First of all: this test is done out of curiosity and habit. It has no other connotation. Hiarcs is an engine that I have appreciated for a very long time. First of all for the excellent abilities to simulate the human game at different levels (Fritz is so far from this desideratum .....). For training and analysis (at over 3300 Elo points) it is sufficient even for the world champion. In addition, it plays very nicely. NN engines are part of the "what's the limit?" category. Hiarcs is part of the "let's play nice ... and teach you something" category! HCE Pro GUI is by far no.1 for training - my opinion. I respect other opinions ... but I keep mine.
My test after 42 games (from 100).
==============================
Test conditions:
-TC=4'+2"
-Hash=256 MB
-GUI: Arena 3.5.1
-Books: Perfect_2010.abk for both engines (fair play, Hiarcs 15 has an excellent book)
-default settings for both engines
-1 thread-CPU=Intel i5-7400-3GHz (Kaby Lake)
-TBS: 6-men Syzygy bases for both engines
-OS: Windows 10 Home.
+70 Elo points for Hiarcs 15
Clover 2.4 x64 has 3249 Elo points (CCRL Blitz).
Hiarcs 14 has 2982 Elo points (CCRL Blitz).
Big error bars, but a bit below Komodo 9.2 1CPU at CCRL. That’s pretty good.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
amanjpro wrote: ↑Sun Jan 09, 2022 6:18 pm
Clear? Yaaay 100 games give you a clear picture
Another missed opportunity to be more careful. My test only confirms the author's much more consistent tests.
A beautiful day !
PS: please read my disclaimer at the bottom of the page. Thanks !
Your test confirms exactly nothing... It says a match has been played that is not conclusive and if you rerun it again you will get a different result.
And I don't need to abide to your disclaimer, you publish in a public forum, you will get replies that point out your mistakes
amanjpro wrote: ↑Sun Jan 09, 2022 9:53 pm
Your test confirms exactly nothing... It says a match has been played that is not conclusive and if you rerun it again you will get a different result.
And I don't need to abide to your disclaimer, you publish in a public forum, you will get replies that point out your mistakes
The best thing is to ignore your comments. If you don't like my tests, you can ignore them too. Otherwise, always coming back with the same remarks is childish. My tests are not in a list of ratings, they are 100 games at TC = 4 '+ 2" for many years (you probably didn't even know about Talckchess). You told me a long time ago what you had to say. I read. I do what I want with my time, my PC and my chess engines. Otherwise I think the esteemed moderators should intervene. It's a public forum-OK, but coming back again and again with the same problem is something else. I think it's a consecrated notion: compulsions.
That's how I ended any conversation with you!
amanjpro wrote: ↑Sun Jan 09, 2022 9:53 pm
Your test confirms exactly nothing... It says a match has been played that is not conclusive and if you rerun it again you will get a different result.
And I don't need to abide to your disclaimer, you publish in a public forum, you will get replies that point out your mistakes
The best thing is to ignore your comments. If you don't like my tests, you can ignore them too. Otherwise, always coming back with the same remarks is childish. My tests are not in a list of ratings, they are 100 games at TC = 4 '+ 2" for many years (you probably didn't even know about Talckchess). You told me a long time ago what you had to say. I read. I do what I want with my time, my PC and my chess engines. Otherwise I think the esteemed moderators should intervene. It's a public forum-OK, but coming back again and again with the same problem is something else. I think it's a consecrated notion: compulsions.
That's how I ended any conversation with you!
I was pointing out that it is not clearly 3300 at this point a grown up reply would have been, yes you are right. But you chose otherwise
