Author: Dann Corbit
Date: 21:09:04 01/17/03
Go up one level in this thread
On January 17, 2003 at 22:30:11, Jorge Pichard wrote:
>On January 17, 2003 at 22:17:33, Jorge Pichard wrote:
>
>>On January 17, 2003 at 20:01:04, enrico carrisco wrote:
>>
>>>On January 17, 2003 at 19:22:56, Dann Corbit wrote:
>>>
>>>>On January 17, 2003 at 18:47:59, Jorge Pichard wrote:
>>>>
>>>>>On January 17, 2003 at 18:39:29, Jorge Pichard wrote:
>>>>>
>>>>>>So the Fritzes are still the best!
>>>>>>
>>>>>>http://www.chessbase.com/newsdetail.asp?newsid=724
>>>>
>>>>None of your conclusions follow from the data presented.
>>>
>>>That's what I love about average user interpretations of the SSDF list. Armed
>>>and dangerous with "theories."
>>>
>>>-elc.
>
>
>What did I missed here? I didn't say better then all the commecials. Simply do
>the comparison. Sorry but this is NOT a theory, it is simply one of the best way
>to compare the strength of these programs, unless you have a better rating
>system than the SSDF.
>
>
>10 Hiarcs 8.0 256MB Athlon 1200 MHz 2679 25 -25 772 55% 2642 11
>Ruffian 1.0.1 256MB Athlon 1200 MHz 2676 39 -37 346 60% 2604 12
>Rebel Century 4.0 256MB Athlon 1200 MHz 2673 29 -29 590 60% 2603
>
>16 Gandalf 4.32h 256MB Athlon 1200 MHz 2650 34 -33 430 54% 2622
>19 Gandalf 5.0 256MB Athlon 1200 MHz 2638 49 -50 202 46% 2670
>20 Gandalf 5.1 256MB Athlon 1200 MHz 2636 25 -25 758 55% 2603
>
>PS: And you can also add that on the same hardware an Athlon 1200 MHz, it is as
>strong as Hiarcs 8.0 and Rebel Century 4.0
>
>Pichard
Conclusions:
1. Fritzes still the best:
Rating + - Games Won Oppo
1 Deep Fritz 7.0 256MB Athlon 1200 MHz 2754 31 -29 614 71% 2601
2 Fritz 7.0 256MB Athlon 1200 MHz 2739 30 -29 574 64% 2634
3 Chess Tiger 15.0 256MB Athlon 1200 MHz 2729 29 -27 660 66% 2614
4 Shredder 6.0 Pad UCI 256MB Athlon 1200 2725 24 -23 911 64% 2627
Highest rated Fritz has central estimate 2754, with an error range of +31 to
-29.
Chess Tiger has a central estimate of 2729 with an error range of +29 to -27
2754 - 29 = 2725 (bottom end of estimate of Deep Fritz 7's strength)
2729 + 29 = 2758 (high end of estimate of Chess Tiger's strength)
It could easily be true that Chess Tiger is stronger than Fritz (though slightly
less probable than the other way around)
Illustration:
I saw a bear that was between 7 feet 9 inches and 9 feet 2 inches in length.
I saw another bear that was between 7 feet 5 inches and 8 feet 11 inches in
length.
Will you conclude that the first bear was bigger than the second bear?
Now, as for Ruffian verses Gandalf:
11 Ruffian 1.0.1 256MB Athlon 1200 MHz 2676 39 -37 346 60% 2604
...
19 Gandalf 5.0 256MB Athlon 1200 MHz 2638 49 -50 202 46% 2670
2676 - 37 = 2639
2638 + 49 = 2687
It could easily be true that Gandalf is stronger than Ruffian (though slightly
less probable than the other way around)
The uncertainty in the measurements is clearly stated.
We can say (with almost dead on certainty) that Deep Fritz:
1 Deep Fritz 7.0 256MB Athlon 1200 MHz 2754 31 -29 614 71% 2601
2754 - 29 = 2725 (bottom end of estimate of Deep Fritz 7's strength)
is stronger than:
24 Gromit 3.11.9 256MB Athlon 1200 MHz 2620 55 -56 162 44% 2662
2620 + 55 = 2675 (upper end of estimate of Gromit's strength)
because 2725 is bigger than 2675. So here we saw a bear that was between 8 and
9 feet long and compared it with a bear that was between 6 and 7 feet long. The
first bear was bigger.
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.