|
Home
NBE 2003 Tournament
Athlon 1.3/256
MB, using Fritz 8 GUI
time control: 120'/60 plus 60'/30 plus
30'
ponder=off, no opening books, no EGTB
no learning, engines 64
MB hash,
games starting with 5 pre-set positions
1.e4 /
1.d4 / 1.c4 / 1.g3 /
1.f4
each engine to play 10 games against all other
total 70
games per engine, total 280 games
Participants: Chess Tiger 15.0 (CB
normal)
Fritz 8.0.0.26, Hiarcs 9, Hiarcs 8 Bareev
Junior
8.0.0.2, List 504, Shredder 7.04, The King 3.23 SKR
Tester: Kurt
Utzinger / Rolf Bühler, Switzerland
NBE 2003 (final standing: each engine 70 games, total 280 games)
| 1 |
Shredder 7.04 |
|
5.5 - 4.5 |
5.5 - 4.5 |
6.0 - 4.0 |
6.0 - 4.0 |
8.0 - 2.0 |
5.0 - 5.0 |
8.0 - 2.0 |
44.0 / 70 |
|
| 2 |
Fritz 8 |
4.5 - 5.5 |
|
5.0 - 5.0 |
5.5 - 4.5 |
5.0 - 5.0 |
6.5 - 3.5 |
6.5 - 3.5 |
9.0 - 1.0 |
42.0 / 70 |
|
| 3 |
The King 3.23 SKR |
4.5 - 5.5 |
5.0 - 5.0 |
|
6.5 - 3.5 |
7.5 - 2.5 |
6.0 - 4.0 |
6.5 - 3.5 |
4.5 - 5.5 |
40.5 / 70 |
|
| 4 |
Hiarcs 9 |
4.0 - 6.0 |
4.5 - 5.5 |
3.5 - 6.5 |
|
5.5 - 4.5 |
6.5 - 3.5 |
7.5 - 2.5 |
7.0 - 3.0 |
38.5 / 70 |
|
| 5 |
Chess Tiger 15.0 |
4.0 - 6.0 |
5.0 - 5.0 |
2.5 - 7.5 |
4.5 - 5.5 |
|
3.0 - 7.0 |
5.5 - 4.5 |
6.0 - 4.0 |
30.5 / 70 |
|
| 6 |
Hiarcs8 Bareev |
2.0 - 8.0 |
3.5 - 6.5 |
4.0 - 6.0 |
3.5 - 6.5 |
7.0 - 3.0 |
|
5.5 - 4.5 |
5.0 - 5.0 |
30.5 / 70 |
|
| 7 |
Junior 8 |
5.0 - 5.0 |
3.5 - 6.5 |
3.5 - 6.5 |
2.5 - 7.5 |
4.5 - 5.5 |
4.5 - 5.5 |
|
3.5 - 6.5 |
27.0 / 70 |
|
| 8 |
List 504 |
2.0 - 8.0 |
1.0 - 9.0 |
5.5 - 4.5 |
3.0 - 7.0 |
4.0 - 6.0 |
5.0 - 5.0 |
6.5 - 3.5 |
|
27.0 / 70 |
|
Erzeugt mit ChessBase
7.0
Has there ever been a tournament [except of course
at blitz level] over a longer distance where the winner was not Shredder
7.04? We can’t remember, see also our RK 2003
tournament. The german program of Stefan Meyer-Kahlen seems
indeed to be the strongest one at the moment. And this time the engines
had to fight from the very beginning, without their opening books, without
tablebases and no learning functions. It is said of Shredder 7.04 that its
strenght has also something to do with the efficient way the program is
using the tablebases. But despite the handicap of the tournament
conditions, it was still Shredder 7.04 to win the tournament with two
points ahead of Fritz 8.0.0.26. There is one thing we would like to
mention: for us it was the best and most interesting tournament we have
ever played and seen. The games started from the first move and so we
could see real fights from the very beginning instead of looking at games
to start at move 20 or even later which in principle is a rather boring
matter especially in those cases where the engines have jumped directly to
the end game. And due to the missing endgame tablebases (EGTB) the engines
had to demonstrate their knowledge in this phase of the game, nothing
could be hidden. The tournament shows some most interesting things and we
do not think being wrong with the following
conlusions:
Shredder 7.04 is indeed an
allrounder being strong in all phases of a game. The German program was
unable to win a single serie of the five series 1.e4 / 1.d4 / 1.c4 / 1.g3
/ 1.f4 but was never placed more badly than on third rank. Not much
chances had Hiarcs 8 Bareev and List 5.04 as both lost their matches with
2.0-8.0, surely the main reason why Shredder 7.04 could win this
tournament. No other program succeeded in defeating two other programs in
such a way.
It seems as if Fritz
8.0.0.26 has become substantially stronger than the previous versions.
Only two draws vs The King 3.23 SKR and Chess Tiger 15 and a close loss vs
Shredder 7.04 prevented the Dutch program to win the tournament. On the
other hand, Fritz 8.0.0.26 could win the series 1.d4 (10.5/14) and 1.g3
(11.0/14) whereas the worst result with only 6.0/14 happened in the serie
1.f4. The best match result of 9.0-1.0 was achieved vs List 5.04.
The reason why the strong The King 3.23
SKR (settings of Sedat, Kurt, Rolf) did not succeed in getting first
place was a loss of 4.5-5.5 against List 5.04 and in particular the very
bad 1.d4 serie with only 5.5/14. On the other hand the program of Johan de
Koning won (like Fritz 8) two series: 1.e4 with 9.5/14 and 1.f4 with
10.0/14 which is somehwat surprising as both openings have a completely
different character.
Never did Hiarcs
9 make an attempt to be amongst the top three programs in this
tournament. The matches vs Shredder 7.04, Fritz 8 and The King 3.23 SKR
went all lost. Nevertheless, Mark Uniacke’s program was the great winner
of the serie 1.c4 (11.0/14) whilst the same opening was a disaster for
Hiarcs 8 Bareev with 4.5/14 which means last place in this serie. However,
there is still a great gap of 8.0 points to Chess Tiger 15 and Hiarcs 8
Bareev as both these programs are on places 5 and 6. In principle we were
satisfied with Hiarcs 9 and with some more luck
in some games, third
or even second place would have been possible. There is no doubt that the
new version has much improved since Hiarcs 8. And we would not be
surprised if Hiarcs 9 playing style is more efficient than other engines
vs humans.
Once more in one of our tournaments did Chess Tiger
15 fight with less success. Losses of 2.5-7.5 against The King 3.23
SKR and 3.0-7.0 vs Hiarcs 8 Bareev and only two wins with 5.5-4.5 vs
Junior 8 and 6.0-4.0 vs List 5.04 had a too negative influence on the
final standing. Chess Tiger 15 had the worst series with 5.5/14 in the
openings 1.e4 and 1.g3 whilst the best serie was 1.d4 with 7.5/14. We are
still of the opinion that the French program is in comparison with
other programs much stronger at lower time controls.
After the
first rounds it seemed that Hiarcs 8
Bareev would even do better than the latest version Hiarcs 9. But the
final score of 30.5/70 (43 %) was cleary worse than the one of Hiarcs 9
with 38.5/70 (55 %). In the series 1.e4 and 1.d4 Hiarcs 8 Bareev got
slightly better results than Hiarcs 9 but a complete disaster for Hiarcs 8
Bareev was the opening 1.c4 with 4.5/14. Contrary to some people, we can’t
imagine that Hiarcs 8 Bareev should play stronger vs human beings than
Hiarcs 9. Nevertheless, Hiarcs 8 Bareev could win two matches vs Chess
Tiger 15 with 7.0-3.0 but even more surprising was the win of 5.5-4.5 vs
Junior 8.
Somewhat a disappointment was the performance of
Junior 8 in this tournament: second last and/or last place
together with List 5.04 with only 27.0/70 (38.5 %). This result did
however not come as a real surprise to us. Already in our RK 2003
Tournament Old it was Junior 8
to take last place. We then wrote in our RK 2003
Report “What however can already be said is that Junior 8
depends more than other programs on its well optimized opening book. This
book surely contributes a great deal to the overall strength of Junior 8.
This can also be seen from our interrupted tournament RK 2003 where all
engines had to use the book remis.ctg. At the time we stopped this
tournament, Junior 8 had played 21 games and took last place with not more
than 7,5 points.” And now we have an even more clear confirmation of this
statement. Hard to believe but six matches went lost and only the match vs
Shredder 7.04 ended with a draw. Junior 8 had indeed only one good serie
in the 1.e4 opening with a score of 8.5/14 whereas the other series speak
for themselves: 1.d4 = 5.0/14, 1.c4 = 6.0/14, 1.g3 = 4.0/14 and 1.f4 =
3.5/14. Even List 5.04 did better. Is this a proof that Junior 8 has not
much knowledge implemented as far as the general rules in the opening are
concerned? Or does Junior 8 needs open and half-open opening systems to
demonstrate its strength? Or have too less games been played? W have no
satisfying explanation but nevertheless very much like the playing style
of Junior 8, for us still a “must have program”.
And eventually
some words about List 5.04 the only
amateur engine in a very strong field of professional programs. The engine
of Fritz Reul has won our respect by winning its matches vs The King 3.23
SKR [5.5-4.5] and vs Junior 8 [6.5-3.5 !] and the draw [5.0-5.0] vs Hiarcs
8 Bareev. Two heavy losses vs Fritz 8.0.0.26 [1.0-9.0] and vs Shredder
7.04 [2.0-8.0] were responsible that the final outcome was not better. And
very soon, nobody will talk about List 5.04 as there seems already to be a
much stronger version.
Some thoughts about the openings played
Are the naked
engines able to create something like normal openings? Of course, some of
the engines can but others fail. Let us have a look how things are on the
basis of some examples.
In the 1st serie
1.e4 (games 1-56) List 5.04 after 1.e4 d5 2.exd5 Qxd5 3.Nc3 chose
3...Qe6? and lost with one exception [due to bad play of Hiarcs 9] all
games. An even worse example showed game No. 1 when Chess Tiger 15 vs The
King 3.23 SKR had a lost game after 12 moves starting with 1.e4 e6 2.d4 d5
3.Nc3 Bb4 4.Qg4? Nf6 5.Qxg7 Rg8 and so on. A fine game, No. 7, [with the
exchange variation in the Ruy Lopez] demonstrated Shredder 7.04 and Hiarcs
9. Against 1.e4, Junior 8 was rather successful with the Sicilian Defence
1...c5.
The 2nd serie 1.d4 (games 57-112) was a hard task
for most engines. Not a single game can be found with the moves 1.d4 d5
2.c4 and only three games with 1.d4 Nf6 2.c4, all played by Chess Tiger 15
vs The King 3.23 SKR, (game No. 57), vs Fritz 8 (game No. 60) and vs
Shredder 7.04 (game No. 112). The programs do not seem to have the
knowledge that it is important to go for c2-c4 or c7-c5 in the early stage
of the 1.d4 games. And so we can admire only three games (No. 93, 101,
112) with a usual course in the opening.
In the 3rd serie
1.c4 (games 113-168) it was funny to see that 49 of the 56 games started
with 1.c4 e5 2.Nc3. The only exception demonstrated Junior 8 with 1.c4 e5
2.d4? exd4 3.Qxd4 Nc6 4.Qe3? but all the games with this rather bad
opening went lost for Amir Ban’s program. And it was also Junior 8 with
the black pieces to choose another line after 1.c4 with the reply 1...e6
but this time getting a very successful score of 5.0/7. And as already
mentioned it was Hiarcs 9 to be the top scorer in this serie with 11.0/14
(78 %) whereas little brother Hiarcs8 Bareev had its worst serie with only
4.5/14 (32 %).
The 4th serie
1.g3 (games 169-224): Here we can “admire” quite a lot of bad games.
Most engines have no idea how to develop pieces in a “correct” manner
after starting with the fianchetto. And again must Junior 8 be blaimed for
playing a very strange opening with White in five games: 1.g3 e5 2.d4?
exd4 3.Qxd4 Nc6 4.Qe3? and so we were not at all wondering that all theses
games ended with a loss for Junior 8. A further intersting detail: Fritz
8, Shredder 7.04 and The King 3.23 SKR chose after 1.g3 e5 2.d4? exd4
3.Nf3 c5 the pawn sacrifice 4.c3 but only Shredder 7.04 and Fritz 8 got a
positive result with this line. Anyway we have not found one single game
which showed a theoretical well known path after 1.g3.
Also less
positive things about the 5th serie 1.f4 (games 225-280) when we saw
similar things to the 4th serie: not one single game that can be found in
theory. And this serie was the worst one for Junior 8 with only 3.5/14 (25
%) or one draw, six losses with White and two wins, two draws with Black.
In game No. 267, Chess Tiger 15 with Black just needed 22 moves to
demolish Junior 8. But also the game No. 269, The King 3.23 SKR vs Hiarcs
9 lastet only 23 moves for Mark Uniacke’s top program.
Above all
it is obvious that even today’s best chess programs lack a lot of
knowledge in the opening phase. From a pure theoretical point of view,
good games were indeed an exception. On the other hand we must also
understand the programmers who concentrate their work to more important
things as the big opening books help to hide the weaknesses of their
engines.
Tournament course
Here can the course of the
tournament be seen. The tournament [280 games] goes over 10 series at 28
games. Each
serie consists of seven rounds at four games. The cross
tables below show the winners of each serie and is again a
good
example to demonstrate that a big number of games are necessary to say
something concrete. Otherwise we
only have winners by
accident.
Serie 1 with 1.e4 (games 1-56)
Winner =
The King 3.23 SKR
NBE 2003 (serie 1.e4)
| 1 |
The King 3.23 SKR |
|
00 |
½1 |
1½ |
11 |
1½ |
1½ |
½1 |
9.5 / 14 |
|
| 2 |
Shredder 7.04 |
11 |
|
10 |
01 |
11 |
½0 |
10 |
1½ |
9.0 / 14 |
|
| 3 |
Junior 8 |
½0 |
01 |
|
11 |
10 |
½0 |
½1 |
11 |
8.5 / 14 |
|
| 4 |
Hiarcs8 Bareev |
0½ |
10 |
00 |
|
½½ |
10 |
11 |
½1 |
7.0 / 14 |
|
| 5 |
Fritz 8 |
00 |
00 |
01 |
½½ |
|
1½ |
01 |
11 |
6.5 / 14 |
|
| 6 |
Hiarcs 9 |
0½ |
½1 |
½1 |
01 |
0½ |
|
½0 |
½0 |
6.0 / 14 |
|
| 7 |
Chess Tiger 15.0 |
0½ |
01 |
½0 |
00 |
10 |
½1 |
|
01 |
5.5 / 14 |
|
| 8 |
List 504 |
½0 |
0½ |
00 |
½0 |
00 |
½1 |
10 |
|
4.0 / 14 |
|
Erzeugt mit ChessBase
7.0
Serie
2 with 1.d4 (games 57-112)
Winner = Fritz 8
NBE 2003 (serie 1.d4)
| 1 |
Fritz 8 |
|
1½ |
½0 |
½1 |
1½ |
11 |
11 |
½1 |
10.5 / 14 |
|
| 2 |
Shredder 7.04 |
0½ |
|
10 |
1½ |
11 |
11 |
10 |
10 |
9.0 / 14 |
|
| 3 |
Chess Tiger 15.0 |
½1 |
01 |
|
½½ |
10 |
½½ |
01 |
01 |
7.5 / 14 |
|
| 4 |
Hiarcs8 Bareev |
½0 |
0½ |
½½ |
|
½½ |
½½ |
1½ |
10 |
6.5 / 14 |
|
| 5 |
Hiarcs 9 |
0½ |
00 |
01 |
½½ |
|
½1 |
½0 |
1½ |
6.0 / 14 |
38.50 |
| 6 |
List 504 |
00 |
00 |
½½ |
½½ |
½0 |
|
½1 |
11 |
6.0 / 14 |
35.25 |
| 7 |
The King 3.23 SKR |
00 |
01 |
10 |
0½ |
½1 |
½0 |
|
½½ |
5.5 / 14 |
|
| 8 |
Junior 8 |
½0 |
01 |
10 |
01 |
0½ |
00 |
½½ |
|
5.0 / 14 |
|
Erzeugt mit ChessBase
7.0
Serie
3 with 1.c4 (games 113-168)
Winner = Hiarcs 9
NBE 2003 (serie 1.c4)
| 1 |
Hiarcs 9 |
|
½½ |
½½ |
½1 |
1½ |
11 |
11 |
11 |
11.0 / 14 |
|
| 2 |
Fritz 8 |
½½ |
|
½½ |
½1 |
10 |
10 |
½½ |
½1 |
8.0 / 14 |
|
| 3 |
Shredder 7.04 |
½½ |
½½ |
|
½0 |
11 |
0½ |
½½ |
1½ |
7.5 / 14 |
|
| 4 |
The King 3.23 SKR |
½0 |
½0 |
½1 |
|
0½ |
10 |
½½ |
11 |
7.0 / 14 |
|
| 5 |
List 504 |
0½ |
01 |
00 |
1½ |
|
1½ |
½0 |
01 |
6.0 / 14 |
40.50 |
| 6 |
Junior 8 |
00 |
01 |
1½ |
01 |
0½ |
|
10 |
10 |
6.0 / 14 |
39.75 |
| 7 |
Chess Tiger 15.0 |
00 |
½½ |
½½ |
½½ |
½1 |
01 |
|
½0 |
6.0 / 14 |
39.75 |
| 8 |
Hiarcs8 Bareev |
00 |
½0 |
0½ |
00 |
10 |
01 |
½1 |
|
4.5 / 14 |
|
Erzeugt mit ChessBase
7.0
Serie
4 with 1.g3 (games 169-224)
Winner = Fritz 8.0.0.8
NBE 2003 (serie 1.g3)
| 1 |
Fritz 8 |
|
01 |
01 |
½1 |
11 |
½1 |
11 |
11 |
11.0 / 14 |
|
| 2 |
Shredder 7.04 |
10 |
|
00 |
1½ |
11 |
10 |
11 |
1½ |
9.0 / 14 |
|
| 3 |
The King 3.23 SKR |
10 |
11 |
|
10 |
00 |
11 |
01 |
1½ |
8.5 / 14 |
|
| 4 |
Hiarcs 9 |
½0 |
0½ |
01 |
|
01 |
1½ |
1½ |
01 |
7.0 / 14 |
|
| 5 |
Hiarcs8 Bareev |
00 |
00 |
11 |
10 |
|
½½ |
½0 |
11 |
6.5 / 14 |
|
| 6 |
Chess Tiger 15.0 |
½0 |
01 |
00 |
0½ |
½½ |
|
1½ |
01 |
5.5 / 14 |
|
| 7 |
List 504 |
00 |
00 |
10 |
0½ |
½1 |
0½ |
|
10 |
4.5 / 14 |
|
| 8 |
Junior 8 |
00 |
0½ |
0½ |
10 |
00 |
10 |
01 |
|
4.0 / 14 |
|
Erzeugt mit ChessBase
7.0
Serie
5 with 1.f4 (Games 225-280)
Winner = Junior 8 / The King
3.23 skr
NBE 2003 (serie 1.f4)
| 1 |
The King 3.23 SKR |
|
00 |
11 |
10 |
1½ |
11 |
11 |
½1 |
10.0 / 14 |
|
| 2 |
Shredder 7.04 |
11 |
|
10 |
0½ |
01 |
11 |
11 |
10 |
9.5 / 14 |
|
| 3 |
Hiarcs 9 |
00 |
01 |
|
11 |
11 |
½0 |
½1 |
½1 |
8.5 / 14 |
|
| 4 |
List 504 |
01 |
1½ |
00 |
|
00 |
01 |
10 |
11 |
6.5 / 14 |
|
| 5 |
Fritz 8 |
0½ |
10 |
00 |
11 |
|
½½ |
0½ |
10 |
6.0 / 14 |
40.00 |
| 6 |
Chess Tiger 15.0 |
00 |
00 |
½1 |
10 |
½½ |
|
½0 |
11 |
6.0 / 14 |
35.25 |
| 7 |
Hiarcs8 Bareev |
00 |
00 |
½0 |
01 |
1½ |
½1 |
|
½1 |
6.0 / 14 |
34.00 |
| 8 |
Junior 8 |
½0 |
01 |
½0 |
00 |
01 |
00 |
½0 |
|
3.5 / 14 |
|
Erzeugt mit ChessBase
7.0
Home
|
