Starting This Weekend- Another High-Powered Match!

[geots]

Could be Sunday, but most likely sometime Saturday. A new match begins with the current No.1 engine version- Stockfish 290613 64 SSE4.2 defending the title he just won against Komodo 5.1r2. This should be an interesting match-up, as both these engines prefer the longer time controls.

There will not be a match for No.1 every week- probably a couple or so times a year. This one will take place because this bug-free Komodo version was just released, and he shall get a chance to show us how he fares against this mighty strong Stockfish.

I have on order another 6-Core system exactly like my Alienware i7 that is being used for this match. After setup- any championship matches will be at longer FIDE controls, with either 4, 6, 8 or 12 Cores for each engine. But then there will be qualifying rounds that will take place first. An engine will have to earn his way for a shot at the gold ring. No rating will get you a title match- you will first have to prove to me you deserve the rating. And it may be that all championship matches are broadcast.

It should be interesting to see if this Komodo version can come thru where Houdini did not. Stay tuned....................................


Best,

[Jim Collins]

Sounds great George. Will you be posting the games here?

[gerold]

Sounds good George. How many game will be played.

I have quit doing the short 25-100 game matches. Now doing 1000 games plus matches. 5/3 tc. The authors i test for prefer the longer matches.

Best,
Gerold.

[geots]

Sounds great George. Will you be posting the games here?

Yes, for sure I will. Thanks.


Best,

george

[geots]

Sounds good George. How many game will be played.

I have quit doing the short 25-100 game matches. Now doing 1000 games plus matches. 5/3 tc. The authors i test for prefer the longer matches.

Best,
Gerold.

Thanks Gerold. I run a lot of blitz when I am beta testing for Don- Komodo, Tom- Djinn, Yuri- Strelka and Giancarlo- Equinox. I think there is a place for short and long controls. Think of it this way. When you see championship human, or engine matches, they play the longer controls. Then you see along with that when the main matches are over- they will often have a blitz tournament. But the interest in and the importance of the blitz matches is like more of a "just have some fun deal." Not that blitz is not important- it is a very important tool. Just depends on your POV as to how important. When you test for authors, they of course mostly go for blitz- because they want lots and lots of games.



All the best,

george

[geots]

I actually had a feeling that this Stockfish version- 290613, had a really good chance of beating Houdini 3. But as for this match- I just have no clue at all. 6 Cores each at 40/40 benched- we shall see. Notice that I have changed the number of games from 50 to 100. I would most likely bet the farm that the engine ahead after 50 games will also be ahead after 100- but I am really excited about this one, and I don't want to see it end quickly. This should really be some great chess, and we will see if Komodo can take the title from Stockfish.

The match began 15 minutes ago, and below are the controls.


Alienware AURORA_R4
Intel i7 w/6 True Cores
Fritz 11 gui
6 Cores each/64bit
256MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating- benchmarked to adapt to 40/40
Match=100 games



Updates as they come in-

[geots]

Just getting into the match- a sort of "feeling out" process goes on. This should be extremely interesting- almost makes me want to pass on sleep and just watch.



Alienware AURORA_R4
Intel i7 w/6 True Cores
Factory overclocked
Fritz 11 gui
6 Cores each/64bit
256MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating- benchmarked to adapt to 40/40
Match=100 games




Stockfish starts with the white pieces:


Game 1- Drawn

Game 2- Stockfish wins

Game 3- Drawn


The target date for my new Alienware AURORA R4 Intel i7 6-Core system to be delivered is on or before July 30. EXACTLY the same as the system being used here- except I was told the processor/s will be quite faster. Someone asked me why I did not wait on the Intel 8-Core system and buy it. Answer: I am waiting and I will buy it. With more cores to rig up and use the end of the month- I may as well go ahead and set up to broadcast certain tournaments soon after that. These games are just too good to not be available to others to watch.


Another update later-

[gerold]

Thanks George. This one will be very interesting.

Best,
Gerold.

[geots]

12 games in the books, with 88 remaining. Still too early to get a handle on which engine might have an advantage- especially considering that of these first 12 games, I probably slept thru 7 or so. But we shall see....................




Alienware AURORA_R4
Intel i7 w/6 True Cores
Factory overclocked
Fritz 11 gui
6 Cores each/64bit
256MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating- benchmarked to adapt to 40/40
Match=100 games

Code: Select all

Stockfish 290613 64 SSE4.2   +2/-2/=8
Komodo 5.1r2 64-bit          +2/-2/=8

More to follow, as they continue to battle-

[geots]

7 more games have been played- with some exciting action. So far I have found nothing that will give me a hint of things to come- they seem to be playing it close to the vest. As of yet, nothing much to separate them.

I will give you this advice. At least in chessbase guis- ALWAYS untick/uncheck the block for "Use Sleeping Threads" in Stockfish parameters before you begin. Marco advises you to do that, and if you don't and are using all cores available for each engine- "System Idle Process" as seen in task manager is going to use one-half to 1 core most of the time when Stockfish should be getting it. So left checked in the box- Stockfish very seldom will get to use the 6 full Cores as Houdini and Komodo do. Just something to keep in mind. Marco says it has been and will be left unchecked in all official releases.

And a big, huge thanks to Ray Banks for pointing out to me how to fix the problem!




Alienware AURORA_R4
Intel i7 w/6 True Cores
Factory overclocked
Fritz 11 gui
6 Cores each/64bit
256MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating- benchmarked to adapt to 40/40
Match=100 games

Code: Select all

Komodo 5.1r2 64-bit          +4/-3/=12   
Stockfish 290613 64 SSE4.2   +3/-4/=12

Barring any unexpected turn of events- this should be the last update for the night-