Author: Robert Hyatt
Date: 08:15:20 08/11/00
Go up one level in this thread
On August 10, 2000 at 13:45:11, J. Wesley Cleveland wrote: >On August 09, 2000 at 09:48:21, Robert Hyatt wrote: > >[snip] >> >>Two things have helped to offset the non-improving memory speed: cache and >>wider bus transfers. Some memory technologies (IE SDRAM) is no faster on >>random access than its predecessor, but if you read a big chunk of memory, >>all of the 8-byte reads (except for the first) are faster. As there is some >>internal buffering going on that helps for large transfers. But that raw >>latency is still there. > >I wonder if it would make crafty faster to reorganize the hash table into 48 >byte chunks, and read all three entries at once. This is on my to-do list... reading 32 bytes would be optimal, but it is advantageous to have one table twice as big as the other so that up to 3/4 of memory can be used. But at present it requires two cache line fills for the two probes (when two are needed of course), where using a 'bucket' would reduce this to 1 (or to maintain the present 1:2 size ratio of the two tables) this would become 1.5, which is not as good as 1.
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.