Your favorite engines?

[AdminX]

I myself really like Dragon Dev.

Course you would, you train their nets for them.

The biggest issue with Dragon is that it costs over a hundred dollars to get one. I personally do not like commercial engines very much (i.e. Fritz, Ethereal, Revenge, etc...). I will most likely wait until Dragon 3 gets released and Dragon 1 is made free like older Komodo versions, before getting a copy of Dragon.

So you only like them when they are made free. Until then, you don't like them. SMH

[matejst]

So you only like them when they are made free. Until then, you don't like them. SMH

Somehow... Me too. I prefer when they are made free.

Kidding aside, I do not understand why the K-developers do not bundle Komodo with a GUI. It would be relatively doable to adapt Chess-x, e.g., and offer it with Komodo. I am simply confused by the prices: on one hand KD2 + Fritz GUI, plus database, plus Opening book - 100 euros, KD2 + subscription ~ 100 euros. Is the subscription worth 60$? But what do I know...

I cannot use Smarthink afther the 1.90 version -- did not try too hard, though. I probably lack some dll on my computer. I also liked a lot of old engines ...

http://www.sdchess.ru/download_engines.htm
The very 2nd line is Smarthink. There is a link to some dll among those links, but the site is quite old, I don't know if all the links are functional. I don't have any dll directly in the directory, so if it needs it, I probably have it installed in Windows 10. I have, among other things, Microsoft Visual C++ for x64, from Microsoft, installed. It used to be free.

Thanks, Pavel.

[wheelman]

My favorite engines are HIARCS (great sparring partner and human-like playing style), Fritz 13 (no good reason; first engine I bought; last engine by Frans Morsch), Lc0 (both CPU and GPU versions; fascinating, original, alien-like playing style), Maia chess (great for training against 1800 and 1900 ELO level), and Stockfish hybrid and pure NNUE.

[matejst]

My favorite engines are HIARCS (great sparring partner and human-like playing style), Fritz 13 (no good reason; first engine I bought; last engine by Frans Morsch), Lc0 (both CPU and GPU versions; fascinating, original, alien-like playing style), Maia chess (great for training against 1800 and 1900 ELO level), and Stockfish hybrid and pure NNUE.

You must be a strong player, Mohammad, since Maia 1900 is quite strong -- at ltc (I tried at 40 mn + 30 sec.), probably around 2200 Elo. Maia 1400 seems to be -- well, 1400 (I managed to beat it blindfold), Maia 1500 is already way stronger. It seems it is very hard to reproduce how humans play. Maia 1900 is positionally too good, and tactically seems sound enough -- it's certainly better than I am.

[wheelman]

My favorite engines are HIARCS (great sparring partner and human-like playing style), Fritz 13 (no good reason; first engine I bought; last engine by Frans Morsch), Lc0 (both CPU and GPU versions; fascinating, original, alien-like playing style), Maia chess (great for training against 1800 and 1900 ELO level), and Stockfish hybrid and pure NNUE.

You must be a strong player, Mohammad, since Maia 1900 is quite strong -- at ltc (I tried at 40 mn + 30 sec.), probably around 2200 Elo. Maia 1400 seems to be -- well, 1400 (I managed to beat it blindfold), Maia 1500 is already way stronger. It seems it is very hard to reproduce how humans play. Maia 1900 is positionally too good, and tactically seems sound enough -- it's certainly better than I am.

I wish I was a strong player, Boban. I just like training specifically against 1800 ELO level; I would be happy to be around that level myself.

[matejst]

My variation of Marquez' Love in the Time of Cholera seems to be Chess in the Time of Corona. Writing and playing chess in these last months made me a lot of good. But, thanks to the spare time a generous university work and old days brought, I manage not only to improve my chess -- in a few months, I hope I will be close to the level I had 25 years ago, when I was playing actively -- but to use and test some new engines. So, now, my top looks like this:

1-2. Wasp 5.00 -- useful settings (e.g. selectivity), very realistic evaluation from a human POV, can be weakened, etc. Tactically very interesting -- very often proposes sacrifices, at least for a few lines. Its stats at Frank's FCP shows that it should be weak in endgames -- a lot of draws, fewer wins and losses than similar engines, but so far I don't know.
1-2. Zahak 9 -- same.
3. Marvin 5.2 -- a bit weaker, and does not have UCI_limit_strength nor levels. Otherwise, multiPV, solid evaluation (noticeably different from Wasp and Zahak). On a good PC, using several engines at the same time make sense, and I use it as a second "voice".
4. Orion 0.8 -- I guess it has the best evaluation, and it is tactically very strong. Lacks all the other things, unfortunately.
5. Tried Seer last night. Passed my "positional" tests, and I like his lines, but the evaluation is way too optimistic. Unfortunately, still no multiPV, etc., something I dare hoping Connor will implement soon. I like it so far, but I want to test it in simple positions and endgame, where I expect it to shine.
6. SCB 2,8 -- The strongest engine on my list, and the one I trust the most in late middlegames. Another engine that is not too fast. I like its GUI, and I would gladly use it with other engines too.

A few years ago, Ed Schröder implemented a way to see the EOC database. A few engines have developed a learning function, but there is no way for a user to see what they have learned, analyzed, etc. A GUI streamlined for playing, analyzing etc. could have a "learning" feature, just sayin'...

Then, I will repeat what I already wrote: an engine with a good static evaluation and a slower search is more useful for learning chess than the search monsters at the top of the rating lists.

[Frank Quisinsky]

Hi,

A bad start for Wasp (fast won games) in FCP Tourney-2022.
After 3.200 gaems for Wasp 4.65 the stats to fast won games are clearly better, just fantastic!!
After Wasp 4.65 ... versions 4.66, 5.00!

King safty is fantastic all the times.

I think a bit random after 500 games and I am very sure Wasp will go in TOP-14 with fast-won-games.

The higher draw quote comes from the endgames.
Here Wasp can make more often a draw in positions older versions lost.

And ... engines in die middle of the field have a higher draw quote.

So, I am very sure Wasp hold the nice playing-style with "Neural-Network".



Best
Frank

[purechess]

For playing against engines:

-CT800
-Honey 14 (Stockfish variant) - great sparring partner with many features to weaken the engine
-Lc0 with maia weights or TinyGyal weights

[matejst]

For playing against engines:

-CT800
-Honey 14 (Stockfish variant) - great sparring partner with many features to weaken the engine
-Lc0 with maia weights or TinyGyal weights

I did not try TinyGyal, but I often play against Maia. In general, from Maia 1500 up I find the engine excellent for training: it is positionally sound, but makes a lot of tactical errors. One can learn his openings, and elevate his tactical awareness against it. It is not really a human opponent, but in many aspects it is better, more useful. Maia 1900 is a tough nut to crack for me at this moment, when I still make too many blunders, it simply plays better than me.

I did not try Michael Byrne's Honey, unfortunately, but I have downloaded it already, so I guess I could have a look.

[towforce]

1. CS-Tal

Just the most fun it's possible to have playing chess!


2. Chess.com bots

More than good enough for my level of play. I expect they'll be rewritten as webassembly apps in the near future, and then they'll become even stronger.


Mini analyser in the PGN app in this site and at the ProDeo website.

Most of the games I look at are in these two sites, and it quickly shows me what happens to the "obvious" move I thought should have been played. I actually think it could well compete with many of the 8-bit dedicated chess computers (which most chess computers were well into the 1990s).