I Am Running Strelka 5.6 tests- Promise Accurate Results!

[geots]

Final results are in and am I glad. My eyes are glazing over from watching too much of these games. Remembering that blitz is where Houdini 2.0c and Critter are supposed to have the big advantage over Strelka.




Inspiron 620 Intel i5-4 True Cores
Fritz 11 gui/Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
4'+2"
Match=200 games

Code: Select all

SP8-Core-i5, Blitz:4'+2"  2/16/2013  

                         
Strelka 5.6 x64       +12    +32/-25/=143   51.75%  103.5/200
Critter 1.6a 64-bit   -12    +25/-32/=143   48.25%   96.5/200

Code: Select all

SP8-Core-i5, Blitz:4'+2"  2/16/2013  

                      
Houdini 2.0c x64   +19    +48/-37/=115   52.75%   105.5/200
Strelka 5.6 x64    -19    +37/-48/=115   47.25%    94.5/200

As I know there are programmers who will attest to this- chasing Houdini can be depressing to say the least. You work your ass off, and then you wake up and the version you have been chasing is now last year's wasp nest- and there is a new release with a whole nuther set of problems to deal with.

Still a few more blitz matches to run, but it has to be done judiciously. A waste of good processor time to run matches ag. engines that Strelka 5.5 just completely tore apart. Plus, I have never been a big fan of that "rub it in your face" attitude. A few matches will be crossed out before going to a longer control- Strelka 5.6 doesn't have a lot left to prove at blitz.




Enjoy-

george

[geots]

4 new 100-game matches have begun, all at 4'+2". They are against:

1.Bouquet 1.5 pop64 (picked over 1.6 because it is rated 20 points higher)

2.IvanHoe 46ei x64 SSE4.2 (Very similar to 50kQ, which played much better against Strelka than 46h did. Fairly new- also compiled by Quocvuong)

3.DeepSaros 3.3 popcnt x64 (exact same version as 3.3b!)

4.Gull II x64 (engine that has played well enough at times to deserve a chance)



Best I can say is we shall see-

george

[geots]

At least I know the IvanHoes like the back of my hand. The 50kQ (new) and its new partner 46ei SSE4.2 can certainly hang in there. Except for 46ei, the other 3 matches below were like a Sherman tank hitting a Volkswagen head-on. Thing is, those that lost have nothing to be ashamed of. They are all 3 A+ engines. The only caveat concerning the 2 IvanHoes above is they have yet to prove themselves at longer controls. And they very well may be able to. It remains to be seen.

Possibly a few more blitz- and then longer controls- where Strelka according to CCRL does its best. I have an ULTRA-Private engine being sent to me- the only request is that it be tested at longer controls. Only thing I can pass on to you concerning it is that there FOR SURE is not a private engine out there that could even stay close to it. Of course the other private ones couldn't carry Strelka's shoelaces, either. Anyway, the matches:







XPS 8500 Intel i5-4 True Cores
Fritz 11 gui/Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
4'+2"
Match=100 games

Code: Select all

XPS-Core-i5, Blitz:4'+2"  2/19/2013  

                             
Strelka 5.6 x64            +70    +31/-11/=58   60.00%   60.0/100
DeepSaros 3.3 popcnt x64   -70    +11/-31/=58   40.00%   40.0/100

Code: Select all

XPS-Core-i5, Blitz:4'+2"  2/19/2013  

                     
Strelka 5.6 x64   +111    +39/-8/=53   65.50%   65.5/100
Gull II x64       -111    +8/-39/=53   34.50%   34.5/100

Inspiron 620 Intel i5-4 True Cores
Fritz 11 gui/Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
4'+2"
Match=100 games

Code: Select all

SP8-Core-i5, Blitz:4'+2"  2/19/2013  

                        
Strelka 5.6 x64      +108    +41/-11/=48   65.00%   65.0/100
Bouquet _1.5 pop64   -108    +11/-41/=48   35.00%   35.0/100

Then the "tough as a boot" IvanHoe-

Code: Select all

SP8-Core-i5, Blitz:4'+2"  2/19/2013  

                             
Strelka 5.6 x64           +31    +26/-17/=57   54.50%   54.5/100
IvanHoe 46ei x64 SSE4.2   -31    +17/-26/=57   45.50%   45.5/100

On a side-note concerning the IvanHoe match above- when I went to bed, in wins and losses only- IvanHoe led Strelka 4-0!

And I still knew its chances of winning were somewhere between zero and non-existant.




Oh well, more matches to begin-

george

[geots]

Strange- all 3 of these engines did a bit better than I expected. And not just that- the results were for all practical purposes about the same for each match.

After posting these results- the match I alluded to will begin: Strelka 5.6 x64 v the Ultra-Strong 64-bit private engine I promised. This match will be run at a control of 5'+3", with 1000 games to be played. This ought to be one hell of a battle! (The extra long match was Yuri's choice).





XPS 8500 Intel i5-4 True Cores
Fritz 11 gui/Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
4'+2"
Match=100 games

Code: Select all

XPS-Core-i5, Blitz:4'+2"  2/20/2013    

                               
Strelka 5.6 x64              +53    +28/-13/=59   57.50%   57.5/100
RobboLito 0.21Q x64 SSE4.2   -53    +13/-28/=59   42.50%   42.5/100

Inspiron 620 Intel i5-4 True Cores
Fritz 11 gui/Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
4'+2"
Match=100 games

Code: Select all

SP8-Core-i5, Blitz:4'+2"  2/20/2013      

                             
Strelka 5.6 x64           +60    +29/-12/=59   58.50%   58.5/100
Vitruvius 1.19 pcnt x64   -60    +12/-29/=59   41.50%   41.5/100

Code: Select all

SP8-Core-i5, Blitz:4'+2"  2/20/2013  

                            
Strelka 5.6 x64          +53    +25/-10/=65   57.50%   57.5/100
IvanHoe 46e x64 SSE4.2   -53    +10/-25/=65   42.50%   42.5/100

After the 1000 game match gets going well, I will post regular updates. Ought to be a good one.



Nite/ or day-

george

[geots]

I am just passing on a message- past that I have nothing to do with it. The author of "UltraPrivate"- playing now in the 1000 game match against Strelka 5.6 has asked me to post this challenge:

"To any private engine listed in Wiki- I challenge you to a 200 game match at 40/40 repeating, in which you will be given the handicap of 'pawn and move.'
Anyone feel lucky?"



Only a message carrier-

gts

[geots]

Initial results are in for the first look at this "strongest ever" private engine. As we speak, he is making some last minute modifications that he says will add 25 or 30 elo to the engine. The agreement was to switch them out at that point in time. (Last minute changes = 30 elo? I am skeptical) At any rate:




XPS 8500 Intel i5-4 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
5'+3"
Match=1000 games

Code: Select all

XPS-Core-i5, Blitz:5'+3"  2/22/2013


Strelka 5.6 x64   +21    +17/-13/=35   53.00%   34.5/65   
d/tella ve0 x64   -21    +13/-17/=35   47.00%   30.5/65

Hmmm- better than I thought,

george

[geots]

We have 80 more games in the books, and this match continues to be extremely interesting. Still waiting on the last minute modifications for this "private" engine. I have a feeling this match will soon turn into one hell of a war.

For anyone who missed any of the previous matches that Strelka 5.6 played- here is a quick overview:

Code: Select all

Strelka 5.6 x64: (Only Wins & Losses)


Fire 2.2b xTreme GH x64     31-13    (+63)
Komodo 5 64-bit             38-16    (+78)
Ivanhoe 46h x64             31-5     (+92)
IvanHoe 50kQ x64 SSE4.2     27-15    (+42)
Rybka 4.1 x64               26-19    (+24)
Stockfish 2.3.1 pcnt x64    37-7     (+108)
Critter 1.6a 64-bit         32-25    (+12)---200 games
Houdini 2.0c x64            37-48    (-19)---200 games
DeepSaros 3.3 pcnt x64      31-11    (+70)
Gull II x64                 39-8     (+111)
Bouquet 1.5 pop64           41-11    (+108)
IvanHoe 46ei x64 SSE4.2     26-17    (+31)
IvanHoe 46e x64 SSE4.2      25-10    (+53)
RobboLito 0.21Q x64 SSE4.2  28-13    (+53)
Vitruvius 1.19 pcnt x64     29-12    (+60)

Now for the results of this match after game 145:



XPS 8500 Intel i5-4 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
5'+3"
Match=1000 games

Code: Select all

XPS-Core-i5, Blitz:5'+3"  2/23/2013


Strelka 5.6 x64   +10    +34/-30/=81   51.40%   74.5/145
d/tella ve0 x64   -10    +30/-34/=81   48.60%   70.5/145

Interesting, but likely to get even more so!



Yippy ki ya-

george

[geots]

Without bringing up my calculator, doesn't 145 games to 235 games mean that 90 more have been played? Whatever, it is the point at which things are very close to being even:

(Both engines are being closely monitored to be sure neither is receiving help from Houdini.)




XPS 8500 Intel i5-4 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
5'+3"
Match=1000 games

Code: Select all

XPS-Core-i5, Blitz:5'+3"  2/24/2013


d/tella ve0 x64   +1    +52/-51/=132   50.20%   118.0/235
Strelka 5.6 x64   -1    +51/-52/=132   49.80%   117.0/235

And we are still 15 games from the quarter pole. Some very good chess yet to be played.




And it is all being brought to you commercial free-

george

[geots]

If my math does not fail me, again 90 more games have been played. And the true definition of a flip-flop. "Almost" impossible to get any closer than this. If you are waiting for one to break loose and get on a run...........................





XPS 8500 Intel i5-4 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
5'+3"
Match=1000 games

Code: Select all

XPS-Core-i5, Blitz:5'+3"  2/25/2013


Strelka 5.6 x64   +1    +71/-70/=184   50.20%   163.0/325
d/tella ve0 x64   -1    +70/-71/=184   49.80%   162.0/325

See you in "90"-

george

[geots]

d/tella has put at least a slight bit of distance between herself and Strelka 5.6. Actually it doesn't look like much in terms of elo difference, but it is the biggest gap by either engine since it started 415 games ago.




XPS 8500 Intel i5-4 True Cores
Fritz 11 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 2012b.ctg w/12-move limit
5'+3"
Match=1000 games

Code: Select all

XPS-Core-i5, Blitz:5'+3"  2/26/2013


d/tella ve0 x64   +6    +97/-89/=229   51.00%   211.5/415
Strelka 5.6 x64   -6    +89/-97/=229   49.00%   203.5/415

Back in "90"-

george