Friday, November 16, 2007

2007 Gothic Chess Computer Championship, Round 1, Game 2

Tornado vs. TSCP Gothic 64, 0-1

We get our first glimpse of the new Tornado program in this game, battling with the "speed demon", TSCP Gothic 64, the first 64-bit Gothic Chess program. We see an odd trend with the Tornado progam: 6 of its first 10 moves are with its Knights! It seems to have an aversion to prefer Knights over Bishops, but later on it does not hesitate to involve them in combinations leading to trading of some heavy pieces.



After each side has castled and black gets in 15...e5 things settle down a bit, as shown in the diagram. White's bad Bishop on the queenside is devoid of prospects, and the pawn structure soon dictates stagnation. The Tornado program makes the misinformed decision to overactivate its Rook with 20. Rg5, but then it deploys it where it can be harrassed mercilessly. TSCP Gothic then wins the exchange after a bit of "razzle dazzle": 27...Bxi4 28.Ng5 Bg2 29.Rxg2 -- not the typical Exchange methods employed in the chess world!

After some further simplication, each side seemed to grope for what was unattainable: a wrest for more material. Tornado was vastly outsearching TSCG Gothic at one point, reporting phenomenal search depths with a score showing no signs of danger.

This had me wondering: Why the unusual play so early in the game when this "powerhouse" search was obviously at its core?

My prediction is that there is some sort of bug in Tornado and we all may get stunned by this program if it can correct its unusual early round behavior. Meanwhile TSCP Gothic 64 with its 1,000,000 node per second move generator is not too shabby either!

In the end, Tornado got "sucked up" in its own overuse of time, and TSCP Gothic pulled out the win. I am curious to see how these programs will do in later rounds, as I am not sure which is the better of the two after this performance.

11 comments:

Martillo Program said...

So how fast was Tornado searching? And why did it run out of time with not even 50 moves played?

The Fitz Man said...

The play just seems too plastic in that game. It's like the programs don't know how to play or they're too chicken to try and risk anything.

DuelingBishops said...

I was thinking the same thing. These programs were playing a boring game. Contrast this agains the smirf and Vortex game and there's a huge difference!

I'd like to know what makes these programs play this way against each other. Against human players they seem to do much better. Are we that much worse than the programs?

Martillo Program said...

Due to technical problems, Martillo must withdraw from the tournament. Much apologies for now. I added some bug and I can't remove it!!!!!

AxeGrinder said...

I know it's still early in the tournament but from what I've seen this won't be that interesting of an event. The play of Vortex is high above the others, even S.M.I.R.F. now is far below. I watched the game where S.M.I.R.F. destroyed Vortex in the blog game and I applauded! It's about time someone kicked it's arrogant programmer's ass!

Knowing the egocentric Trice, he probably hired a team of consultants to take the thing apart bit by bit and redo it all just to appear respectable in the tournament. And that whole "We lost the sponsors now" crap, does anyone really buy that bunch of bull? S.M.I.R.F. beat you bad, so you did not want to risk paying out any more, end of story, loser!

As for the other programs, sorry guys, you're all weak. S.M.I.R.F. and Vortex will run the show. The rest of you will make your weak moves and the weakest of the weak will sink to the bottom.

That's my prediction anyways.

And I expect this post will be deleted by the Gothic Nazis that run the show, so enjoy it while you can.

geography_teacher said...

Axegrinder, your name is very appropriate, but most of what you said is misplaced anger based on conjecture and innuendo. There were sponsors involved but there was some form of "blowup" on the German discussion boards that were read by someone in the company. They pulled out because their company name was surrounded by inappropriate trash talk that would be able to be found in a google search. Guess what? I'd pull out my money if the same junk was going on and my company's name was linked by such ill tempered misfits!

Why don't you think Ed was able to debug his own code? Wouldn't that make more sense than what you suggest? To me, your post is ridiculous!

Vortex was fixed, it beat smirf, so what? It's a contest between programs and the better one usually comes out on top.

These other programmers are just getting started. Give them a chance to compete and learn for Christ's sake!

And leave your "bad ass" attititude somewhere else. It reads like you are just jealous of Ed. In fact, you sound an awful like those other two morons from chessvariants.org I forget their names for now.

Smirf said...

To AxGri: SMIRF's goal here was and is to come out as second best. And as usual for goals, that mainly is adressing the intended direction. Where it will be in the end still is open, and I also could live with a bad result, if that should happen.

To become able to write good (10x8) chess programs requires special people with a sufficient portion of self confidence. If you are misunderstandig this as an arrogant behaviour, you merely are outing some lack of understanding. I leave it to the readers to decide, which of our postings might be most arrogant ...

DuelingBishops said...

Axegrinder: Dude, get a life!

You know ZIP about what went on with the sponsor thing. You don't know jack about what went on with the Vortex programming team. Everything you wrote was guesswork at best, and at worse it was pure lies trying to make someone else look bad. What purpose does that serve?

If Ed worked on his his program and it's doing better now, who cares who he hired to fix it? It's so unlikely that strangers could fix something so specific as a chess variant program that a designer built. The designer lives or dies by their own code.

You sound like some angry person who's got other problems. Even your choice of posting name is suspect.

Cartaphilus said...

I've known Ed from "the net" for over 2 years now. If there's ever been a fighting, knock down, drag out rivally, it's been between the two of us. I throw everything at him every game we play. I try everything to distract him when we play, including debasing insults and claiming every move of his loses.

He just plays and offers little rebuking commentary though sometimes he makes funny remarks about my insults.

I don't think I've ever won a game but more than a few have been real close battles. He never showed "ego" by rubbing this in my face or telling others that I am like 0-8 against him.

I don't get you axegrinder. You obviously don't post here that much and when you do you're all piss and vinegar. Your claims about Ed just don't match up to the person I've played.

And if you think the other programs are weak, why don't you create a strong one and show us how it's done!

GothicChessInventor said...

Reinhard,

I think what Axe was saying was that he was glad SMIRF won against Vortex when it did, and he was referring to me when he was talking about the
"arrogant programmer."

It amazes me that someone with whom I have never interacted imagines ficticious dialogs to the point where a personality ascertation can be abstracted.

And about Vortex: There is no "team", and I have given abundant credit to those who helped me construct it.

Dr. Robert Hyatt published the Crafty 8x8 chess source code. My colleague, Gil Dodgen, shouldered the burden of "stealing" the Crafty code and porting the engine to an 80-bit mode. He gives credit to his brother Steven for coming up with a "trick" known as "operator overloading" to ease or transition by an order of magnitude.

Steve Dodgen also wrote the graphics library that I used to build the Gothic Vortex graphical user interface.

Later on, Mark Bourzutschky modified Eugene Nalimov's endgame database engine, and we both computed 5-piece tablebases for Gothic Chess.

Andrew Kadatch then assisted by making the run-time probing code that allows the tablebases to be used by Vortex.

These are not "full time" programmers. They did it for fun, on a part time basis, and all in a relatively short period of time.

The evaluation function, search engine tuning, and countless add on features have been written by me over the last several years.

Sometimes I would make "theoretical' changes to the program without sufficient testing. While, on a superficial level, these "should have" led to programming improvements, for the most part, they did not. For years Gothic Vortex 1.2 was the "king of the hill" when it came to running engine matches. No matter what I tried, I just couldn't create something that would beat it.

Then I starting making the board more ornate, adding other features to the program, and, somewhere along the way, I "broke" something. It was searching slower. It would produce occasional strange evaluations.

And then a hard drive crash denied me access to the source code for several months.

Enter Drive Savers -- they recovered it all, amazingly!

And this was very close to the start of the tournament.

I ran what I thought was a repaired Vortex against SMIRF, and it lost, big.

So, I basically "threw out" all of the King Safety, Development, and Search Extension changes I did, and I started over.

The result: The optimized version of Vortex that is killing programs left and right today.

No team. No hired contractors.

Just time, and patience.

I wrote a "log" mechanism that chronicles every variable used in the search, and I spit out the results for every position seen during a game.

Not every position searched, just every position at the root.

This has helped tremendously in the tuning of the new software.

Smirf said...

Ed, indeed sometimes it would help to go back to the roots. That kind of action I am intending to do with my SMIRF "zombie". Moreover, because my old computersystem was not working well since my son decided to use its big RAM and to replace it by some silly parts, I just have begun to rebuild that old Windows computer by integrating some new components. But technology has changed during the last years and thus it has been a little bit more desasterous than expected. Now luckily the machine is working again, but the process of installing all my tools still has to be completed. And I still have not decided, whether to have only old XP there or also XP-64 or Vista-32. You see, I am somehow busy after having my SMIRF finally released. This year I intended to use MAC only, but it seems, that a lot of people are working against my detaching from Microsoft OSs.