Novag Obsidian vs Saitek Travel Champion 2100

sje · Post by **sje** » Fri Nov 01, 2013 7:15 pm

For its last move, the machine ran for only fifteen hours or so because I had restarted the search. I don't think the early interruption made any difference. The evaluation was -0.24 pawns for White.

A correction: The Schach-Computer wiki says that the Obsidian has 1,024 bytes of RAM, not 768 bytes. So maybe each move takes three (not two) bytes of storage on the move stack.

Wardy · Post by **Wardy** » Fri Nov 01, 2013 7:20 pm

13).......Qc7

Still in book

sje · Post by **sje** » Sat Nov 02, 2013 6:35 pm

[Event "Correspondence Game"]
[Site "http://hiarcs.net/forums/viewtopic.php?p=62613"]
[Date "2013.10.30"]
[Round "-"]
[White "Novag Obsidian"]
[Black "Saitek Travel Champion 2100"]
[Result "*"]

1 e4 e5 2 Nf3 Nc6 3 Bb5 a6 4 Ba4 Nf6 5 O-O Be7 6 Re1 b5 7 Bb3 O-O 8 c3 d6 9 h3
Na5 10 Bc2 c5 11 d4 cxd4 12 cxd4 Bb7 13 Nc3 Qc7 14 dxe5 *

[fen]r4rk1/1bq1bppp/p2p1n2/np2P3/4P3/2N2N1P/PPB2PP1/R1BQR1K1 b - - 0 14[/fen]
The search finished early, but I don't know how early.

Wardy · Post by **Wardy** » Sat Nov 02, 2013 7:02 pm

The tc2100 is finally out of book with this move and initially rates the position as +0.3 for white..... At 8 ply.

Move tomorrow.

Wardy · Post by **Wardy** » Sat Nov 02, 2013 8:19 pm

After an hour and at the commencement of 10 ply the scores falls to +0.1 for white, averaging around 3k moves per second currently.

Wardy · Post by **Wardy** » Sun Nov 03, 2013 7:03 pm

14)........... dxe5

with a score of +0.1 for white

The TC2100 reached 12ply but didn't get past considering the 1st move at this depth, obviously the branching factor and only 3k nodes/sec reach a point where the time required goes well beyond 24 hours.

the expected continuation is

15) a4 Rfd8
16) Bd2......

though for a very long time it was fixated on

15) Nd5 Nxd5
16) exd5

sje · Post by **sje** » Sun Nov 03, 2013 7:34 pm

At first glance, the Obsidian is looking at 15 Nd5 Nxd5 16 exd5.

The machine does not display the node frequency. However, the Novag Citrine does display this using nearly the same program, CPU, and speed. The Citrine averages about 4,000 nodes/second in the middlegame, so I guess the Obsidian is doing the same.

In a simple endgame, the Citrine can search about 14,000 nodes per second. Unfortunately, the Obsidian and the Citrine lack some endgame knowledge that Novag had in its Super Constellation twenty years earlier. An example: the Super Constellation could always win a won KBNK endgame, but the latter pair of machines can't.

sje · Post by **sje** » Mon Nov 04, 2013 1:44 pm

The Obsidian finished early again, this time by at least seven hours.

[Event "Correspondence Game"]
[Site "http://hiarcs.net/forums/viewtopic.php?p=62613"]
[Date "2013.10.30"]
[Round "-"]
[White "Novag Obsidian"]
[Black "Saitek Travel Champion 2100"]
[Result "*"]

1 e4 e5 2 Nf3 Nc6 3 Bb5 a6 4 Ba4 Nf6 5 O-O Be7 6 Re1 b5 7 Bb3 O-O 8 c3 d6 9 h3
Na5 10 Bc2 c5 11 d4 cxd4 12 cxd4 Bb7 13 Nc3 Qc7 14 dxe5 dxe5 15 Nd5 *

[fen]r4rk1/1bq1bppp/p4n2/np1Np3/4P3/5N1P/PPB2PP1/R1BQR1K1 b - - 1 15[/fen]

I read about the "H8 bug" reportedly present on a number of dedicated computers including the TC2100. Have you seen any evidence of this bug on your machine?

http://www.schach-computer.info/wiki/in ... ugs#H8_Bug

Wardy · Post by **Wardy** » Mon Nov 04, 2013 7:01 pm

I'm aware of the H8 bug but have not seen it manifest itself in actual play. Hopefully we won't see this settled via a bug!

Wardy · Post by **Wardy** » Mon Nov 04, 2013 8:23 pm

What was the Obsidian expecting in reply?

sje · Post by **sje** » Mon Nov 04, 2013 8:24 pm

15 Nd5 Nxd5 16 exd5

Wardy · Post by **Wardy** » Mon Nov 04, 2013 8:33 pm

At 9 ply it's thinking of playing Qd6......still +0.2 for white at this point

Wardy · Post by **Wardy** » Tue Nov 05, 2013 5:26 pm

15)............ Nxd5

An accidental key press whilst poking about in the display ended the search after 21+ hours, however at 20 hours it had just started to search at 12 ply so I doubt I've hindered the machine at all.

Machine rates the position as 0.0 with the continuation being

16) exd5 Rad8
17) Nxe5

Deep Fritz 12 has the same continuation with a tiny advantage for black at 30 ply or so. I don't have my usual machine available for analysis.

I've also discovered that it's possible to change the batteries quickly without losing game data.Result!

sje · Post by **sje** » Wed Nov 06, 2013 2:32 pm

[fen]r4rk1/1bq1bppp/p7/np1np3/4P3/5N1P/PPB2PP1/R1BQR1K1 w - - 0 16[/fen]
With 3:25 left, the Obsidian is at ply 11 and saying 16 exd5 f6 17 Bd2 with a -0.41 score.

In a week or so, I'll be getting a copy of Fritz 13 to run on my 1.7GHz AMD E2-1800 dual core 64 bit CPU 4 GB RAM notebook. The program can provide some more commentary.

My own Symbolic says:
[-0.414/14/4:21.727/173,828,415/0] 16 exd5 Rad8 17 Nxe5 Rxd5 18 Qe2 Bd6 19 Nf3 Rc5 20 Bd3 Nc4 21 b4 Rh5 22 Bd2 Nxd2 23 Nxd2

sje · Post by **sje** » Wed Nov 06, 2013 5:14 pm

[Event "Correspondence Game"]
[Site "http://hiarcs.net/forums/viewtopic.php?p=62613"]
[Date "2013.10.30"]
[Round "-"]
[White "Novag Obsidian"]
[Black "Saitek Travel Champion 2100"]
[Result "*"]

1 e4 e5 2 Nf3 Nc6 3 Bb5 a6 4 Ba4 Nf6 5 O-O Be7 6 Re1 b5 7 Bb3 O-O 8 c3 d6 9 h3
Na5 10 Bc2 c5 11 d4 cxd4 12 cxd4 Bb7 13 Nc3 Qc7 14 dxe5 dxe5 15 Nd5 Nxd5 16
exd5 *

[fen]r4rk1/1bq1bppp/p7/np1Pp3/8/5N1P/PPB2PP1/R1BQR1K1 b - - 0 16[/fen]