PVS inside the tree

[adieguez]

Hi, I have never used straight PVS inside the tree(yes in the root where the window is too big, but not inside the tree, where I use the window as it is most of the time). I know you are going to say that my move ordering may be bad... I will check the statistics (and Gian-carlo may say my program is buggy ). But what happens if you avoid PVS and do not make the window smaller if it is too small in the first place and searching killers for example?

[bob]

Hi, I have never used straight PVS inside the tree(yes in the root where the window is too big, but not inside the tree, where I use the window as it is most of the time). I know you are going to say that my move ordering may be bad... I will check the statistics (and Gian-carlo may say my program is buggy ). But what happens if you avoid PVS and do not make the window smaller if it is too small in the first place and searching killers for example?

The most efficient search of all is one that uses the A-B window of N, N+1, because every search from that position either fails high or fails low, which is extremely efficient. PVS inside the search simply exploits this characteristic by using a null-window, since in most cases all we want to know is "is this move better or worse than the best so far?"