geots wrote:I call it the "Super Match"- being played for No.2 in the world: Komodo CCT 64-bit vs. Strelka 5.6 x64. It is being played at 40/40 repeating, benchmarked to CCRL and CEGT controls, except that CEGT no longer uses 40/40.
For anyone who thinks 50 games is not enough, let's take every trophy and title away from all the WCCC world champions in history- because none of them in winning played close to 50 games all total.
Next to last- Critter is not in this struggle for the No.2 spot. Simply put, IMO and in my testing- it is not strong enough to fight for No.2
And lastly, 8 games have been played. Updates will come every 10 games.
Best,
I don't have Strelka 5.6 to test. Good luck and hoping to get Strelka 5.6 to test as it is one of my favorite engines.
Have you run a test Critter 1.6 vs. Komodo cct. I think Critter is much stronger than the cct.
Best,
Gerold.
Critter stronger than CCT? Not in a million years- trust me.
Best,
On two other testers lists it shows CCT below Critter and even SF and others. Its about number 14 on one main testers list. Its not in my top ten. I just test the top ten engines most of the time.
Best,
Gerold.
P.S. Keep up the good work.
That is fairly cut and dried. They are wrong. Now the tester who has it as No. 14 needs to find something else to occupy his time.
BTW, take a look at CCRL's latest 1CPU list and see how far down you have to go to find Stockfish and Critter!
geots wrote:I call it the "Super Match"- being played for No.2 in the world: Komodo CCT 64-bit vs. Strelka 5.6 x64. It is being played at 40/40 repeating, benchmarked to CCRL and CEGT controls, except that CEGT no longer uses 40/40.
For anyone who thinks 50 games is not enough, let's take every trophy and title away from all the WCCC world champions in history- because none of them in winning played close to 50 games all total.
Next to last- Critter is not in this struggle for the No.2 spot. Simply put, IMO and in my testing- it is not strong enough to fight for No.2
And lastly, 8 games have been played. Updates will come every 10 games.
Best,
I don't have Strelka 5.6 to test. Good luck and hoping to get Strelka 5.6 to test as it is one of my favorite engines.
Have you run a test Critter 1.6 vs. Komodo cct. I think Critter is much stronger than the cct.
Best,
Gerold.
Critter stronger than CCT? Not in a million years- trust me.
Best,
On two other testers lists it shows CCT below Critter and even SF and others. Its about number 14 on one main testers list. Its not in my top ten. I just test the top ten engines most of the time.
Best,
Gerold.
P.S. Keep up the good work.
That is fairly cut and dried. They are wrong. Now the tester who has it as No. 14 needs to find something else to occupy his time.
BTW, take a look at CCRL's latest 1CPU list and see how far down you have to go to find Stockfish and Critter!
Best to you as well,
george
Maybe it depends on under what conditions the engines are being tested. And what book etc. , time, etc. Seems my testing of engines going head to head may differ a bit from some of the other lists that are testing different engines at the same time.
One thing i did note on the CCRL is Rybka vs. SF 3 is equal in elo now. That is the engines i am testing now. My results show they are the same elo also. Good to see SF improving. Houdini will have to have an upgrade soon.
I see the Ivanhoe group still has a time bug of some kind when trying to test at very fast tc.
Last edited by gerold on Sun May 26, 2013 10:40 pm, edited 1 time in total.
I tried to make it easy, especially for the "whatever you call him" who has CCT ranked No. 14. If you know him- suggest he find another hobby to occupy his time.
Obviously the reason for running this match for Number 2 in the world- CCRL does not have Strelka 5.6 to test. So their rankings don't reflect what Strelka 5.6 "might and might not be able to do"'
Lastly, any tester who is way off base here: if his excuse is "he was including multi-core testing" and thinks anyone buys that knowing neither have an MP version refined yet to release as "official"- he is as delusional as his testing is.
Lastly, just to add that there is a Strelka 5.7MP that I will test when I can get to it.
The match is being run on my new Alienware AURORA R4 Intel i7 with 6 true cores. As Ray Banks said- "blindingly fast". But not the point. Important is that I am running it as 40/40 repeating. And I have it benchmarked to 40/19 repeating using the same benchmark CCRL uses. So there is no "different hardware" issue.
Actually it benches to 40/17 or 40/18- but after a lengthy discussion with my mentor (yes, even the best have one) Ray Banks, we decided because of the turbo-boost issue- 40/19 would be best.
Strelka 5.6 cannot shake Komodo CCT- and on the flip side Komodo has not been able to shake Strelka. Each time one of these engines gets a lead and it appears he may be about ready to run away with it- here back comes the other. So here we are after 92 games:
Alienware AURORA_R4
Intel i7 w/6 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating (Benched to adapt to 40/40) Match=200 games
If I am correct, 9 more games are in the books, and we have just passed the halfway mark. I think if they could talk, they would both be thinking what does it take to shed myself of this "sob". This match could in effect be over in 30 or 40 more games, and it might go right down to the wire. About the only thing I have been correct on is that the leader after 50 games is also the leader after 100 games. At any rate............
Alienware AURORA_R4
Intel i7 w/6 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating (Benched to adapt to 40/40) Match=200 games
It seems this 10 game update could be called something similar to a mirror image of the previous 10-game update- as Komodo ground out another 1 point advantage. Make no mistake- with 79 games remaining, Strelka has no reason to push the panic button. OTOH, in each of the last 3 updates, Strelka has lost ground in each of them. At some point he will have to find a way to reverse the trend. Or...............................
Alienware AURORA_R4
Intel i7 w/6 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating (Benched to adapt to 40/40) Match=200 games
At the 135 game mark- the one thing Strelka could not afford to have happen- happened. Komodo CCT increased his lead by 3 MORE games. Now, with 65 games to play, Strelka is- what I would describe as- in a lot of trouble. Against an engine like Komodo CCT, it's bad enough to play from even. But when you are in a position where drawn games are not acceptable- what are you gonna do. (rhetorical)
Alienware AURORA_R4 Intel i7 w/6 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating (Benched to adapt to 40/40) Match=200 games
9 more games played, which puts the total at 144. Actually, in this last 9-game set, Strelka 5.6 knocked 2 elo points off Komodo CCT's lead. If there is any good news for him- that would be it. Unfortunately, the lead still remains at 8 games. And instead of 65 games left to make them up in- now it has dropped to 56. Strelka is swiftly running out of time as Komodo methodically holds to his lead. One hell of a good match!
Alienware AURORA_R4
Intel i7 w/6 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
40/19 Repeating (Benched to adapt to 40/40) Match=200 games
