And Stockfish 2.2.2 Beats the Odds!

[geots]

DeepSaros ver.2.3iT vs. Stockfish 2.2.2

A very interesting match with, for me, an unknown engine. 2.3j is the latest DeepSaros version. This is 2.3i, the previous version, using the 2.3iT executable- the "T" meaning "turbo"- an experimental version that is supposed to be tuned especially for blitz games. As to its success, so far I have not seen a lot of evidence supporting that. So the jury is still out.

Quite often I see a particular engine win 3 games in a row. It doesn't happen in every match, but it is not a rarity. However, with 30 games already drawn in this match, Stockfish trailed 10 - 7 after 47 games. He could not afford a loss or even a draw. He had to win all 3 remaining games to tie the match.

The only thing in Stockfish's favor was that DeepSaros did not realize a mere draw would seal the victory. If he had, he could have played all 3 games attempting at least 1 draw. At any rate, Stockfish managed to roll off 3 consecutive victories when he needed it the most.



1CPU/32bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games


DeepSaros ver.2.3iT.......0........+10/-10/=30.......50.00%.......25.0/50

Stockfish 2.2.2..............0.........+10/-10/=30.......50.00%.......25.0/50


An interesting match. Saros 2.3iT certainly did well enough to earn another match at some point in time. Maybe very soon.



Going, going, gone-

g
e
o
r
g
e

[geots]

DeepSaros ver.2.3iT vs. Stockfish 2.2.2

A very interesting match with, for me, an unknown engine. 2.3j is the latest DeepSaros version. This is 2.3i, the previous version, using the 2.3iT executable- the "T" meaning "turbo"- an experimental version that is supposed to be tuned especially for blitz games. As to its success, so far I have not seen a lot of evidence supporting that. So the jury is still out.

Quite often I see a particular engine win 3 games in a row. It doesn't happen in every match, but it is not a rarity. However, with 30 games already drawn in this match, Stockfish trailed 10 - 7 after 47 games. He could not afford a loss or even a draw. He had to win all 3 remaining games to tie the match.

The only thing in Stockfish's favor was that DeepSaros did not realize a mere draw would seal the victory. If he had, he could have played all 3 games attempting at least 1 draw. At any rate, Stockfish managed to roll off 3 consecutive victories when he needed it the most.



1CPU/32bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games


DeepSaros ver.2.3iT.......0........+10/-10/=30.......50.00%.......25.0/50

Stockfish 2.2.2..............0.........+10/-10/=30.......50.00%.......25.0/50


An interesting match. Saros 2.3iT certainly did well enough to earn another match at some point in time. Maybe very soon.



Going, going, gone-

g
e
o
r
g
e

What I am trying to determine is if the "T" executables for DeepSaros that are supposed to be for blitz- actually are better. As they are in Carotino's words, "experimental versions".

The above version against Stockfish is the "T executable" for version 2.3i. Shortly I will post the match that followed in which the default 2.3i version was used against same Stockfish version.


george