Depth 1 - Engine Tournament

Since I'm on holiday now, I have some free time. So I've decided to conduct an engine tournament.

Instead of using those boring standard time controls in which a single game takes ages to finish, I've decided to let each engine think for just one half move (depth 1).

The tournament will be in Round Robin format. We can even calculate their elo ratings after enough number of games have been played.

The engines that are participating in this tournament are -

• Komodo 10
• Stockfish 7
• Houdini 4 Pro
• Gull 3
• Deep Rybka 4.1
• Critter 1.6
• Strelka 5.7
• Fire 4

I'm willing to add up to 20 engines to the tournament. So if you guys want to see any other engine participating, feel free to let me know.

All games will be played on Arena 3.5; RAM - 8GB; Intel i5 Processor.

The results of the first round is out. There are a few surprises -

Rybka is definetely a force to reckon with. Its amazing to see how the top engines fail to deliver at depth 1.

Round 2 in progress.

You ran all of those games in 30 minutes? Yes, I understand what "depth 1" means.

In my opinion, a "Depth 1" engine tournament is totally meaningless, due to the different ways an engine can handle things like search extensions. Plus, there's also the possibility that Rybka could be lying about the actual ply depth reached.

Better to make a tournament based on short time controls. Just my two cents.

The one good thing about Depth 1 tournaments is how fast they get, you simply cannot follow the games because they are lightning fast. Anyway round results are out too.

So far, Rybka is demolishing the competition.

Yes, that's how fast it is.

This tournaments aims to judge how well the engines play at depth 1, It doesn't matter how they handle search extension - it's their problem.  

And what do you mean Rybka is lying about the actual ply depth reached?

You're right EscherehcsE, After closely analysing the analyses of Rybka, it turns out that it is actually operating at depth 2 - 3

Rybka has been DISQUALIFIED from the tournament.

After a closer look it seems like Strelka and Critter are also using more than 1 depth.

Strelka and Critter have been DISQUALIFIED.

EDIT: OK, you replied before I posted, heh.

There have been discussions about why Rybka takes more time to reach higher depth than other engines, even though its strength was comparable to them. If it really was slower to reach the same depth, it would be much weaker.

So people believe that Rybka misreports the depth reached or simply that it works in a different way and that the "depth" means something different.

If I were you I would run a tournament where all engines use 2 threads, the same memory (about 256mb), ponder off of course and exactly 8 seconds for each move. Not sure about books or egtb, probably without those (?)

I want to see whether Komodo 10 or Stockfish 7 would win. 

Only 5 engines remain - 

• Komodo 10
• Stockfish 7
• Gull 3
• Houdini 4 Pro
• Fire 4

I'm happy to inform that these engines are playing a fair game by not operating at depth 1 only.

I can do that, but those games would take too much time.

If you multiply the number of games with the average moves and multiply with the seconds, how much time will it take? Maybe set it to 4 seconds? Or take out one of the weaker engines.

I'll give it a try, let us first get the results of the depth 1 tournament. I'm going to run 20 rounds.

Okay the results after 20 rounds are out -

EloSTAT rating estimation -

This idea actually did make me laugh out loud, so I applaud that. 

Why? So I can laugh too :)

Can you reccomend an interface that allows me to control all that? Something other than Arena.