Author: James T. Walker
Date: 10:41:27 11/29/02
Go up one level in this thread
On November 29, 2002 at 12:36:47, m.d.hurd wrote: >On November 29, 2002 at 11:47:34, James T. Walker wrote: > >>On November 29, 2002 at 03:05:19, m.d.hurd wrote: >> >>>On November 29, 2002 at 02:05:16, James T. Walker wrote: >>> >>>>On November 28, 2002 at 19:27:19, Jeremiah Penery wrote: >>>> >>>>>On November 28, 2002 at 17:11:08, James T. Walker wrote: >>>>> >>>>>>On November 28, 2002 at 16:54:25, mike schoonover wrote: >>>>>> >>>>>>>could it be that the master drive determines the access speed? >>>>>>>or,the system gives the master priority,hoging resources? >>>>>>>regards >>>>>>>mike >>>>>> >>>>>>Hello Mike, >>>>>>The only problem with that theory is the computer where the TB's are on the >>>>>>"Master" drive is the slow one. The fastest one uses the 8G/5400rpm as drive C >>>>>>(Master) and the 40G/7200rpm as drive D (Slave). The chess programs are on the >>>>>>"C" drive in both cases. >>>>>>Jim >>>>> >>>>>I would guess that the reason for the slowdown is because the virtual memory >>>>>(pagefile) is on the same drive as the tablebases. So, when it caches >>>>>information, it's copying on the same disk, while also trying to access other >>>>>tablebases. On the other machine, it copies to a different disk, which will be >>>>>much faster. >>>>> >>>>>Other things that can affect the access speed of tablebases are disk >>>>>fragmentation and the physical position of the tablebase files on the disk. If >>>>>the files are heavily fragmented on one system, and are contiguous on the other, >>>>>the fragmented ones should be quite a bit slower. Files near the beginning of >>>>>the disk have faster access times and faster transfer rates than files near the >>>>>end of the disk, as well. >>>> >>>>Thanks for trying to help. I did de-frag the drive on the slow computer and it >>>>seemed to help some. It now starts out at almost the same speed but slowly lags >>>>behind. For instance, both computers will reach 14 ply in 15 seconds on one >>>>computer and 16 seconds on the other. But when getting to say 16 ply at 2:15 on >>>>one computer it takes like 3:10 on the other. Also as you might guess, the NPS >>>>on the computers gets further apart with time. >>>>Jim >>> >>>Hello Jim >>> >>>One thing worth checking is the EGTB cache size, which is set in Options. Mine >>>was set at 1 mb. It might be worth experimenting with different sizes to see >>>what difference each size increase makes. The help file says to use a value of 1 >>>to 8 mbs. However I seem to recall someone mentioned 16 mb was a good size to >>>use. I suppose the bigger the cache, up to the optimum figure, the less disk >>>access. >>> >>>Regards >>> >>>Mike. >> >>Hello Mike, >>I'm using 16M I think. Anyway it's the same for both computers. I have gone >>through all Bios settings, all ".ini" files and everything I can think of to >>make sure the two computers are set up the same. The only difference is the >>video cards plus the hard drives are backwards from each other. It's a lot of >>work but I'm thinking of swapping the drives around in the slow machine to see >>if it helps. I don't think the difference is critical since it only shows up in >>late endgames when the tablebases are being hit hard. >>Jim > >I would imagine that the optimum EGTB cache would be different for each hard >drive. Some are faster at reading than others, some have a faster access time. > >So setting both machines at 16 mb might not be optimum. > > >Regards > >Mike I assumed it would be since the tablebases are on the 40G/7200rpm drives on both machines. I think I'm going to just reformat the HDs and start all over with the slow machine. Jim
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.