Author: Dieter Buerssner
Date: 15:28:18 04/01/03
Go up one level in this thread
On March 31, 2003 at 20:28:16, Robert Hyatt wrote: >On March 31, 2003 at 15:45:38, Dieter Buerssner wrote: >>Alex was speaking about hard disk caching of the OS. This can have an influence >>in situations with many TB hits. >Perhaps. I would use the internal LRU cache built into EGTB, rather than an >external >disc cache program myself... It is probably better. It ist not that clear. I made several tests here. I think the situation is comparable to L1 and L2 level memory caches. If you have the position in the internal cache, of course the access will be very fast. The OS hard disk cache can store many more positions in the same RAM however (because it is compressed at that stage, while the internal cache is uncompressed). OTOH, fetching a position from the HD-cache will need CPU time to uncompress it (but still faster than fetching it directly from HD with a physical disk access, obviously) So, if we don't have unlimited memory available, some advantageous compromise can be done. And this fits Alex' question (the chunk sizes the HD cache gets organized by the OS). Regards, Dieter
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.