More Details:
Atlas wins 3 times already with 1 core over 6 cores, amazing...bravo to A. Manzanares
Andscacs managed to eliminate one the Top engines: Chiron, well-done to D. Queralto
Texel, Hannibal are another qualified Top engines, which perform very good too ...
And so far, a superior performance by Houdini, Stockfish, Komodo, Synapse
Note also that the next Round 4 (Octa-Final) will be played at 150 Min +30 Sec
Later, the Semi-Final is planning to be played at 300 Min + 30 Sec
And only a few days are left to the biggest FINAL, which will be at 600 Min + 30 Sec
More Details:
First of all, I'd like to congratulate all players...!!
No BIG surprise that the current four engines are qualified to the semi-finals
Note that finely, Atlas is eliminated by Komodo, but however even with 1 core:
Atlas managed to draw the 1st game over K91 6 cores, plus if we compare the Elo of both engines
And to be honest, that means a lot for me...!
Yes..really I wonder a lot, who will be WSET champion, very hard to guess, let's see...
Houdini is eliminated, so Komodo is the qualified player for the FINAL !
My special congratulations to Komodo team !!
About Stockfish vs Synapse match,
The semi-final's result is still not clear...
The 1st game ended as draw, 2nd game is still active...
Note that both engines are using the same opening (reversed colors)
Breaking News:
Synapse RZ4 WSET MP x64 eliminates Stockfish 170815 MP x64
Special Congratulations to Alejandro Torres too !
More Details:
Synapse won with a score 4.5 - 3.5 (on extra games)
Firstly a match of 2 games are played (300 Min +30 Sec with reversed colors),
Where both games are ended as draw, later a match of 4 games are played (15 Min + 10 Sec with reversed colors),
The results were still drawn, finely Synapse won in a match of 2 games (3 Min + 2 Sec with reversed colors)
Unfortunately, this time the reigning WSET Champion 2014 (Stockfish) lost its title...
To be honest, since a long time, I have never seen such exiting engine tournament!
And very soon I plan to start WSET 2015's Final : Komodo vs Synapse
FINAL (600 Min +30 Sec) - The Current Position after 12 Moves
Details:
The semi-finalists, finalist used to play under Arena 3 GUI, due to I set giant slow time controls
In this way, e.g in case of power cut: I can resume the game, exactly at that position...
Because Arena GUI has a superior option: it saves current game after every played move
Also I have another possibility, to copy/paste and to publish the current position over chess forums
I set openings (based on Perfect 2015x book) mainly up to 5-6 moves for the quarter-finals, final
Probably the current game will take approx. 1 (one) day...
And depending on my free time...I plan to make often updates
WSET 2015's Final (600 Min + 30 Sec) - Komodo vs Synapse
It seems,
Komodo has really a very BIG chance to be the champion !
According to my experience,
Usually the latest top engines suffer (as Blacks) with D44 Semi-Slav
That's why D44 is not allowing for playing in Perfect 2015 books
Note that I set a short line: 1. d4 Nf6 2. c4 e6 3. Nf3 d5 4. Nc3 c6
So..I am not the reason about why Synapse preferred D44 Semi-Slav
For example, Komodo preferred D43 with playing 6.h6 (during his turn)
And after all,
I am not much surprised that the current score is in favor for Komodo
The Shortest Wins:
1)Rhetoric 1.4 x64 vs Fizbo 1.5 MP x64 0-1 (Fizbo as Blacks wins after 29 moves)
*The opening is A34 English: it is too risky and dangerous for both sides
2)Komodo 9.1 MP x64 vs Senpai 1.0 MP x64 1-0 (Komodo as Whites wins after 34 moves)
*The opening is D47: Semi Slav (well-balanced for both sides, in case of near Elo strength)
* And I think that the power of Komodo played the biggest role
3)Komodo 9.1 MP x64 vs Atlas 3.80 x64 1-0 (Komodo as Whites wins after 34 moves)
*The opening is A33 English: In my opinion it is a strong opening for both sides
*I noticed that, Atlas missed to play 10.Bc5, this is the reason about why Atlas lost in short moves...
----------------------------------
The Longest Wins:
1)Andscacs 0.82 MP x64 vs Chiron 2 MP x64 1-0 (Andscacs as Whites wins after 243 moves)
*The opening is B67...sometime some Top engines (as Blacks) suffer with this opening
2)Synapse RZ4 WSET MP x64 vs Stockfish 170815 MP x64 1-0 (Synapse as Whites wins after 136 moves)
The opening is C11: French, it is risky opening (e.g similar to be risky as B67 Sicilan)
Both engines are tested twice with the same opening C11, where Synapse managed to draw...
3)Hannibal 1.5 MP x64 vs Octochess 5190 MP x64 1-0 (Hannibal as Whites wins after 109 moves)
The opening is E59: Nimzo-Indian (balanced opening for both sides, in case of near Elo strength)
*And I think that here the strength of Hannibal played the main role about why...
---------------------------------
Games : 116 (finished)
White Wins : 47 (40.5 %)
Black Wins : 22 (19.0 %)
Draws : 47 (40.5 %)
Unfinished : 0
White Perf. : 60.8 %
Black Perf. : 39.2 %
ECO A = 12 Games (10.3 %)
ECO B = 31 Games (26.7 %)
ECO C = 29 Games (25.0 %)
ECO D = 38 Games (32.8 %)
ECO E = 6 Games ( 5.2 %)
The overall draw percentage is approx. 40%
And who said that Perfect book's opening lines are too drawish ?)
The number of draws percentage looks good...of course it is due to the Elo differences...
More Details
TTs official ranking includes only (since 30.08.2015):
1) Public and strongest versions
2) Engines, which are up to 55% similarity (+56% out)
* Exception there will be for:
- Private engine, in case of better performance than the leader (Houdini 4 tactical)
- For derivative engines: in case of +20 points better than based original engine
- For clone engines: in case of +100 points better than based original engine
* Houdini 4 Tactical (based mainly on Rybka) managed to do that...why not other engine?!
* Otherwise, TTs rankings will be based mainly on Stockfish/Rybka derivatives or clones
* Thanks for your understanding !
A few notes more,
Stockfish 170815 performed better than all previous SF development versions
*But however, even this new SF release could not do better than Stockfish 6
Synapse RZ4 WSET Tactical x64 has been tested with 'Tactical mode' enabled
Ginko 1.2 x64 performed 14 points better than previous Ginkgo 1.0e x64
Giraffe 280815 x64 performed 29 points better than Giraffe 010815 x64
Congrats to all engine authors who managed to improve their engines...
Unexpected and disappointed tactical results by some newer versions:
- Komodo 9.2 x64 (17 points less than Komodo 9.01 x64)
- Tornado 7 x64 (12 points less than Tornado 5 x64)
- Andscacs 0.82 x64 (8 points less than Andscacs 0.81 x64)
- Arminius 060815 x64 (1 point less than Arminius 2014 x64)
Some new engines which have analyze bugs:
Alfil 15.8 and Fizbo 1.5 have the same analyze bug as Protector 1.8 and Stockfish 5
I don't know exactly, but it seems some authors copy or study also the bad parts...
Also since today,
I will not use any polyglot adapter for those buggy UCI engines, due to we are in 2015
And I hope the newer versions can analyze and to be better in tactics !)
Note that I decided to work completely new...
Just because the latest SF version supports up to 128 cores
Plus I expect to see better MP scaling by the current version...
Yes...now it's your turn...and I hope to see many new SF benchmarks !)
More Details,
-Houdini 3 Tact performed incredible strong, slightly weaker than Houdini 4 Tact
-StockfishTS 240915: best performance so far, comparing with other SF versions
Some engines, which crashed on my systems:
Embla 0.5, Neurone XXIII
SCCT - TTs (3 seconds per position) - All Versions:
https://sites.google.com/site/computerschess/tts-3sec
After some hot discussions...!)
I've decided to run a new hardware speed test !
And here are the results:
1st Hardware Speed Test
Conditions:
Intel Core i7-980X
Hyper-Threading OFF
Large Pages OFF
Arena 2.01 Chess GUI
Stockfish 290815 x64 modern
3 Seconds per move
128 MB hashtable size
600 Tactical positions
kN/s values are based on WSET 2015's bench position
Conditions:
Intel Core i5-3210M 2.50GHz
Hyper-Threading ON
Large Pages OFF
Arena 2.01 Chess GUI
Stockfish 290815 x64 modern
3 Seconds per move
128 MB hashtable size
600 Tactical positions
kN/s values are based on WSET 2015's bench position
Details,
This hardware speed testing is different than all, never played before !)
The goal of this test is that: the kN/s values are quite useful for measuring...
Note also Stockfish 290815 breakes the formula rules under the current conditions !)
As we see, Stockfish 2.01GHz 2 core performed better than Stockfish 4.02GHz 1 core
More Details,
Some experts have different views, but I hope after checking the current results:
- They will change their opinions... because we are in 2015 !)
For any questions about the current test, please feel free to ask...
In shortly,
I will test Stockfish 290815, 1 core, with same 600 positions, on i7-980X 3.33GHz
But this time, I will set Time Control: 11 seconds per move
Really I wonder a lot about what will be the results...
And what is your estimation ?)
SF 1 core 11 sec/move will be performed stronger than SF 4 core 3 sec/move ?)
Note: SF 4 core 3 sec/move i7 980X 3.33 GHz managed to get: 374 points
The speed hardware test has n completed, so here is the final result:
Stockfish 290815 x64 1 core 11 sec/move i7 980X 3.33 GHz: 395 points
What I can say more,
This time, SF 1 core 11 sec/move performed clearly better than SF 4 core 3 sec/move
For example, SF 1 core 11 sec/move's performance is very close to SF 6 core 3 sec/move
More Details,
Used unsolved positions in 3 sec/move 1 core by the current chess engine versions
Note that (for better comparing) I decided to test Houdini 4 6 core 6 sec/move too
As I expected, SF 6 cores 6sec/move performed very close to 1 core 30 sec/move
Note also that,
I tested SF 6c 6sec/move with 128 MB hash too: 43/132
Also SF 6c 6sec/move tested with 1024 MB hash: 42/132
What I can say more,
We can not use any right 'formula' in case of comparing the hardware speeds !
It all depends on our used conditions...some engines perform better some worse...
But however (in case of measuring the hardware speeds)
Stockfish 290815's KN/s values seems to be not bad choice...
Of course, one of the best choices is that: running similar test suite positions
Or running engine matches in Auto232 mode or via LAN etc...
n completed, so here is the final result: