Vienna Symphonic Library Forum
Forum Statistics

196,770 users have contributed to 43,031 threads and 258,438 posts.

In the past 24 hours, we have 3 new thread(s), 14 new post(s) and 93 new user(s).

  • nick, you're right PC3200 = DDR400, sorry for the confusion
    christian

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

    @Another User said:

    at last nobody was willing to confirm that logic or any other application will _use_ the 8GB until now and i doubt it will, because it's still a 32bit application - how should it address more than 2GB? yes, there are some workarounds to address more (~2.8GB) using 32 bit, but the application and the OS has to support this - nothing can be found which confirms this

    please can somebody contradict and point me to a respective source?


    I can't contradict you, but what I have read is that Panther is not a 64-bit OS - it's a 32-bit OS with 64-bit memory addressing. That's how they're getting all the current programs to work on it.

    I hope to find out about Logic and 8GB at the AES convention tomorrow (he says, nervously watching the time and realizing he still has to pack and catch a plane in three hours!).

    CM, I also want to explain what you probably are aware of: the size of the orchestrations is only one factor. As a matter of fact, last week I finished a huge orchestral/techno cue with everything but the kitchen sink in it, and the entire orchestra - all VSL - doesn't even fill up my 2GB Giga machine (all that's running on the Mac is Reason and a couple of tracks of Kompakt drums). Only a couple of the instruments are performace legatos, but the point remains: you want enough of the orchestra loaded and ready to go while you're writing. Otherwise it's like a violinist stopping to take another instrument out of the case every time he/she switches from long to short bowings!

    Finally, Evyn, can you explain how you're loading 5GB into EXS! I want one of those!

  • last edited
    last edited

    @evanevans said:

    ... LOGIC doesn't access memory, it REQUESTS it. .... That's how right now on a 2GB G4 I can have 5GB loaded into LOGIC with no problems.

    evan, there must be some misunderstanding going on, but maybe i'm simply not clever enough to understand it ...
    exs uses logic's audio-engine to stream samples, having a descrete portion of them in ram, so it's impossible to *load* 5GB into 2GB ram. it _is_ possible to access an amount of 5 GB samples loading a certain percentage/part of each sample (depending on your settings each sample takes 100-170 Kb memory for buffering) so if one has 1 GB free memory you can access about 8000 samples, independently how large each single sample is.
    nevertheless, if logic is able to address more than 2GB of requested memory or not is still not answered - but we will find that out soon
    christian

    I am telling you that based on my work under-the-hood of OSX, namely looking inside the UNIX based core Darwin, that no application truly accesses RAM. When LOGIC loads up the start-buffers for disk streaming for instruments, it tells the system I need THESE readily available. That's all it can really ask. The way the computer handles that request is up to the OS, in this case OSX, and Panther is going to be even more efficient ... but, basically what happens is the system says to itself, OK, you are requesting 5GB of readily available space, so I will put 1GB of that into RAM and 4GB onto a disk buffer. But, let's say in theory that hard drives were faster than RAM. Well it's going to be up to the OS to decide how to manage the 5GB of info that LOGIC is requesting get high priority so it's readily available. So in that theoretical case the OS will make the best performance decisions and put it all on the hard drive.

    Anyway, you don't have to wait to find out if this is true. I am telling you it's the case. Last night I loaded 7GB worth of streams into LOGIC. That was split into 5.8GB of disk buffer and 1.2GB RAM.

    I am going to say something here and I want you to repeat it like it's a mantra, all together now:

    "THERE ARE NO RAM LIMITS IN OSX. APPLICATIONS DON'T ACCESS RAM."

    It's hard to wrap your mind around it, but it's based on sophisticated new system architecture. When announcements of OSX were made years ago and they said it is a revolutionary system, unprecedented, and the best readily available commercial one available on the planet, they said that for good reasons. Eliminating RAM issues, was one, pre-emptive multitasking was another, stability, ... the list goes way on and on. If you aren't a OSX user then it's hard to appreciate because most of what makes it such a great operating system goes on in the background. But once you have used OSX for some intensive tasks, going back to XP Pro, ME, OS9, or whatever, just feels retarded, in the classical sense of the word.

    Evan Evans

  • evan, besides i'm using osX since it's beta and also IRIX and BSD (which is the true origin of osX btw), multitasking is not a sensation ourdays - no?
    i'm willing to follow your mantra, but tell me: what sense would be in getting samples from a harddisk, read them into a buffer on (another) harddisk and make an application beleive they will be immediately available - none (ok, using a solid state disk this would probably work). means: i can't beleive logic allows the OS to store streaming buffers on a harddisk ... ahh - you're talking maybe about audio-tracks?
    anyway - i'll report back, how much samples we can fill into a G5, because this has been one of the most requested answers during the last weeks
    christian

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

    @Another User said:

    ...tell me: what sense would be in getting samples from a harddisk, read them into a buffer on (another) harddisk and make an application believe they will be immediately available - none (ok, using a solid state disk this would probably work). means: i can't beleive logic allows the OS to store streaming buffers on a harddisk ... ahh - you're talking maybe about audio-tracks?


    no I am not talking about audio-tracks. I am talking about Audio Instruments, nameley EXS24mkII with VSL samples loaded. let me explain in detail:

    The sense is that without having to rewrite their application, LOGIC could take advantage of any faster methods of data I/O that the OS would allow. You have to think with an open mind. Do not presume to think that RAM might be the fastest I/O method around. Soon we may see some new technology on a PCI card that stores data in crystals using laser data retrieval. Who knows? But, with the right drivers and the right OS update, OSX will do what it needs to give any program that needs HIGH PRIORITY I/O, the data from that location.

    The point is, the computer is supposed to be the monster machine not the application. The application is supposed to provide functionality. And the OS in Apple's case is what makes it all happen. The OS manages prioiritized tasks as best as it can. If the OS ever gave control of it's host machine's RAM to an applicataion you would surely have machine crashes. But since OSX traps every application to their own space in virtual-land, it doesn't harm the system in any way because all I/O from the application goes through the OS. This is what gives us our new smile [:D] when an applicaiton crashes, and we don'thave to restart our computer and debug conflicting extensions and all that old school stuff. You just re-launch the app.

    So, to answer your question, it makes all the sense that making OSX in the first place makes to have the data requested by LOGIC that it expects to be in RAM to be treated with the highest priority by the OS. Although the OS puts some of what's supposed to be in RAM onto the HD, with modern computers it is becoming to be irrelavant. But instead of having Emagic make Logic aware of how every different system reacts to different RAM installations etc., the OS just does it all. So as new computers come out and new technologies come out, all your old apps work without rewrites. Because if Apple comes up with some new more efficient way of passing fast data I/O of Applications, then Apple will supply the update, and ALL of teh applications running on that computer will benefit.

    hoenstly, it's the way it should be. The technology should be harnassed by the OS first. It's not the application's place to assume that a certian computer or configuration is best for itself. if that were the case you'd see applications written only for specific computers/configurations.

    OSX really is a beauty. I cannot even fathom all of it's beauty. Very few can. You'd have to know all that's going on under-the-hood to really appreciate it.

    Now, to continue to answer your question about the effectiveness of such management, I think Emagic put a limit on how many samples you can have open to sort of optimize the live playback of their EXS sampler. Imagine if Logic had no limit on the number of samples that could be open. You could load 200,000 samples and OSX would put maybe 1.5GB in RAM and 15GB on disk buffer. When you playback, each stream would start with 10% in quick RAm playback and 90% coming from disk. This would result in anyof the following:
    1) stuttered playback intially
    2) Logic reporting disk too slow
    3) overload alert
    I think since Emagic knows how OSX works, they didn't wnat that happening. Which is why (in my other posts) they imposed a maximum of 8,192 files open at one time. And maybe that's based on installed RAM. They query that on opening logic (ever see the message "setting up memory buffers????) and than give you a scale(proportion) that would keep the disk stream buffers mostly in RAM to the point where live playback wouldn't suffer. The limit that Emagic imposes (if they do, it is still my presumption), would automatically ensure that most strams are coming from the RAM.

    So ,if that's true, than Emagic did there own sort of "forcing" the system to use RAM, but ina creative way. unfortunately their creative way was to limit how many samples you can have loaded, even though OSX would let you load (based on my calculations with 2GB Ram), about 5-times as many, or around 50,000 samples. That's the practical limit under OSX right now. With Panther it will become UNLIMITED. Only dependant on how much you want to bog down your computer. (now do you know why companies use these boxes as servers? because they won't crash when there's lot's of users. Just go slow. And Panther will destroy the competition in this area)

    You might be able to load 50,000 samples on current G4 systems, and the CPU might be able to handle playback, but the disk retrieval may be too slow to make it happen live without stuttering or completely stopping. So you mixdown the whole thing, or solo tracks, or deal with some occasional stuttering.

    I much prefer a working unlimited system that works better when I use it the better way, than a system which LIMITS me. Say "NO" to limitations!

    Evan Evans

  • Wow! Is this true, or the raving of a maniac? Maybe both... it certainly sounds interesting.

  • last edited
    last edited

    @William said:

    Wow! Is this true, or the raving of a maniac? Maybe both... it certainly sounds interesting.


    LOL. [[;)]]

    No honestly, I am very intimate with the UNIX core and the bus architecture, the capabilities and limitations of OSX, G4s, G5s, LOGIC, EXS24mkII, and especially how it all related to VSL users.

    So please tell me, William, which thing did you want me to go into? You say "is this true?" and I said a lot of things. I will elaborate or confirm validity if you specify what you want me to answer.

    thanks,
    Evan Evans

  • Evan,

    I guess it's what you suggest about unlimited number of samples possible. (Though of course you're still limited in polyphony, aren't you?) If the operating system does this data management so well in relation to RAM, etc., and it can be done practically. I find the Mac os and EXS interesting because the problems I had with Gigastudio on XP pro tainted my view of those systems permanently.

  • last edited
    last edited

    @William said:

    Evan,

    I guess it's what you suggest about unlimited number of samples possible. (Though of course you're still limited in polyphony, aren't you?) If the operating system does this data management so well in relation to RAM, etc., and it can be done practically. I find the Mac os and EXS interesting because the problems I had with Gigastudio on XP pro tainted my view of those systems permanently.


    Indeed. There is a thread I started here somewhere about how OSX really has unlimited RAM. Think I called it something like the "G5 has more than 8GB of RAM", but it applies to any system running OSX. So read that thread too. It's VERY exciting stuff. But since OSX was released 3 years ago there was still a limit to how many files oyu could have open on the system. Panther eliminates that, REPORTEDLY. I should have that definitive answer maybe a week before it's release.

    You are limited in polyphony, but you have 64 Audio Instruments each with 64 note Polyphony. Isn't 4,096 notes of polyphony enough for you? When you start to get up there you are of course limited to the CPU power for LIVE playback of all teh notes. But of course bouncing down the master output always works fine. And you can always freeze tracks. But i never need to. I have a G4 DP 1.42Ghz with 2GB Ram. And I am using about 10GB of RAM-Space (read that other thread) without CPU performance problems.

    And having everything working in a protected OS branded by Apple is really a plus. As you said XP Pro, Giga and the rest can have their dilemmas.

    PS. You can AIM me anytime. I do VideoChat with iChat AV as well. My AIM address is at the button below my posts, always.

    Evan Evans

  • I did read the other thread - very interesting and I will eagerly read your results after some practical testing. I like what you say about science not being the same thing as skepticism - if you look at most of the great innovators, they were believers in something first, and skeptics second (though both attitudes are necessary in the long run).