Hydra vs. Todays Top

[Werewolf]

I recall that around 2007 Hydra was still playing in some online events or games (I forget details), and the conclusion was that somewhere around that time Rybka passed Hydra in strength.

Yes.

Then years later history repeat itself as the Rybka Cluster appeared, and then the conclusion was that somewhere later Houdini (in regular hardware) passed the Rybka Cluster in strength.

It didn't really though. The point is that on equal hardware Houdini 3 was crushing Rybka 4.1 and the Rybka Cluster on double digit cores. But unlike Houdini which practically topped out at 12 cores (the best available at the time IIRC), the Rybka Cluster could go up and up. I tested a version on 188 cores and it was better than Houdini 3 on my 12 core IIRC.

[Ovyron]

It didn't really though. The point is that on equal hardware Houdini 3 was crushing Rybka 4.1 and the Rybka Cluster on double digit cores. But unlike Houdini which practically topped out at 12 cores (the best available at the time IIRC), the Rybka Cluster could go up and up. I tested a version on 188 cores and it was better than Houdini 3 on my 12 core IIRC.

I now wonder which one was the first Stockfish version that on a 12core was stronger than the Rybka Cluster on 188 ones. Moore's Law stood and passed the test of time in computer chess, but it did so on the software size.

[Werewolf]

I've no idea.

I'm pretty sure SF15 today on only one core would be stronger than the 188 core Rybka Cluster though. (With the possible exception of tactical test solve times because of the cluster's broad search - but in actual play I bet 1 core SF15 would win)

[swami]

Would love to see a match with any of todays top engines (Dragon, LC0, or Stockfish) versus Hydra. Do you think it will ever come out and play again? The 2005 version was strong, wonder what a current version would do.

I was there during the Hydra - Adams match. I spoke to Chrilly about Hydra and had emailed him a few times beforehand. He very kindly let me test it on my tactical testsuite. I have numerous positions where Hydra had a go. It was much stronger than anything at the time.

Immediately after the match with Adams, Chrilly told me privately he thought Hydra was 3100 elo and that no human stood a chance.
Let's assume he's correct about the elo: Stockfish today kills it.

Tactically as well, SF kills it. But what you want to know is what would happen if we used modern FPGA cards. Unfortunately, I think we have our answer with SF vs SF at 10:1 time ratio...

That's about the strength of Zappa/Naum. Today's top engines are far ahead, even with hardware handicap they could outperform.