My point was that most of the speed increase is due to the zero width searches being faster than normal width searches of standard alpha-beta.
It is possible that more speed increase is due to finding more cutoffs in this method of search. It was a seat of the pants estimate, and I could easily be wrong. The best way to find out for sure it to measure it yourself, and I also suspect that the answer may even vary from program to program.
principle variation search
Re: principle variation search
It's called principal variationnak3c wrote:principle variation