward in relation to the last generation macPros, actually not even compared to the 4-cores. i'm not even sure everything is scaling as expected across the 8 proc. intel themselves releases some *code optimization tool* lately to allow full access to the capabilties of these processors ...

    regarding several reports about the new firewire chip and strange USB behaviour it would maybe a good idea to let their first release pass by until everything is clear ...

    as i read the news the Q9000 (quadcore penryn) series is a little bit delayed but should be delivered soon.

    my 2c only, christian


  • You're warming my quad-core heart![:)]

    PL 


  • last edited
    last edited

    @nomad said:

    haha - its not a lack of enthusiasm, it is just that they do not perform (as we understand performance currently) - you can't open as much instances on a G5 as you can on a macPro ... is it the little vs. big endian thing or the processor pipeline, don't ask me why in detail ...

     

    the 8-cores ... if you like to hear my opinion ... they are only a minor step forward in relation to the last generation macPros, actually not even compared to the 4-cores. i'm not even sure everything is scaling as expected across the 8 proc. intel themselves releases some *code optimization tool* lately to allow full access to the capabilties of these processors ...

    regarding several reports about the new firewire chip and strange USB behaviour it would maybe a good idea to let their first release pass by until everything is clear ...

     

    as i read the news the Q9000 (quadcore penryn) series is a little bit delayed but should be delivered soon.

    my 2c only, christian



    The "Harpertown" in the new 8-core Mac Pro _is_ a quad core Penryn - as I understand it, the Q9000 is a "Core" (not "Xeon") line of the Penryn family, which means it will be targeted for future iMacs, not Mac Pros. El Reg clears things up a little:

    http://www.reghardware.co.uk/2007/09/19/idf_intel_penryn_decoder/

    See: clear as mud! 😉

    But, I have to say a "minor step forward" they most certainly are not. Lower power (and virtually silent), faster, cheaper, pci-e 2.0, and so on... I'm also really curious about rumors suggesting that OS X 10.5.2 will somehow make use of SSE4 - if it does, then we should see significant performance gains in certain areas. Time will tell, I guess. I bought the day after this revision was released - in spite of long history with Apple, which has made me fully aware of the potential dangers of buying "rev a" anything. However, my feeling was that most of the development in this revision was coming from Intel, not Apple, so the logic board _design_ shouldn't have changed too much (just the various chips running the show). Anyway, no USB or Firewire problems to report, at this early date. My audio interface is Firewire (Apogee), and it's been flawless.
    As revisions go, obviously the next gen will be "better", but it will probably continue to build on the current Penryn line. If you have the patience of a saint, and basically don't need to get anything done this year, then wait for Nehalem - that promises to be a real beast. For myself, I had already waited several months for Penryn, and I'm glad I finally "jumped in".

    J.

  • good to hear everything works flawless with your 8-core, jbm - its just that several reports reached my desk that USB gets lost unexpectedly.

    i've been under the impression the new macPro is using the E5472 XEON with an intel 5000 family chipset and here i'm considering the processor bus to be too small for 8 cores (or better: for what 8 cores could process in the sampling world)

    christian


    and remember: only a CRAY can run an endless loop in just three seconds.
  • Hey Christian.

    Yeah, it's a 5400 of some kind... I'm not sure exactly which one (you're probably right about 5472). I actually did have my mouse disappear when I first booted the machine, but I updated to the latest Logitech drivers, and it's been fine since - so I assumed it was a driver issue. I'll be interested to see if it continues to stay put now. But my m-Audio keyboard (class-compliant USB, thank god - no need to rely on m-Audio _writing_ a driver) has been totally fine. What do you care about USB anyway! ;-)

    Interesting about the bus - I wasn't aware of that potential bottleneck. Mind you, it doesn't seem like much of anything in the software world is even close to managing "what 8 cores could process" anyway, at this point. I'm curious to see how much improvement I see over the coming year, or so, as apps get proper Leopard updates, and start to take advantage of all 8 cores - not to mention SSE4. Should be interesting. I remember some pretty astonishing performance boosts, in my PPC days, when altivec first started showing up in apps. I had a G3, and upgraded it to a G4 with only a slightly higher clock speed. But the performance gains were astonishing. Some apps which were almost useless without altivec were dramatically improved once optimized to take advantage of it. Mind you, I'm kind of new to the Intel world, so I'm not absolutely clear on how comparable SSE is to Altivec. It also doesn't seem certain that many developers are even going to bother with it. Do the VIs take advantage of SSE4? Will they? Is there any point in VSL implementing it, or does it have to be done at a lower level? I'm genuinely curious about all this...

    J.

  • oh yeah... and ALL of my Vienna Keys have stay put! They worked right from first install, and are happily reporting licenses every day!!!

    And the really big thing for me?... 16GB of RAM!!! It's a dream come true! Bliss, really. Though, of course, a 64-bit VE for Mac would be the icing on the cake. Some day soon, I hope. ;-)

    J.

  • Although, "Hah, Hah!" I throughly enjoy sarcasm directed at me, my actual question remains unanswered. I can't help observing, from my own direct experience, that the G5 machine has somewhat limited capabilities in comparison the the Intel MacPro machines. I was asking if there were reasons beyond this that make it unsuitable as a slave when used wit VE 3.Obviously, if my intent is to use it as a slave I would not expect it to run a tremendous number of samples. But, would I be better off selling it replacing it with two MacBooks each with 4GB of RAM? Or with something else? Perhaps someone would be gracious enough to forgo the sarcasm and answer this question. Thanks, Stephen

  • I'd personally go and sell, but only as Apple's focus will be on Intel so at least you'll have a good base to move forward on.  I can see a point not too far in the future when PPC code is no more and apps developers will focus on Intel only.

    You do raise an interesting point though around buying Macbooks - I've never really considered buying a laptop as a slave as I always saw the cost as too high when you can buy 2 minis or an iMac for the same price.  Why not buy 4 x 2GB mini's and control them with leopard's screen sharing?

    And I do wish VSL would lift the veil of silence on all this - or at least give us a countdown on the front page.....


  • last edited
    last edited

    @stevesong said:

    Perhaps someone would be gracious enough to forgo the sarcasm and answer this question

    sorry if it came over that way ...

    we just noticed a significantly higher CPU load with PPC and the strange thing is, it puts also higher load on the (intel-)host.

     

    our test setup was the special edition demo #1 (24 tracks IIRC) and this has been too much ....

    christian 


    and remember: only a CRAY can run an endless loop in just three seconds.
  •  Christian:

    Thanks for the answer.

    Tim:

    I thought of the MacBooks because they are relatively inexpensive and accomodate 4GB of RAM whereas the current Mac Mini is limited to 2GB of RAM. So the two Macbooks might double as mobile system as well being slaves in the studio. First thing is to obtain a MacPro and see what can be done with it - - I'm waiting a bit until some of the kinks are worked out with firmware updates and OS 10.5.2. Perhaps by then there will be new Mac Minis or some other solution.  


  • Hi Christian,

    How would a Mac Mini cope with being a VE 3.0 slave? I was thinking of purchasing 2 mac minis, upgrading the hard disk and upgrading the RAM to 3GB.

    Would these upgraded minis suffice?

    Thanks