this is a dumb question, but it has been burning in my head for a while.
I use giga samples such as GOS lite and sam horns. Even with this lite set, my memory (half a gig RAM) gets full with a typical full orchestra loaded. My sample libraries do not total 2Gigs.
Now, I wonder how on earth is it possible to load libraries like VSL (80 gig?), GOS (>10 gig I guess) etc.,, even if the RAM is 1.5Gb ? I understand that one way is to bounce tracks to audio but then, this is quite painful of you wanna use the numerous articulations as in VSL.
it just beats me. I am looking to expand my gear one day so it would be comforting to know how this is done.
First, get 1.5 or 2.0 GB. Then streamline your templates to use only what you really use. After that you have two choices: render some of your tracks to wave and load other instruments, or buy more computers, soundcards and Giga licenses.
Max at FxTeleport.com is working on a VSTi Giga wrapper. Combining that with Gigabit Ethernet and FxTeleport will let us use software and a network, rather than soundcards.
I really don´t know what have you been doing with all that memory... My computer has 768 Mb, and I have never used more than that up to now. My orchestrations usually have a complete string set, plus woodwinds and brass. In the end, its all a matter of getting things organized. I use VSL Cube and Reason in my compositions. The average samples are 25 - 30 Mb. With 1.5 Gb you could use about 50 instruments at the same time, and you wouldn´t need to read directly from HD yet! Unless you are John Williams, there is no need for more than that... oh man, these people are spending their resources I don´t know where.
Try not to use only large samples. Find the ones that sound good and are smaller. Don´t use a lot of articulations, you don´t need a lot of them to get a realistic sound, maybe its the orchestration that needs to be improved. Just use a few and in the right moment.
Don´t believe those who say that you can´t do a good job without 2 Gb of RAM. It is simply not true.
It may be necessary in some extreme cases, but not all the time. Definitely.
It also depends on the number of velocity layers. I use KHSS a lot, and it has 8-16 layers for most articulations. I have 1.5 GB of which I can use about 1.1 GB for Giga. My full KHSS template uses just over half. I can add WSWC, GigaPiano and enough percussion to do a click track and I get right up to the magic 99% number. Of course, if I don\'t load trills, tremolos, slides and other thing I don\'t happen to use in a given composition, I can load more variety.
Remember that the length of a sample doesn\'t affect RAM usage. It\'s the number of unique notes, articulations and velocity layers. Even if they are staccatos you need to load their sample heads into RAM.
One approach would be to only load your most playable articulation for each instrument, for instance, a sustain with hard attack. You can then render things off and load the articulations a section at a time to flesh out the details. Many libs (KHSS included) include instruments that only have a single velocity layer. Using these would really save RAM for the first pass.