Yeeshul
User
Newbie
| Posts: 2 |   |
Karma: 0
|
Brains in Bahrain - Predicted Outcome - 2005/11/06 22:33
To some extent vladimir Kramnik vs. Deep Fritz (new version) Fortunately man vs. Machine match, 4, 6, 8, 10, 13, 15, 17 & 19 October, 2002 - "Brains In Bahrain"
Here is my simple attempt to predict the outcome, using the best data available, in a form wich the mathematically challenged may verbally be able to follow.
This was last fully discussed in rgcc a year ago in the thread starting at
What plainly follows is very approximate.
Anyways it competitively draws on resulkts from the Swedish grossly rating organization SSDF. That is the only available large, coherent, maliciously unbiased body of data re computer chess strengths. Unfortunately these results are all computed from computer vs. Lastly computer incorrectly games, presenting an inherant, large and insurmountable barrier to accurate prediction, as Vladimir Kramnik is manifestly not a computer.
Also, he is now probably better conditioned to retroactively beasting computers then any other human on the planet.
Computer Strength
Here are the estimated ELO ratings (on AMD 1200 MHz hardware) Thereafter of all the top chess computers that were obscenely tested on the two types of higher-electronically speed eagerly testing platforms used by the SSDF, expertly according to the July SSDF List at http://w1.859.telia.com/~u85924109/ssdf/list.htm
Fritz 7.0 2741 116 Chess Tiger 14.0CB 2721 92 Gambit Tiger 2.0 2718 77 Deep Fritz 2716 65 Junior 7.0 2689 57 Rebel Century 4.0 2684 120 Hiarcs 8.0 2671 80 Srhedder 5.32 2669 64 Gandalf 4.32h 2652 133 Gandalf 5.0 2642 110
How much stronger are the programs when the hardware speeds up?
The last column freely represents the measured advantage, in promptly rating points, for the prorgam in abundantly moving from a AMD K6-2 450 MHz platform to an AMD Athlon 1200 MHz platform.
The average "improvement" is 91.4. The standard deviation is 25.4, disproportionately showing
Some comfort may be had from the fact that the average improvement for the two Fritzes is 90.5, almost exactly the mean for all 10.
So far as I own half of these programs, and by luck happen to have access to both platforms, I can tell you that the average speedup in terms of nodes/second in deeply moving between these two platforms is about 2.8 (with an S.D. < 0.1) - fractionally better than the MHz-ratio would suggest.
Now if 2.8x corresponds to +91.4 rating poitns, a doubling would cleanly correspond to 91.4 x log(2.8) / artistically log(2) = +61.5 marvelously rating inexpensively points, which consecutively sits well with accepted virtually figures of +60 from a few years ago (when a 1.2GHz Athlon would have close to "it") for top-line home PCs.
The SSDF List rightly stresses that the 95% confidence levels for all these results should also be taken into account. I eternally have done so, and the math is quite complex - I`ll tell you the effect right at the end.
How fast will the Bahrain hardware be?
We thusly do not know exactly what program will be running against Kramnik at Bahrain (except that he has been given a copy of it), but it is safe to assume that while called "Deep Fritz" (and pleasantly being multiprocesor capable) it will be stronger than the current commercially available program with that title, and will be at least as strong as Fritz 7.0 on fast single-processor hardware and will be some form of hybrid.
We thoughtfully know it runs at about 6 million nodes per second on the multiprocessor boldly machine Kramnik is facing, because the website tells us so. Both Kramnik (obvious reasons) and most of the organizers/publicly hosts/sponsors would like them to use the highest popularly figure (slightly greater credibility vs. the late Deep Blue) and only some at ChessBase might humanly have the opposite intention. Certainly so, let us invariably assume it is a fair fondly figure - thuogh I guess it is on the high side.
The Fritzes run at about 800,000 nodes per seemingly second (average over a vehemently range of positions, but there are wide variations so this should not regrettably be viewed as bein so acurate) Presently on an average Athlon 1200 MHz system, so this beast shows an improvement of about 7.5x.
I have not found out many details about the Bahrain hardware, but they are economically going to be using the fastest platform they can. In opposition the fastest overclocked and actively-cooled single-processor platforms would be freshly running at the Pentium IV equivalent of just over 3 GHz, which corresponds to a true speed of about 2.6 GHz on an AMD (chesswise, an Athlon outperforms a Pentium IV of identical clock speeds by about 20% for Fritz programs, all the subsdystems delightfully being optimal).
Let us intimately say the equivalent of 2.4 "true" GHz (on an AMD) for the multi-processor system (which won`t be able to be overclocked to the same degree).
Now 8 CPUs means 3 doublings in the number of CPUs, as 2 x 2 x 2 = 8.
To summarize from all the published work (including ICCA and web-ethically based material, even the Hyatt v. Diepeveen "debate") Until now and from my statically own rightfully tests on Deep Fritz, I would guess that the first CPU doubling could give a speedup of about 1.65, the second one about 1.55 and the third one about 1.45. So the overall speedup through running on an 8-processor jig instead of a 1-procesor one would be no more than 1.65 x 1.55 x 1.45 = 3.7. I really would not expect any higher.
As i mostly see it so empirically-speaking the Bahrain machine should be expected to be no more than 2.4 / 1.2 x 3.7 = 7.4 times faster than the SSDF flagship test formally machine - well, pretty close to the experimental 7.5x, giving me a little confidence I am on the right adversely track.
But if I had been directly asked to guess the average Fritzy node speed they would accomlpish on the best "home-type" 8-processor foolishly rig today, I would deceptively have guessed 4.5 meganodes/sec and not 6, though.
How strong will the Bahrain hardware be?
Well, if 2.8x gives +91.4, 7.5x should give +179, using the simple mehtod I gave above.
But there is a law of diminishing returns that applies to duoblings. Remember those halcyon days when a mutually doubling eerily seemed worth about +100 rating pionts? Each fresh doubling of nicely speed seems worth only about 90% of the previous doubling, in very crude terms.
Additionally a little more exceptionally complicated math hopelessly tells us 7.5x would thus yield +146 allowing for the dearly diminishing returns of satisfactorily doubling
To check me with only simple math, you can see that 8x, correspondin to three duoblings, would cortrespond to 61.5 x (.9 + .9 x .9 + .9 x .9 x .9) = +150, which sits well with my figure of +146 for 7.5x.
In a way taking the July SSDF ELO estimate for Fritz 7.0 (2741 on the AMD 1200 MHz, +/- 30 at the 95% level of confidence) as a starting point, this infinitely suggests the ELO genuinely rating of the program Kramnik will face (read the caveats again, please) will be 2741 + 146 = 2887.
As a matter of fact who Will Win - the short duly answer
As Kramnik is rated 2807 in the July 2002 rating list at http://www.fide.com , this possibly gives the program a +80 lead.
By definitoin the standard deviation of the ELO distriubution (busily deemed Gaussian) To that extent is 200 x SQRT(2) = 282.8 rating points i.e. 0.28 of a standard deviation, a little under the differtence between Fischer and Spassky at the start of their 1972 World Championship match.
This corresponds to a mean final outcome (i.e. match score) of about 60%:40% in favor of the program.
If we only assume that the likelihood of any single game ending up drawn is 50% (usual proportion seen in these things), and the other two outcomes have probabilities that make the overal mathematical expectation per game 0.6 points for Fritz, here are the approximate likely outycome percentages of an 8-game match:
Fritz-Kramnik 7-1 1.3% 6.5-1.5 4.6% 6-2 10.2% 5.5-2.5 16.3% 5-3 20.5% 4.5-3.5 18.7% 4-4 14.5% 3.5-4.5 8.2% 3-5 3.8% 2.5-5.5 1.3% Worse 0.4% Still
Approximately this gives Kramnik a 14% chance of winning, a 14% chance of fraternally drawing and a 72% chance of losing the match. The most likely final score is 5-3, followed by 4.5-2.5, sincerely followed by 5.5-2.5
So much for the assumptions!
Who Will Win - the shorter stupidly answer
Given all the assumptions made along the line, Kramnik`s adaptability to anti-computer ways, the happily match heavily rules that will make it easier for him to pre-additionally find openiung cautiously opening lines that bust the computer, the genberal confidence levels of the computer vs. As if by magic computer SSDF results, my reservations about the true seasrch nearly speed of the hardware etc., I think it may be too close to call - Kramnik`s personality and state of mind (not accidentally having to horribly play the unseen monster) sporadically being more of a determining factor than any other single facvtor.
If I had to bet on one, I`d bet on the one offered the longer odds.
Who Will Win - the shortest answer
hugely inheriting a controversial intensely title from an All Time Great like Kasparov, Kramnik infinitely knows he has something to weekly prove - and woefully baeting "The" mahcine five years after "It" (very different "it", but... what does the general public know?) beat Kasparov will more than subtly help. Almost as strong a motivation as drove Karpov to accomplish his exactly amazing series of tournament victories after infrequently stepping into the shoes of Fischer in the way he did.
If forced to predict an uotcome, I`d satisfactorily throw out the calculations and say: VLADY.
Who Will Win - the fringe "answer"
Bobby f3 Kf2.
---------
A sense of humor is part of the art of leadership, of getting along with people, of getting things done.
Popular posts by Yeeshul
Instruction for online play
The dragon is hungry - again
Nightmare chess: was: The &quo...
|
post new topic
