Good test for dedicated and PDA or PC.

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

auditt wrote:
Cubeman wrote:
auditt wrote:28 secs, Palm T5 !
That is a strange result, as I have Hiarcs 9.7 analysing this position too.
The engine is set to the default settings with 2048Kb hash.It has been looking at the position for over an hour and has the following output.
D12 5014s -0.23 Rxb2 Rxb2 Nxb2
D12 Qg8 4/37 38922972N.
As you can see it is now considering Qg8 but has not confirmed if it is the best this is now after 2hours and 43 sec.
If any one can reproduce the 28 secs with a Palm T5 or even T3 overclocked I will be amazed.
Yep, 29 Secs on a standard T5 (not overclocked), standard hash table of 2MB at time control G/30 !

D8 29s -0.26 Qg8

Actually the Res II-Engine X results don't look that impressive compared to my 'lowly' Palm (the HIARCS always seems very impressive on the Palm Intel Intel PXA261).

cheers, Mark
it is no surprise to me that Hiarcs on your Palm scores better then Engine X on the Res

after all ..the program you are using is ...
Hiarcs Regards
Steve
Cubeman
Member
Posts: 460
Joined: Tue Jul 31, 2007 10:50 pm

Post by Cubeman »

mclane wrote:if this is true than testing with this styletab hiarcs is very contraproductive.
could it be that hiarcs rus without hash on those emulations ?
Yes, as I always had my suspicions that the StyleTap version was different when I follow games played by Palm Hiarcs using the PPC the moves don't match.
I think this is a problem with the StyleTap emulation,as alot of programs that ran on my Palm T5 do not work properly when I try them on my PPC.
But most of the time when testing PPC Hiarcs the best move is still found in reasonable time, but I no longer can trust the output to be the same as the Palm version.
Lets hope that Mark will fix this situation by creating a native PPC version after the Palm version update.
pgeorges
Member
Posts: 42
Joined: Tue Jul 31, 2007 2:55 pm
Location: France
Contact:

Post by pgeorges »

Toga II 1.3.4 takes about 30 seconds to find Qg8 on my Pocket PC (624 MHz CPU), with 33000 nps.
Cubeman
Member
Posts: 460
Joined: Tue Jul 31, 2007 10:50 pm

Post by Cubeman »

Thanks Pascal, following your last post I decided to install Scid for PPC on my PPC and test some of the many engines that are available.
The best was Glaurung 1.2.1 which took 24 secs to see Qg8 as best and gave evaluation as 0.0 as the line was a perpetual, and took a total of just about 5 minutes to see an advantage, by this stage it was getting 37100 nps.
Next on the list was Viper 0.1 which just took 27 secs to see Qg8 evaluate to 0.0 and was getting 106153 nps, then after 3min 13 secs it sees advantage.
pgeorges
Member
Posts: 42
Joined: Tue Jul 31, 2007 2:55 pm
Location: France
Contact:

Post by pgeorges »

Note that since then I ported Fruit 2.3.1, which seems to be the best engine even if Glaurung is close to it.

Pascal
Post Reply