_In my experience_, as long as you don\'t have more than 512 mb ram, just forget the whole thing. A small vcahce causes less polyphony, and a larger vcache causes higher latency. For me everything works best overall, when not using any specific vcache settings.
Interesting question there, Mihmar. The MaxFileCache setting is a very important parameter relating to disk buffer size, which has a big effect on overall computer performance. The trick though is that the right MaxFileCache setting depends on how the computer is used. For most applications, the higher the better since the disk is accessed fairly randomly and a higher buffer size will more likely have any requested data. However, a higher number is also taking valuable RAM away. A good rule of thumb according to the SiSoftware Sandra benchmark utility software is a MinFileCache of 1/8 of your RAM size and a MaxFileCache of 1/4 of your RAM size.
Note however that for computer audio applications, the situation is very different. Audio applications do a lot of serial disk read and writes so having a huge disk buffer will adversely affect audio track performance. Companies like Cakewalk and Frontier Design actually recommend a very low MaxFileCache of 4096. Also, a high MaxFileCache will take away valuable RAM from music sample loading (as well as from other applications).
Unfortunately, this does not seem to be the end of the story, because it seems that a low MaxFileCache will reduce GigaStudio polyphony. This is a strange observation, but it seems true. See the topic on \"can\'t handle polyphony\" in the Hardware Issues section. So now there is a trade off between low MaxFileCache and good audio track performance and high MaxFileCache and good GigaStudio polyphony. The right MaxFileCache value will need to be optimized by finding the lowest one that gives good GigaStudio polyphony and sustain pedal use. GigaStudio polyphony is probably more directly related to CPU processing speed so if your machine and disk is fast enough, you may be able to get 160 voices with a recommended MaxFileCache of 4096. In my case (850 MHz Pentium III), I can maximize GigaStudio polyphony and sustain pedal use with a MaxFileCache of around 24576.
[This message has been edited by Mega (edited 06-18-2001).]
I think the vcache should be set if you are doing audio applications. I wouldn\'t trust Microsoft Windows in the context of audio (or even in other contexts) to manage such a vital use of RAM as caching data from the hard drive. Otherwise, the vcache can just keep growing and growing and unlike the virtual memory swap file (which by the way should be overridden to a constant value like 256 Mb or 2 x RAM for audio applications), vcache never seems to let go of the memory that it grabs. I have seen it grab hundreds of Megabytes of RAM without ever decreasing. This may result in speedy performance for many applications, but not for audio, and even for other applications it is not an efficient use of RAM. So for general applications, some sort of vcache setting is useful, like the MinFileCache of 1/8 RAM and MaxFileCache of 1/4 RAM as recommended by the Sandra benchmark utility. In audio applications, the lower the MaxFileCache the better (with MinFileCache set equal to MaxFileCache) such as a MaxFileCache of 4096 as recommended by Cakewalk and Frontier Design. If you notice that GigaStudio polyphony can be improved by increasing the MaxFileCache, then the value can be increased but by the smallest amount possible to keep it optimized. As Blob pointed out in \"can\'t handle polyphony\" in the Hardware Issues section, the ChunkSize is also an important parameter. I get good results with 512, while he seems to prefer 1024.