On Thursday, Jul 17, 2003, at 08:03 America/Guayaquil, Berin Loritsch wrote:
[skipping nice parallel in digital audio]
The concept translates over to the adaptive cache approach at almost 1:1.
The actual individual cost sample is not important. It is the windowed average that is important.
Well, on second thought, I think you misinterpreted my pseudocode. There is only *one* value of efficiency per resource. This is updated as we go, incresing or decreasing depending on the sampled efficiency but we don't need to keep the history of it, just the final sum of all the terms. for this reason we don't need a window, nor the system will slow down later on.
-- Stefano.
