PVS in QSearch

[theturk1234]

So I'm probably overlooking something obvious, but should I be using PVS in QSearch as well as AB?

Your thoughts?

David Cimbalista

[hgm]

I thought normally PVS was ot used below a depth of 2 ply, because the probability that the first move you try is the best one is too small to make it pay. Note that at those low depth you usually don't have a hash move, as the positions were not yet visited on the previous iteration.

[Sven]

Wouldn't this fit into the programming forum?

[D Sceviour]

I thought normally PVS was ot used below a depth of 2 ply, because the probability that the first move you try is the best one is too small to make it pay. Note that at those low depth you usually don't have a hash move, as the positions were not yet visited on the previous iteration.

What you write certainly makes theoretical sense. However, I tried PVS below depth 2 and it slightly increased the number of nodes searched in some bench tests. A depth of 1 made no difference at all. Is there a position test to show it would make any difference?

Sven,
You are correct. It probably would be better but there is always room for mistakes.

[theturk1234]

Sorry guys, I put it in the wrong forum
If any of you would be kind enough to tell me how to move it I'd be glad to!

We should have excellent move ordering in Qsearch because of MVV-LVA. Good move ordering = fast PVS.
That's as far as my thinking takes me.

[D Sceviour]

Sorry guys, I put it in the wrong forum
If any of you would be kind enough to tell me how to move it I'd be glad to!

There is a 15-minute edit period available to change your post. After that, it is locked. I do not know if the moderators can change it. There should be a special request to fix irritating bugs that are discovered too late.

[theturk1234]

Ok, sorry about that