Fractional plies and transposition tables

Discussion of chess software programming and technical issues.

Moderators: hgm, Harvey Williamson, bob

Post Reply
brtzsnr
Posts: 426
Joined: Fri Jan 16, 2015 3:02 pm
Contact:

Fractional plies and transposition tables

Post by brtzsnr » Sat Apr 18, 2015 2:30 pm

Which depth should I store in the transposition table?

1) Remaining fractional depth
2) Remaining plies left (excluding extension)?

I use the second and I remember trying the first option. However it doesn't seem natural to me that I need to track both ply and fractional depth.

bob
Posts: 20340
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Fractional plies and transposition tables

Post by bob » Sat Apr 18, 2015 3:30 pm

brtzsnr wrote:Which depth should I store in the transposition table?

1) Remaining fractional depth
2) Remaining plies left (excluding extension)?

I use the second and I remember trying the first option. However it doesn't seem natural to me that I need to track both ply and fractional depth.
Has to be the remaining depth, how ever you use it. Otherwise you can never get an accurate comparison when you ask "is the draft sufficient?" When I did fractional plies I stored fractional plies. It's the only thing that makes sense.

User avatar
hgm
Posts: 22075
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Contact:

Re: Fractional plies and transposition tables

Post by hgm » Sun Apr 19, 2015 8:25 am

As Bob says, the only purpose of this item is to judge if the stored data is good enough to satisfy the probe request. That means it should also encode differences between PV searches and cut-node or all-node searches, if your engine prunes those differently.

brtzsnr
Posts: 426
Joined: Fri Jan 16, 2015 3:02 pm
Contact:

Re: Fractional plies and transposition tables

Post by brtzsnr » Mon Apr 20, 2015 9:52 pm

bob wrote: Has to be the remaining depth, how ever you use it. Otherwise you can never get an accurate comparison when you ask "is the draft sufficient?" When I did fractional plies I stored fractional plies. It's the only thing that makes sense.
Thanks. Tried fractional depth and it gave a 6ELO increase. I wonder why it failed the first time I tried.

Post Reply