Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A Very Good Test Position For Dedicated Units

Author: Drexel,Michael

Date: 01:10:42 07/03/04

Go up one level in this thread


On July 02, 2004 at 22:49:29, Bruce Humphrey wrote:

>On July 02, 2004 at 08:59:46, Mark R. Anderson wrote:
>
>>On July 01, 2004 at 22:55:30, Bruce Humphrey wrote:
>>
>>>On June 30, 2004 at 18:48:13, Mike Byrne wrote:
>>>
>>>>On June 30, 2004 at 08:56:28, Daniel Jackson wrote:
>>>>
>>>>>[D]2r4k/ppqbpp1p/3p1npQ/3R4/2r1P1P1/2N2P2/PPP1N3/1K5R w
>>>>>
>>>>>
>>>>>I tested this position with the Mach III Master back in 1989, it took about
>>>>>1hr.09min. to find Rf5!! with the double threat of Rxf6! and Nd5!
>>>>>
>>>>>Daniel
>>>>
>>>>Pocket PC Genius                1:58
>>>>Pocket Grandmaster (Ruffian)  ~12:00
>>>>Pocket Fritz 2                ~31:20 (on the 12th ply )
>>>>
>>>>
>>>>Dell Axim 590 Mhz (oc)
>>>
>>>Are you sure? You should be getting better results (I expect)! I thought genius
>>>2.1 for palm was the same soft (basically) than pocket pc genius? Either that,
>>>or it is true that Pocket PC (operating system) is a resources hog (in
>>>comparisson with palm and symbian)
>>>
>>>In my palm T3 (400 Mhz arm)
>>>-Hiarcs 9.32 palm:          depth 7, 23sec*
>>>-Genius 2.1 (arm):          depth 8, 01:44 , 61880 n/sec
>>>-Tiger 15.1:                depth 8, 04:33**
>>>
>>>*- Hiarcs does 'see it' at 18sec, but doesn't select the move until 23 sec
>>>
>>>**- In Tiger depth 8, at 180 sec Rf6 is move 31/51. Then at 232 sec it gives it
>>>a ! (meaning it has discovered it is a lot better than previously thought) and
>>>it ends selecting it at 273 sec
>>>
>>>Bruce Humphrey
>>
>>Bruce,
>>
>>How do you like your palm T3?  I have Hiarcs 9.32, Genius 2.1 and Tiger 15.1,
>>but they are running on my trusty old M130 (33 mzh, non-ARM).  I thirst for more
>>mhz and optimized engines!  Do you recommend the T3?  It's quite pricey and I
>>only want to get one if it's worth it.  Another option I am mulling over is a
>>palm Zire 71 or the newer 72.  Any thoughts on those?
>>
>>One thing I like about going with the T3 or the 71 (perhaps the 72 also) is that
>>I can still use my Palm "Power To Go".  This is a battery that attaches to the
>>back of the unit to recharge it on the go, but can also be used as a battery
>>extender ... I usually leave it attached to my palm and it then has a much
>>bigger battery capacity.  You can even put the palm unit with the power-to-go
>>attached in the hotsync cradle to hotsync your palm and recharge both units at
>>once.  It's *wonderful* for chess!  It keeps my unit charged up while I play and
>>gives it (I would estimate) about 3-4 times the power.  It can be purchased for
>>only about $37 (no tax or shipping) from Amazon.  I would HIGHLY recommend it.
>>Any of you people who complain about low battery life on the Tungsten or Zires
>>... this would make your concern *go away* for about $37.  It works with the
>>Tungsten "T", "W" and "C" series (not E), and the Zire 71 (and perhaps some
>>older Zires), as well as my M130.  Enjoy!
>>
>>Mark Anderson
>
>First of all, thank you for the "power to go" information. I have seen it in
>same place I bought my palm T3 unit.
>
>I "upgraded" from an HP Jornada 728 (clamshell with keyboard... keyboard was a
>must then for typing text). I simply got fed and tired of Pocket PC (I worked
>with Cassiopea and Dell Axims) and decided to switch to palm. Now I find palm
>design is wonderful, a lot easier and comfortable than pocket PCs.
>
>About upgrading to ARM... well... Genius is around 20 times faster there! Chess
>Tiger is not arm optimised yet, but it will be. And HIARCS, well, HIARCS is a
>dream come true talking about power in your hand!
>
>According to http://www.chesstiger.com, info on different hardware section:
>The difference in ELO from Z71 to T3 is: 2155 to 2293.
>This is, of course, current chess tiger, non ARM version. This is an evaluation
>based on a formula Theron explains in his website.
>
>According to http://www.chessgenius.com, arm version is 400 elo points stronger
>than non arm version.
>
>According to ssdf, genius is 1870, chesstiger is 2101 (on a palm m515).
>
>Didzis Cirulis published his match results when testing genius arm. That was
>+6-1=5 for Genius.
>
>So this is, according to elo tables, a performance of 2259 vs a 2101 opponent.
>That would mean genius has gone up (yes, I know I'm using ARM results
>extrapolating on a non arm ELO) 389 elo points! (quite near the 400 claimed at
>the website)
>
>From http://www.hiarcs.com , we get this elo comparison between the 3 products,
>which is, in its turn, based on the elo for T3 published in both websites
>(genius' and tiger's):
>HIARCS: 2591
>GENIUS: 2447
>TIGER:  2293
>
>My estimation is that tiger for arm will have 2563. Tiger 15 has in SSDF 28 elo
>less than Hiarcs 9, so... I just used that figure to get the estimate.
>
>Someone said in this group they can't believe a 2591 rating for hiarcs in a
>palm... well, if you look at SSDF list (2004-04-22) you can see:
>
>25 Chess Tiger 15.0 128MB K6-2 450 MHz  2666 34 -33 433 52% 2651
>35 Hiarcs 8.0 128MB K6-2 450 MHz  2588 33 -34 440 42% 2648
>
>Can we believe this numbers? Because they are 'close' in bare speed and memory
>to a palm T3 (half the memory, 400 MHz).
>
>To get an approximation for Hiarcs 9.0 in a 450, I base my calculation in the
>elo it says it has for a 1200, 2747. Tiger 15 in said machine has 2719. So, once
>again extrapolating, we could say Hiarcs 9 for 450 Mhz machine has 2666+28 =
>2694. So, once again I ask... if we do believe SSDF list for pentiums, why not
>100 less elo for palm??? It sound right to me.
>
>About your initial question... T3 or Z71, here is my thought:
>
>The T3 is 4 times faster (almost) than the Z71, but... there is a new palm
>coming out soon, with new OS 6, and that will be faster. Don't know what that
>will mean for prices, but probably T3 will go down :-) Of course, it is the
>usual dilemma: should I wait? I couldn't...
>
>I'm sure my math and logic have some glaring error I have not taken into
>account, and know they will be pointed out soon :-)

Yes, of course :-)

They talk about estimated Fide ELO, not SSDF rating.
There is a huge difference.

According to your logic Shredder 8 has about 2818 ELO Fide on A1200.
Now we assume (simplifying) 50 ELO gain for doubling the clock speed.
So Shredder 8 has about 2968 ELO Fide on a Quad Opteron.

You get the point?

Btw,a Palm T3 400Mhz is close in speed to a Pentium 100 for Chess.
Hashtables at a Palm are also not nearly the 80MB used by SSDF for 128MB
machines.

Michael


>
>Bruce Humphrey



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.