Author: Koundinya Veluri
Date: 04:41:50 09/26/03
Go up one level in this thread
On September 26, 2003 at 06:58:07, Sune Fischer wrote: >I think there are two downsides to doing this. > >One, you get a more complex code. Not that this needs to be an issue, just an >observation. I fancy simple things now, unless there is a really big win in >doing it complicated. A theoretical 1% win seems not that big to me. > >Two, you need to probe around in several tables instead of just one table, and >wouldn't this incur a huge latency penalty? > >-S. > For me, the advantage of using two tables is much higher than the latency penalty. Especially in long searches, I found that with just one hashtable using higher-depth replace scheme, the entries increase in depth enough that nothing will be stored near the leaves anymore. I haven't tested how much loss in NPS I get from using two tables instead of one, but I'm happy because solution times usually decreased. Regards, Koundinya
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.