Author: Chris Carson
Date: 04:28:03 03/17/03
Go up one level in this thread
On March 16, 2003 at 21:49:01, Christophe Theron wrote: >On March 16, 2003 at 19:53:10, Jason Waugh wrote: > >>I'll contribute what I can, for what it's worth: >> >>Using the test position posted.... >> >>Chess Tiger 15 (Palm Tungsten T, 1.5MB HT) - 390 NPS >>Chess Tiger 15 (Pentium II 400MHz w/Win2K, 384MB RAM, 96MB HT} - 82,000 NPS >> >> >>On the Tungsten T, CT15 runs under PACE and runs at the same speed as it does on >>a 33MHz palm device. This isn't just "the word on the street" - I can vouch for >>this, as TigerMark (CT's built in benchmark) displayed the same score on my m505 >>as it does on my Tungsten. I cannot test overclocking since I no longer have my >>m505. When we can see how well it runs with some ARM code depends on >>Christophe. ;) >> >>Since similar engines are not available for Pocket PC and Palm, then the >>comparison can't exactly be scientific - but if it helps, I tested that position >>on my PC with Ruffian (96MB HT) and achieved 185,500 NPS >> >>(hmmn, interestingly enough, despite the fact that Ruffian was more than "twice >>as fast" - CT15 was well into 14 ply depth before the minute was up, Ruffian >>only in the middle of 12. Mind you, that doesn't necessarily mean anything >>either.) >> >> >>Jason. > > > >A small advice for the sake of reliability: > >The NPS in Chess Tiger is influenced by the amount of hash tables. I think any >other engine out there is similarly influenced. The more hash table, the slower >NPS you get. > >So you should test CT15 on Palm and on PC with the same hash table size (for >example 6Mb because it is reasonably possible to get this on the Tungsten-T). > > > > Christophe Thanks Christophe, good advice and keep up the great work on CT.
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.