MCGE Classic London Book v Champion Book

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
herO
Member
Posts: 342
Joined: Wed Nov 12, 2014 8:49 am

Post by herO »

Yarc wrote:I do know that it is calibrated to the speed of the PC it's running on but that is all.
This statement is not accurate. Hiarcs weaker ELO opponents are not calibrated only to the speed of the PC, Hiarcs has "smarter" calibration. I also made few tests with Hiarcs weaker opponents. If you set Hiarcs eg. to 2000 ELO, you will get the SAME result on 2 core old i5 CPU as on 16 core AMD Threadripper.
User avatar
Yarc
Senior Member
Posts: 1051
Joined: Sat Apr 11, 2009 8:13 am
Location: United Kingdom

Post by Yarc »

herO wrote:
Yarc wrote:I do know that it is calibrated to the speed of the PC it's running on but that is all.
This statement is not accurate. Hiarcs weaker ELO opponents are not calibrated only to the speed of the PC, Hiarcs has "smarter" calibration. I also made few tests with Hiarcs weaker opponents. If you set Hiarcs eg. to 2000 ELO, you will get the SAME result on 2 core old i5 CPU as on 16 core AMD Threadripper.
I meant that Hiarcs was calibrated to the speed of the PC but that is all I know.
This means that it should give consistent results on any machine. What I do not know is how accurately it simulate a particular ELO setting.
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

Yarc wrote: Keep dedicated for dedicated use!
Amen Regards
Steve
herO
Member
Posts: 342
Joined: Wed Nov 12, 2014 8:49 am

Post by herO »

Yarc wrote:
herO wrote:
Yarc wrote:I do know that it is calibrated to the speed of the PC it's running on but that is all.
This statement is not accurate. Hiarcs weaker ELO opponents are not calibrated only to the speed of the PC, Hiarcs has "smarter" calibration. I also made few tests with Hiarcs weaker opponents. If you set Hiarcs eg. to 2000 ELO, you will get the SAME result on 2 core old i5 CPU as on 16 core AMD Threadripper.
I meant that Hiarcs was calibrated to the speed of the PC but that is all I know.
This means that it should give consistent results on any machine. What I do not know is how accurately it simulate a particular ELO setting.
I played many games against Hiarcs and I have to say that elo opponents from 1000 - 2000 ELO are quite accurate. Honestly I do not know how accurate is 2400 ELO, because I do not play on IM level. In my opinion, Hiarcs will play tactically stronger than IM but weaker positionaly.
User avatar
Yarc
Senior Member
Posts: 1051
Joined: Sat Apr 11, 2009 8:13 am
Location: United Kingdom

Post by Yarc »

herO wrote: I played many games against Hiarcs and I have to say that elo opponents from 1000 - 2000 ELO are quite accurate. Honestly I do not know how accurate is 2400 ELO, because I do not play on IM level. In my opinion, Hiarcs will play tactically stronger than IM but weaker positionaly.
Thanks, but I was hoping for an accurate rating to calculate ELO rating differences (if any!) between using the two opening books of the Millennium ChessGenius Exclusive. If I forgo calculating the ELO then I could use PC Hiarcs at a lower ELO rating and make a comparison between the MCGE's books.

@Steve: I am wondering if calculating the ELO difference between the two books is a futile effort. I have started asking myself the following:

Is the peformance of other dedicated machines also affected by coming out of book sooner? In other words, do they perform better when the opening is short? I can imagine this being possible as they would then be playing in their own style sooner, unless the book opening happens to be in keeping with their style. How many dedicated machines truly understand the main idea of these book openings anyway?

Maybe if I am witnessing an actual difference between using the MCGE's big Champion book and the much smaller Classic book, this is something that would be common to most if not all dedicated machines if they had a choice of book sizes? My only machines with this option are the MCGE and MCGP. I don't see much point pitting these two together, it's the same engine!

I could try a two matches with and without opening books. Not sure how varied the games would be without books. Perhaps a better option would be using a test suite of openings, but only 4 or 5 moves into each opening?

Opening quandary Regards
Ray
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

Yarc wrote:
I could try a two matches with and without opening books.

well the interesting idea from my perspective was to see if the opening books in the MCGE make a difference in rating ...so a "no opening books" match would not address that issue

even though you have early results that show a big difference in the rating my guess is that at the end of the day..there will probably be no significant difference ..as we saw with the MCGP

So just pursue whatever interests you and enjoy yourself

Have Fun Regards
Steve
User avatar
Yarc
Senior Member
Posts: 1051
Joined: Sat Apr 11, 2009 8:13 am
Location: United Kingdom

Post by Yarc »

Steve B wrote: well the interesting idea from my perspective was to see if the opening books in the MCGE make a difference in rating ...so a "no opening books" match would not address that issue
Yes, and that is also my interest. However, I was begining to wonder that if I am seeing a difference in rating between the big and small book that playing against another dedicated with a small book may effectively make the MCGE's big book obsolete. They will both tend to be out of book sooner. In this way, maybe your next statement is correct.
Steve B wrote: even though you have early results that show a big difference in the rating my guess is that at the end of the day..there will probably be no significant difference ..as we saw with the MCGP
The Stardiamond has a reasonably large book and so both machines will tend to be in book longer.

I need to refresh my memory of the MCGP games. From my limited testing it was better with the London book, but there are no doubt plenty of games to show little or no difference with the books.

The idea with the no book vs big book was to test the concept of is less better? However, I agree that this should be another test as it could cloud the issue.
Steve B wrote: So just pursue whatever interests you and enjoy yourself
Ah, definitely doing that! The MCGE is proving to be a very nice machine and so much stronger than my other machines. I tried a match against the Advanced Talking Chess Academy which is rated 1900 ELO from selective search. I set the MCGE to run at 50Mhz but even slowed down 6 times, it was a walk in the park for it against the talking machine. I have only played two games, but the Chess Academy has not been putting up enough fight for my liking. This is too one sided, and so maybe I could try turning ponder off on the MCGE, but even that may not be enough!

I need more strong machines in my collection Regards
Ray
Post Reply