Genius5 destroys MChess

[Sylvanus McLeod]

Hi all,

I aquired the Auto232 program a few months ago and did a test between
Genius5 (in dos mode) and MChess 6.0

The machines used was a Cyrix 200MX with 64megs of ram (I will call this
machine A) The second was an AMD K6 233MMX also with 64megs of ram (This I
will call machine B)

MChess 6 playing white on A vs Genius5 playing black on B in 24 games got a
score of 6(W) - 13(L) - 5(D) for a score of 8.5 - 15.5

MChess 6 playing black on A vs Genius5 playing white on B in 24 games got a
score of 3(W) - 17(L) - 4(D) for a score of 5 - 19

The total score with MChess 6 on A vs Genius5 on B is an unbelievable 34.5 -
13.5 in favour of Genius5.

After that I switched the machines around. Genius5 was now on A and MChess
6 on B.

Genius5 playing white on A vs MChess 6 playing black on B in 16 games got a
score of 10(W) - 2(L) - 4(D) for a score of 12 - 4

Genius5 playing black on A vs MChess 6 playing white on B in 24 games got a
score of 13(W) - 5(L) - 6(D) for a score of 16 - 8

The total score with Genius5 on A vs MChess 6 on B is also an unbelievable
28 - 12 in favour of Genius5.

This brings the total score in 88 games to 65.5 - 25.5 in favour of Genius5.
All games were played in the standard 40/2hrs and with all defaults set.

I them compared my test with that of the SSDF to find that in 20 games
MChess defeated Genius5 by a score of 11.5 - 8.5.

I find the SSDF tests as an insult to the teams of MChess6 and that of
Genius5. I will also conduct test on Hiarcs6, Rebel8 and Rebel9. These
results will be posted as they become available.

[Komputer Korner]

It would be interesting to know how much opening books played a part in
this. I am not questioning your results. Genius 5 is strong.
--
- -
Komputer Korner

The inkompetent komputer

If you see a 1 in my email address, take it out before replying.
Please do not email both me and the r.g.c.c. at the same time. I read all
the postings on r.g.c.c.

Sylvanus McLeod <smc...@cyberglobe.net> wrote in article
<34494...@news.cyberglobe.net>...

[Robert Hyatt]

Robert Sullivan (bob4...@earthlink.net) wrote:
: Korner and Ledoc ask interesting questions that wouldn't occur to this
: unsophisticated fan. I wonder if Enrique, Bob, Chris, Ingo, Howard, et al.,
: could agree on a list of variables which must be controlled for in order to
: get some reasonably useful data. Also, what would N need to be?
: Robert

I really can't answer this. Computer Chess seems to somehow lie outside
the normal realm of statistics. IE, I'd thought that Crafty had been playing
pretty well of late, winning about 30+ games against GM's with no losses at
all. Then along came WchessX and a series of 5 games ended in 4 losses and
one draw. I went off and studied the games, but when I got back, I found
that Crafty had won 5 in a row. So I don't know how many games you have to
play to get a real estimate of relative strengths...

[Harald]

quoting a mail from smc...@cyberglobe.net

> I aquired the Auto232 program a few months ago and did a test between
> Genius5 (in dos mode) and MChess 6.0
>
> The machines used was a Cyrix 200MX with 64megs of ram (I will call this
> machine A) The second was an AMD K6 233MMX also with 64megs of ram (This I
> will call machine B)

snip

> Genius5 playing black on A vs MChess 6 playing white on B in 24 games got a
> score of 13(W) - 5(L) - 6(D) for a score of 16 - 8
>
> The total score with Genius5 on A vs MChess 6 on B is also an unbelievable
> 28 - 12 in favour of Genius5.
>

> This brings the total score in 88 games to 65.5 - 25.5 in favour of Genius5.
> All games were played in the standard 40/2hrs and with all defaults set.

Interesting.

> I them compared my test with that of the SSDF to find that in 20 games
> MChess defeated Genius5 by a score of 11.5 - 8.5.

Are these games on 200MMX or P90?

> I find the SSDF tests as an insult to the teams of MChess6 and that of
> Genius5. I will also conduct test on Hiarcs6, Rebel8 and Rebel9. These
> results will be posted as they become available.

My autoplayer-tests confirmed the SSDF almost right but I don't have
enough games, only 20 per match.
I had MCP6-Genius5 4.5-5.5
Genius5-MCP6 2.0-8.0

which makes 12.5-7.5 for MCP6
Important is the hardware: MCP6 on P133/30MB hash and Genius5 on 486-133/
15MB hash. So Genius5 did well! But I doubt that your results can be
confirmed.

Harald Faber