Vienna Symphonic Library Forum
Forum Statistics

194,283 users have contributed to 42,914 threads and 257,948 posts.

In the past 24 hours, we have 2 new thread(s), 17 new post(s) and 86 new user(s).

  • Much as I'd like to, Plowman, I don't think I can take credit for this method, and I don't think you noticed me being cool to the idea of one computer being all we need at all. I'd like that very much. It's just that I don't see it happening soon - although it's getting better.

    If you look at the Activity Monitor dumps above, you can see the CPU percentage. The second one with all kinds of stuff going appears to be at about 80% - but that's of one of the two CPUs. i'll load those sequences when i get a chance and see what the CPU reading is with the transport stopped.

  • This topic demonstrates what I was fearing and explaining on this same board a few weeks ago. Going "full" 64-bit will *not* automatically give us the ability to use 16+ GB of samples.

    I suspected that there would be a tremendous amount of rewriting before that would happen.... it could turn out that I feared right!

    Jerome

  • Well, either way, I think it's fair to say that the industry would do well out of us consumers as a whole if it did certainly provide greater memory access for graphical, video and obviously music applications via whatever means works best. I don't see why there can't be multiple VSL servers controlled by a central administrative server that does not access the memory of the other independant apps, then we could have as many VSL servers running as necessary to utilise the amount of RAM in the machine. Then you could use 16Gb easily no problems. Of course I have no actual idea abou the technical implimentation of this but it seems to me that you could do it, after all, games do this: multiple servers controlled by a central server, only in this case, you have each server as a memory access device which accesses 1.2 Gigs RAM, not more, and once it is full it starts off handling memory holding to the next server (or rather than admin server does that).

    (edit) just to add of course the host app (logic etc) would interface with the memory via the admin server. The admin server would be like a router between the memory holding servers and the host. Might add some latency but on a fast machine, wouldn't matter. On a quad intel, with 16gigs of ram the problem is ram access not cpu anyway, you could just run your system on a lower buffer setting -seems worth it to be able to access all that ram.

    Miklos.

  • Anyway, my personal main problem is not enough cpu... The ram optimiser and freeze functions combined mean you can quickly enough optimise VI and I find it works excellently, not too hard to re-open , reset, re optimise when necessary too - with this feature you can get quite a lot out of a single instance as it is... The problem then becomes that each instance uses an amount of RAM in the server so you have to be efficient in using each instance to it's fullest in terms of loading sounds... Having multiple VSL servers seems like it would work on the surface but of course I'm sure there is a complex programming reason why it would not work well. Hats off to the masters at VSL for making VI as superb as it is. I for one am not complaining only wishing to be able to use it to the maximum capability or at least have that freedom when writing.

    Miklos.

  • last edited
    last edited

    @mpower88 said:

    .... On a quad intel, with 16gigs of ram the problem is ram access not cpu anyway, you could just run your system on a lower buffer setting -seems worth it to be able to access all that ram...

    Miklos.


    That sort of reinforces something that has been at the back of my mind for a long time.

    We have clearly reached a technological impass of sorts when some aspects of hardware meet or exceed the demands of software while other necessities of code structure prove to be counter productive. As long as VI users are willing to buy and sustain a network of comptuers then it may not be a problem.

    But at some point, something has got to give-- the need alone will force a change sooner or later-- always does.

    So maybe users of orchestral VIs represent a minority of computer users, but the term Pro User as far as the Mac Pro is concerned is in a strange way doing as much to alienate users as it attracts them.

    The Cube is an awesome accomplishment-- its sound exceeds anything believed to be possible on a computer, imho. We're just in a bit of a 3GB limbo for a while yet.

  • To me a network on computers is unfeasible and a pain to set up and administer and work with. I need that aspect to be as simple as possible when writing.

    Perhaps we simply the problem then solutions will follow from that.

    1. We need to access more RAM from VI - lets go with todays maximum - 14Gb (plus 2 for system).

    What can be done to make this happen in the near future? If people are able to run multiple standalone VI's and access more RAM then *surely* this is not as impossible as it seems, it's only impossible with the existing set up, can that be modified so that we have multiple VSL servers running? Or one VSL server with memory nodes as separate apps, that it uses to store memory in RAM? That would solve all the problems NOW and no need to go 64bit at any stage, and would work with existing hosts and OS.

    Miklos.

  • "I don't think you noticed me being cool to the idea of one computer being all we need at all."

    You have been cool to Apple's touting of 64-bit when so much remains to be done after it premieres. Consequently, you have noted that the one computer solution is not imminent. I know you don't like Apple "plastering" 64-bit on their marketing materials. Don't make me go into the archives.

    Not to be a pest, but you mentioned "sequences" playing. The numbers actually become less helpful in that case because we don't know how many voices and DSP's are affecting the CPU.

    The baseline I'd really like to see is the CPU hit you take when Logic's transport is running with full RAM loaded and nothing playing. Then I'd know if I got a 2 x 2.5 how much CPU I'm losing right out of the gate. As I mentioned, with loaded RAM on a dual 1.8, I sacrifice about 35% total CPU just by hitting play, even before I start writing. It's nearly prohibitive. I'm hoping on a dual 2.5 with the load you've illustrated, you'd only lose about 15 to 20% CPU with the SPL running / no music.

    Thanks for all you do, Nick and everyone else. And Jack, what a time you'll have. After you recoup (money, sleep, speaking voice), I'd love to know your take on things. Maybe you'll get a glimpse of Herb in the MIR. Wish I could be there.

    Think of all the free dinners Nick must get.

  • No such thing as a free dinner, Plowman. [:)]

    I'll find the seq and load it up.

    Oh, and you're right about what I said about 64-bit marketing. But I don't think I'm going far out on a limb - it's obvious that this isn't all going to change overnight; we're going to be using multiple computers for the forseeable future. And I say that even before taking into account that developers always write software that pushes the boundaries and brings current machines to their knees.

    Case in point: MIR.

  • last edited
    last edited

    @Nick Batzdorf said:

    No such thing as a free dinner, Plowman. [:)]

    I'll find the seq and load it up.

    Oh, and you're right about what I said about 64-bit marketing. But I don't think I'm going far out on a limb - it's obvious that this isn't all going to change overnight; we're going to be using multiple computers for the forseeable future. And I say that even before taking into account that developers always write software that pushes the boundaries and brings current machines to their knees.

    Case in point: MIR.


    I think you're right, Nick. The extra hype about Leopard's 64-bit capabilities had many wondering if networks would finally get smaller at least. I never thought they'd go away, but it did instill a rather false sense of optimism for the short term which has proven to be a little disappointing.

    But now that a year of Intel-Macs has passed we can at least budget more appropriately. I don't mind investing in a network if I know that I'll get 2-3 years out of it, but if computers were to really take a flying leap forward over the next 6-12 months and I've tossed tens of thousands of dollars into soon-to-be-antiquated technology-- well, let's just say it becomes less a matter of the actual dollar amount spent rather than the manner and purpose for which is might have been wasted.

    I really feel a lot better now knowing what hardware upgrades will be most practical over the next year or two. This doesn't change my wish list, but it does inspire a more realistic approach rooted in fact instead of spec.

  • last edited
    last edited
    Okay, I didn't even have to load it again. This is from the article in our current issue (Virtual Instruments mag). I see from my own figure caption that the sequence isn't playing:

    @Another User said:

    Fig. 1: 7GB on one machine, and it’s running reliably. This Activity Monitor screen dump shows 6.96GB of memory access in a G5 with 8GB loaded. The VSL-Server program is listing samples loaded into the plug-in versions of the VSL Vienna Instruments player running inside Logic Pro, which is accessing 1.38GB. (Logic is also running other programs, including Spectrasonics Stylus RMX.)

    Four stand-alone copies of the Vienna Instruments player (all named differently, which you don’t see) are running outside Logic. In addition, Native Instruments Kore is running several instruments, including Toontrack’s EZ-Drummer (a sample-playing instrument).

    While the sequence isn’t playing here, you can see from the idling percentages that the CPU hit isn’t going to be outrageous. That’s what’s changed recently to make this possible.


    http://homepage.mac.com/virtualinstruments/.Pictures/seq%20not%20playing.png">

  • Kore uses a fair amount of CPU, by the way, but the VSL and Logic programs aren't.

  • The fact is though that how many users need to have access to 3+ GB of memory for one single process? Besides high-end scientific and pro applications, no one.

    It's really a niche market, so I wouldn't be surprised that this takes some time to work.

    I'll always remember the words of one of Logic's top developers when I met with them about a year agor: "More than 3GB of Memory? What for?"

    (He wasn't joking).

    Jerome

  • last edited
    last edited

    @Jerome said:

    I'll always remember the words of one of Logic's top developers when I met with them about a year agor: "More than 3GB of Memory? What for?"

    (He wasn't joking).

    Jerome


    Hey-- let's head up to Cupertino, chain the guy to a 2.0 dual, hand him VSL's Cube....

    Then, let's insist that he "Mac Pro This" where the sun doesn't shine and see if 3GB out of 550+GB is not a tight fit! j/k

    Seriously, we're talking Apple and Logic. We're talking Pro Applications (Apple's term). We're now talking Mac Pro machines (also Apple's term). So, does Apple's definition of a Pro User not include those of us electronic musicians using such things as the Cube and other high end sample libraries?

    Just keeps me wondering and inspires greater patience.

  • I think the problem is that many if not most Apple's pro users do not absolutely need to access more than 3GB of memory... Think Final Cut studio apps or Aperture... granted, it'd be nice, but there are many other features more important to them. Teechnically, they are more looking for faster processors than more ram.

    Even when thinking about Logic, its main target is people in the recording industry (songwriters, producers, etc) or electronic musicians - not "traditionnal composers" (or film composers) per say. These "composers" are the only one who really need to load a bunch of samples, and can hardly be satifsied with the 3GB barrier.

    Jerome

  • last edited
    last edited

    @Jerome said:

    I think the problem is that many if not most Apple's pro users do not absolutely need to access more than 3GB of memory... Think Final Cut studio apps or Aperture... granted, it'd be nice, but there are many other features more important to them. Teechnically, they are more looking for faster processors than more ram.

    Even when thinking about Logic, its main target is people in the recording industry (songwriters, producers, etc) or electronic musicians - not "traditionnal composers" (or film composers) per say. These "composers" are the only one who really need to load a bunch of samples, and can hardly be satifsied with the 3GB barrier.

    Jerome


    LOL!

    Why is it that I now want to take RAM *out* of my computer? [:)]

    I had a long talk with a friend of mine who doesn't use VIs at all, but just software plugins and audio tracks. We have the same machine-- Dual PPC 2.5. He manages to do huge sessions with 6 or more Altiverbs among many other things-- and I-don't-know-how-many aux busses (20+?).

    2GB RAM. That's all he uses. I've tried to run the same size session on my computer with 4+ GB-- no dice.

    I think I should send my G5 in for a check up.

    (Determination is so exhausting.)

  • "While the sequence isn’t playing here, you can see from the idling percentages that the CPU hit isn’t going to be outrageous."

    If by "idle" you mean the transport isn't running, my concerns remain.

    When my slow-as-Mac-o-lasses dual 1.8 is loaded with 1 GB for VI and 0.5 GB for Logic, "idle" records a CPU of about 13% for VI and 31% for Logic. When I hit play for an empty song (RAM loaded but no music), VI's CPU goes to 25% and Logic goes to 60% with nothing playing. That's approximately double the "idle" CPU numbers.

    And what an irony we face, looking forward to 64-bit and expanded memory. Imagine the CPU you'll lose just getting out of the starting gate when you hit "play" with 10 GB loaded. But hey, maybe Clovertown will blow that away in an instant. Interesting, hopeful times. But cautious times as well.

  • Plowman, I'm exhausted now. [:)]

    You want to see it moving with nothing playing. Okay, maybe sometime I'll do that.

    But in all honesty I don't really care - the point is that it works on my ancient 2 x 2.5. The CPU doesn't go way up when I push play, by the way.

    Also note that it goes higher than 100%, depending on the situation, since there are two processors.

  • last edited
    last edited

    @Nick Batzdorf said:

    .... my ancient 2 x 2.5. ...


    [:'(] [:'(] [:'(]

    It's a shame when you have lightbulbs that burn for longer than a computer will stay current.

    ("current"--- no pun intended-- really!)

  • Plowman,

    I have to say that I'm seeing similar problems with my dual 1.8... maybe it's just a "dud"? It was fine for a couple of years, but the VIs appear to have pushed it over the edge. As I understand it, Apple did some board-level tweaking to the G5s after the dual 1.8 (mine is the original with 8 ram slots), but I've never had this confirmed. Anyway, I'm seriously considering a Mac Pro, it's just a question of when. We'll see what Apple announces later today -- hopefully something worth buying.

    J.

  • [H] MACWORLD