Chess program Tournament - 4 Matches, Round Robin - COMPLETED !!
Alex_Lux
IMO, 'base' for marks/rating should be results on Tournament among
chess programs. Of course playing ONE game isn't enough! Playing TWO,
one as black, one as white is better... but 1st good 'solution' is to
play FOUR games (2 white, 2 black).
So I decided to make such Tour one day...
In the next few posts here, I'll present results, leaderboard etc. now
it's time for some extracts (features):
* 12 programs in competition, after qualification matches, 7 made to
the Round-Robin Tournament!
* few screenshots for every (!!!) of 84 matches, moves listings
whenever is possible;
* playback files (RZX) for 63 matches;
* fair play settings for response time (approx. 120 minutes for 40
moves, i.e. 180 second settings for programs like Colossus 4.2 or
Super Chess 3.5)
* good ideas "what to do next"!!.
and many more...
BTW, I have 2 kinds of playback, real-time (approx. 1MB) and replay-
mode (e.g. every 5 seconds, one move, such RZX is approx. 30kB, very
convenient for web storage!)
More details, how I played it, how I set all, and results from 1st
Round in next mail.
One more thing, as my main 'Speccy-action' ATM is publishing site
'bout Database with ratings/marks for 'all' Speccy games (which is 98%
done) if someone is interested to make site about this (chess)
Tournament, I can send him all files (and write all needed comments,
match-reviews etc.) or be patient and wait for me to do it. (I guess
it be in one month, maybe 3 weeks?).
Aleksandar
Alex_Lux
- Well known, strong chess programs, entered the Tour automatically.
It was: Cyrus 2, Sup.Chess 3.5, Colossus 4.2, and Clock Chess '89;
- in 1st 'qualifier', Cyrus 1 'demolished' Chess (Psion);
- in 2nd 'qualifier', Spectrum Ch.2 (Artic) and Turk played DRAW (both
win as black!), Turk advances as 'nicer' (not only by look, but due to
its playback, listings, time shown..);
- in 3rd 'qualifier', Turk and Chess Player (quicksilva) played DRAW
(now both win as WHITE !!!), you guess, Turk advance;
- in 4th 'qualifier', Chess Player had a 2nd chance, with Cyrus 1...
clean sweep for Cyrus 1;
- in 5th 'qualifier', Master Chess '87 was obliterated by Cyrus 1;
- in 6th 'qualifier', Psi Chess - Chess Master 2000 1.5:0.5, Psi
qualifies as I needed one more 'new' program.
So, Cyrus 1, The Turk (Oxford CP), and Psi Chess (Edge) joins 'fab
four'
* Emulator (or how I 'played' that Tour)
- Spectaculator was used, 2 opened 'windows', one program as white in
one window, its 'challenger' as black in other window;
- I used 'speed up' shortcuts - Ctrl & + for 1000 % speed, Ctrl & -
for normal speed, thanks to this, an match with aggregate thinking
time of 6 or 7 hours can be played whitin 2 hours (I like to reset to
normal spped when it's time to transfer from one window to another,
i.e. for 'human' move).
* Program settings:
- Cyrus 2 - level 8 (claims 4 min/move but plays much faster e.g.
167min-62mov.);
- Psi Chess - level A7 (6 min/move);
- Sup.Chess 3.5 - 180 seconds;
- Colossus 4.2. - 180 seconds, "Think ahead" during opponents move set
ON;
- Clock Chess '89 - 180 seconds, 128kB version used, "Think ahead"
during opponents move set ON;
- Turk - level 4 (claims 5 min/move, during 'test' games, it varies
from 180min-43mov. to 226min-31mov.!!);
- Cyrus 1 - level 7 (claims 3 min/move).
More about Cyrus 1 - I played several test matches as white and black,
on lev.7 and lev.8, and measure time 'manually', in order to find
which level is appropriate...
...I'll show you one detail, average time for 'crucial moves' (i.e.
from 12th move to 17th). On level 7, both Cyrus 1 and 2, take
respectively, as white 2.7 and 2.5 min/move. As black, 4 and 1.7 min/
move. On level 8, as white it was 6.5 and 4 min/move. As black, 4 (cyr.
1) and 3 (cyr.2) min/move.
Now you see, it's 'fair' to use level 7 for Cyrus1 (2.7 & 4 min/move)
and level 8 for Cyrus2 (4 & 3 min/move).
Just to mention settings for progs which fail to qualify:
Chess (Psion) - level 5 (4 min/move), Spectrum ch.2 (Artic) - 300
seconds (!!!) (compared to this, Cyrus1 is handicaped with its shorten
response time, but even that can't 'save' Artic's program!), Chess
Player (Quicksilva) - level 3 (7 min/move), Master Chess
(Mastertronic) - 180 seconds (this DON'T WORK!! use 'space' to force
move after I measure 3 minutes!), and finally ChessMaster 2000 - level
7.
This post tends to be too long, so I'll stop here. 1st round results
in next post.
Alex_Lux
Turk - Cyrus2 0:4
Clock Chess '89 - Psi chess 4:0
Colossus4.2 - Sup.Chess3.5 1.5:2.5
Turk - Cyrus2:
1st match, Turk white, 37 moves, Turk time 4:07:29, Cyrus2 time
1:28:39, Cyrus2 wins.
2nd match, Turk white, 54 moves, Turk time 4:30:05, Cyrus2 time
2:11:15, Cyrus2 wins.
3rd match, Cyrus2 white, 43 moves, Turk time 4:45:14, Cyrus2 time
1:44:49, Cyrus2 wins.
4th match, Cyrus2 white, 17 moves, Turk time 0:42:06, Cyrus2 time
0:24:47, Cyrus2 wins. (expess win!)
Clock '89 - Psi:
1st match, Clock'89 white, 24 moves, Clock'89 time 0:53:25, Psi time
1:35:59, Clock'89 wins.
2nd match, Clock'89 white, 21 moves, Clock'89 time 0:46:07, Psi time
1:44:37, Clock'89 wins. (almost express
3rd match, Psi white, 24 moves, Clock'89 time 0:53:42, Psi time
1:18:32, Clock'89 wins.
4th match, Psi white, 42 moves, Clock'89 time 2:05:34, Psi time
2:16:19, Clock'89 wins.
And finaly, 'derby' of 1st Round, best program in 1985 versus strong
challenger from 1986...
Colossus4 - Super Ch.3.5:
1st match, Sup.Ch.3.5 white, 75 moves, Colossus time 5:01:01,
Sup.chess3.5 time 3:36:21, Super wins. (almost 9 hours!)
2nd match, Sup.Ch.3.5 white, 93 moves, Colossus time ???, Sup.chess3.5
time 4:40:23, drawn.
3rd match, Colossus white, 27 moves, Colossus time 1:42:55,
Sup.chess3.5 time 1:07:38, Super wins.
4th match, Colossus white, 70 moves, Colossus time 3:32:50,
Sup.chess3.5 time 3:21:28, Colossus wins.
Summary:
Two expected 'sweeps', Cyrus2 use 250 % less time then obsolete Turk,
but destroys Turk. (silly) Overlook in match-3, makes things look
desperate for Colossus after 3 matches, but final score is real
relation. We eagerly waits to see was Super Chess 3.5 de-throned in
1986 by another challenger (Cyrus2)...
Who is the strongest 'player' among Speccy's chess programs ? ...
Round 2 comes...
DanSolo
> How to make ratings for CHESS GAMES (???) :-O
Hey, just to potentially start a flamewar and cause trouble in
general, why not try some Spectrum chess programs against some C64
ones? That'll sort the Kasparovs from the Shorts...
Brian Gaff
fetishist I think..
Brian
--
Brian Gaff - bri...@blueyonder.co.uk
Note:- In order to reduce spam, any email without 'Brian Gaff'
in the display name may be lost.
Blind user, so no pictures please!
"Alex_Lux" <ale...@writeme.com> wrote in message
news:0605b255-3e6a-4d12...@v25g2000yqk.googlegroups.com...
Chris Young
to MC DanSolo:
The obvious thing to do is a similar tournament for C64 chess games.
The winners from each tournament can then compete to find the GRAND
CHESS CHAMPION.
(I suggest that the results are never published if the Speccy loses)
Chris
--
+-------------------------------------------+
| Unsatisfactory Software - "because it is" |
| http://www.unsatisfactorysoftware.co.uk |
| Your Sinclair: A Celebration |
+- http://www.yoursinclair.co.uk -----------+
DISCLAIMER: I may be making all this stuff up again.
DanSolo
wrote:
> The obvious thing to do is a similar tournament for C64 chess games.
>
> The winners from each tournament can then compete to find the GRAND
> CHESS CHAMPION.
http://www.gamebase64.com/search.php?a=5&f=5&id=64&d=36
Argh, 56 of them. Maybe we can just try the more popular ones (more
popular is better, right? :-) ) and "send forth out champion"...
Chris Young
to MC DanSolo:
> On Feb 16, 8:57�pm, "Chris Young" <chris.use...@mail-filter.com>
Perhaps somebody from comp.sys.cbm can do the tournament for those,
that then minimises the time Speccy people are being infected with C64
crap :)
DanSolo
wrote:
> Perhaps somebody from comp.sys.cbm can do the tournament for those,
> that then minimises the time Speccy people are being infected with C64
> crap :)
I'm guessing there's only three real contenders on "our" side:
Colossus, Chesssmaster and Sargon. I'll round robin them at the
weekend...
Chris Young
to MC DanSolo:
> > Perhaps somebody from comp.sys.cbm can do the tournament for those,
> > that then minimises the time Speccy people are being infected with C64
> > crap :)
>
> I'm guessing there's only three real contenders on "our" side:
> Colossus, Chesssmaster and Sargon. I'll round robin them at the
> weekend...
Note that Super Chess 3.5 beat Colossus4 in the first round, Psi
Chess beat Chess Master 2000 in the qualifiers...
DanSolo
wrote:
> Note that Super Chess 3.5 beat Colossus4 in the first round, Psi
> Chess beat Chess Master 2000 in the qualifiers...
Hmm, who knows if the algorithms are ported exactly between machines
though...
Alex_Lux
> On Tue, 16 Feb 2010 08:06:24 -0800 (PST) da kidz on comp.sys.sinclair were rappin'
>
> The winners from each tournament can then compete to find the GRAND
> CHESS CHAMPION.
>
> (I suggest that the results are never published if the Speccy loses)
> Chris
My intention was (and still is) to do a such thing. 'Round robin' is
'a must', as results in later rounds will show, simply, "program A"
will beat "program B", while "program B" will do same to "program C"
and... you guess, "A" should 'demolish' "C" right? But no (NO!),
surprisingly "program C" will win "program A" :-)
So for GRAND champion, IMO, should be event with 2 best programs from
Speccy plus 2 from C64.
There are many other conclusions from tournament, e.g. some programs
are well written (balanced play as white and black), some plays better
as white, and some (surprisingly) as black! So playing one match in an
'duel' is waste of time ;-)
Now, it's time for Round 2...
Alex_Lux
In this round, derby "Who's the boss in CP Software stable?"...
plus another derby - "Who's the strongest in Cyrus family?"... and
last but not the least... derby of 'back part of Leaderboard'...
...
Psi - Turk:
1st match, Psi white, 46 moves, Psi time 2:19:29, Turk time 2:29:36,
Turk wins.
2nd match, Psi white, 50 moves, Psi time 2:43:40, Turk time 3:05:05,
Turk wins.
3rd match, Turk white, 109 moves, Psi time 9:18:25, Turk time 5:29:37,
draw! (marathon, almost 15 hours!)
4th match, Turk white, 58 moves, Psi time 2:14:36, Turk time 1:45:48,
Turk wins.
Clock'89 - Sup.Chess 3.5:
1st match, Clock'89 white, 47 moves, Clock'89 time 2:22:22, Sup.ch.3.5
time 2:21:50, Clock'89 wins.
2nd match, Clock'89 white, 57 moves, Clock'89 time 2:53:38, Sup.ch.3.5
time 2:45:35, draw!
3rd match, Sup.Ch.3.5 white, 57 moves, Clock'89 time 2:48:51, Sup.ch.
3.5 time 2:53:40, Clock'89 wins.
4th match, Sup.Ch.3.5 white, 53 moves, Clock'89 time 2:30:23, Sup.ch.
3.5 time 2:32:34, Sup.Ch.3.5 wins.
Cyrus2 - Cyrus1:
1st match, Cyrus2 white, 62 moves, Cyrus2 time 2:47:14, Cyrus1
time ???, Cyrus2 wins.
2nd match, Cyrus2 white, ?? moves, Cyrus2 time ???, Cyrus1 time ???,
Cyrus2 wins. (I can't find screenshots from this one)
3rd match, Cyrus1 white, 48 moves, Cyrus2 time 2:00:02, Cyrus1
time ???, draw!
4th match, Cyrus1 white, 47 moves, Cyrus2 time 2:02:07, Cyrus1
time ???, Cyrus2 wins.
As you see, Cyrus1 doesn't show its thinking time, so that's why "???"
appears always when Cyrus1 plays. In general, you can presume it's 10
% less then Cyrus2's time. (actually it's between 5% and 10% according
to my measurements, so 10% is conservative presumption).
Psi - Turk 0.5:3.5
Clock '89 - Sup.Chess 3.5 2.5:1.5
Cyrus 2 - Cyrus 1 3.5:0.5
Standings - after 2.round
=========
Cyrus2 ________ 7.5
Clock '89_______ 6.5
Super Ch.3.5____ 4
The Turk _______ 3.5
Colossus4.2 ____ 1.5 (-1 match)
Cyrus1_________ 0.5 (-1)
Psi Chess ______ 0.5
Summary:
Clock'89 isn't perfect, first loss of points, oldie Super Ch. made a
quite 'damage'. Cyrus1 manage to avoid 'sweep' from its successor.
Turk? What a dissapointment for "Edge"... obsolete Turk played 'cat &
mice' with Psi... is it 'Tour's highlight' for The Turk?
Stay tuned ;-)
Alex_Lux
some 'difficulties' with net connection at home, so I'll post "Round
3" as 'scheduled', today :-) i.e. right after "Round 2"... sorry for
my slow 'response'
Round 3:
======
Today on 'menu'... another strong challenge (Colossus) for
Clock'89? Another chance (Turk) for 'Super' to demonstrate its power,
and one more derby of 'back part of LB' i.e. would Cyrus1 'recover'
after heavy loss?
I'll start with... ABSOLUTE SHOCK !!!
Super Chess 3.5 - Turk:
1st match, Sup.Ch.3.5 white, 70 moves, Sup.ch. time 3:06:03, Turk time
4:23:48, Turk wins.
(note: after 53 moves, SC 3.5 time was 2:40:05, and it was in very bad
position, Turk was on 3:47:29)
2nd match, Sup.Ch.3.5 white, 38 moves, Sup.ch. time 1:55:02, Turk time
2:49:01, Turk wins. (Turk did it again!)
3rd match, Turk white, 50 moves, Sup.ch. time 2:20:25, Turk time
3:26:25, Sup.ch.3.5 wins.
4th match, Turk white, 65 moves, Sup.ch. time 3:14:25, Turk time
5:26:35, Sup.ch.3.5 wins.
Clock'89 - Colossus4.2:
1st match, Clock'89 white, 52 moves, Clock'89 time 2:05:58, Colossus4
time 2:53:14, Clock'89 wins.
2nd match, Clock'89 white, 58 moves, Clock'89 time 2:43:17, Colossus4
time 2:47:27, Clock'89 wins.
3rd match, Colossus white, 92 moves, Clock'89 time 4:35:53, Colossus4
time 4:57:18, draw!
4th match, Colossus white, 35 moves, Clock'89 time 1:40:46, Colossus4
time 1:26:40, Colossus4 wins.
Cyrus1 - Psi chess:
1st match, Cyrus1 white, 39 moves, Cyrus1 time ???, Psi time 3:50:41,
Cyrus1 wins.
2nd match, Cyrus1 white, 37 moves, Cyrus1 time ???, Psi time 3:47:14,
Cyrus1 wins.
3rd match, Psi Chess white, 30 moves, Cyrus1 time ???, Psi time
3:29:05, Cyrus1 wins.
4th match, Psi Chess white, 67 moves, Cyrus1 time ???, Psi time
7:41:23, Cyrus1 wins.
Super Chess 3.5 - Turk 2:2
Clock'89 - Colossus4.2 2.5:1.5
Cyrus1 - Psi chess 4:0
Standings - after 3.round
=========
Clock '89 _______ 9
Cyrus2_________ 7.5 (-1)
Super Ch.3.5____ 6
The Turk________ 5.5
Cyrus1_________ 4.5 (-1)
Colossus4______ 3 (-1)
Psi Chess______ 0.5
Summary:
Now it's obvious that Clock'89 is vulnerable as black, problem is in
attempting to play 'by book' too long, so after 7 or 8 moves Clock'89
(as black!) is 'in minus' position i.e. 'in chase', waiting for
opponents error... and smart programs doesn't grant it easily! After
1st Turk's win as black over Super Chess I was in shock, after 2nd, in
disbelief! Someone more competent for chess then me, should analize
it. Yep, I have RZX playbacks of both matches. In 1st, Turk made
'sacrifice' at start and gains a 'tempo', in 2nd, after 20 moves Turk
was 3 figures 'down' but gain position/tempo, so around move 25, Turk
gets 'quality' (Super lost both towers). Psi Chess is 'nailed down' on
bottom of Leaderboard after this round, and will stay there to the
final.
In the next round... "Battle of the Worlds" - CP software vs.
Cyruss'... part 1 - "Revenge" (could Super ch. 3.5 'revenge' Super ch.
2? Strike on Cyrus1!!)
Alex_Lux
Today... '1986 (programs) in action', Psi vs. Cyrus2, colossus4 vs.
oldie Turk. And... "Battle of the Worlds" when chess on Speccy is in
question, CP software vs. Cyruss'... is Super Chess 3.5 capable to
take crown from Cyrus1 (in 1985) and avenge Super Chess 2's debacle?
Enjoy!
Turk - Colossus 4.2:
1st match, Turk white, 31 moves, Turk time 3:28:59, Colossus4 time
1:47:03, Colossus4 wins.
2nd match, Turk white, 41 moves, Turk time 2:53:24, Colossus4 time
2:19:10, Colossus4 wins.
3rd match, Colossus4 white, 36 moves, Turk time 3:15:39, Colossus4
time 1:59:10, Colossus4 wins.
4th match, Colossus4 white, 42 moves, Turk time 2:55:30, Colossus4
time 2:12:52, Colossus4 wins.
Psi - Cyrus2:
1st match, Psi white, 28 moves, Psi time 1:34:30, Cyrus2 time 0:54:28,
Cyrus2 wins.
2nd match, Psi white, ?? moves, Psi time ???, Cyrus2 time ???, Cyrus2
wins.
3rd match, Cyrus2 white, 44 moves, Psi time 2:41:49, Cyrus2 time
1:45:16, Cyrus2 wins.
4th match, Cyrus2 white, ?? moves, Psi time ???, Cyrus2 time ???,
Cyrus2 wins. (I can't find screenshots from this one)
Super chess 3.5 - Cyrus1:
1st match, Sup.Ch.3.5 white, 43 moves, Sup.ch. time 2:10:56, Cyrus1
time ???, draw!
2nd match, Sup.Ch.3.5 white, 73 moves, Sup.ch. time 3:39:40, Cyrus1
time ???, Sup.ch.3.5 wins.
3rd match, Cyrus1 white, 79 moves, Sup.ch. time 3:47:44, Cyrus1
time ???, Sup.ch.3.5 wins.
4th match, Cyrus1 white, 48 moves, Sup.ch. time 2:14:51, Cyrus1
time ???, Sup.ch.3.5 wins.
Turk - Colossus 4.2 0:4
Psi - Cyrus2 0:4
Super chess 3.5 - Cyrus1 3.5:0.5
Standings - after 4.round
=========
Cyrus2 _______ 11.5
Super Ch.3.5 ___ 9.5 (+1 match)
Clock '89_______ 9
Colossus 4.2 ___ 7
The Turk _______ 5.5 (+1)
Cyrus1_________ 5
Psi Chess_______0.5 (+1)
Summary:
Now we see that Super Chess 3.5 was written 'to kill' Cyrus1... after
calvary with Turk, 'Super' achieved convincing victory over Cyrus1
which isn't logical, as Cyrus1 is much stronger program then Turk!
Things be more obvious after Turk-cyrus1 match Super Chess 3.5 has 2
very hard matches (70+ moves) with Cyrus, used more time to think [1],
and few times 'Super' changed its move to better one 'in last
moment' [2]!!! Both Cyrus programs tends 'to make important' and
respond very, very fast... that's charming feature for human opponent
but... Other 2 duels are not worthy of comment.
[1] it's hard to make 100% equal times for all programs, especially
with Cyrus1.
[2] Super Chess 3.5, same as Colossus4 and Clock'89, shows 'flow of
thought'
In the next round... "Battle of the Worlds - Part 2 - "Cyrus Family
Strikes Back!", Cyrus2 vs. Super Chess 3.5. Then, should we see
routine victory for Colossus4 (after 'heroic' resistance to Clock'89)
vs. cyrus1. Round close Turk vs. Clock'89, maybe new upset for CP
Software by Turk?
Alex_Lux
'Today' we expect 2 duels where old, inferior chess programs have
little or no chance vs. their opponents (Turk-Clock'89 & Cyrus1-
Colossus4.2) and derby - in 1986, it was duel for 'Champion's Belt' in
category of chess programs on ZX Spectrum - defender: Super Chess 3.5
- challenger: Cyrus 2.
Turk - Clock'89:
1st match, Turk white, 57 moves, Turk time 3:47:36, Clock'89 time
2:38:36, Clock'89 wins.
2nd match, Turk white, 75 moves, Turk time 4:45:53, Clock'89 time
3:42:29, draw!
3rd match, Clock'89 white, 33 moves, Turk time 2:16:24, Clock'89 time
1:30:38, Clock'89 wins.
4th match, Clock'89 white, 59 moves, Turk time 3:49:29, Clock'89 time
2:49:32, Clock'89 wins.
Colossus 4.2 - Cyrus1:
1st match, Colossus white, 59 moves, Colossus4 time 2:52:02, Cyrus1
time ???, Colossus4 wins.
2nd match, Colossus white, 96 moves, Colossus4 time 4:58:35, Cyrus1
time ???, Cyrus1 wins. !!! Bravo!
3rd match, Cyrus1 white, 77 moves, Colossus4 time 4:01:06, Cyrus1
time ???, Cyrus1 wins.
4th match, Cyrus1 white, 42 moves, Colossus4 time 2:21:04, Cyrus1
time ???, draw!
And now, match for 'world Title'...
Cyrus2 - Super chess 3.5:
1st match, Cyrus2 white, 52 moves, Cyrus2 time 2:20:36, Sup.Chess 3.5
time 2:32:45, Cyrus2 wins.
2nd match, Cyrus2 white, 61 moves, Cyrus2 time 2:25:45, Sup.Chess 3.5
time 2:56:08, Cyrus2 wins.
3rd match, Sup.Chess 3.5 white, 45 moves, Cyrus2 time 1:24:39,
Sup.Chess 3.5 time 2:21:29, Cyrus2 wins.
4th match, Sup.Chess 3.5 white, 78 moves, Cyrus2 time 2:39:42,
Sup.Chess 3.5 time 3:34:15, Cyrus2 wins.
Hmm... New comparsion for "Destroyed": destroyed, more destroyed, mega-
destroyed, super-mega-giga-destroyed, destroyed like cyrus2 destroys
sup.chess 3.5 :-)
Turk - clock'89 0.5:3.5
colossus 4.2 - Cyrus1 1.5:2.5
Cyrus2 - Sup.ch.3.5 4:0
Standings - after 5.round
=========
Cyrus2_________ 15.5
Clock '89________12.5
Super Ch.3.5_____ 9.5 (+1 match)
Colossus 4.2_____ 8.5
Cyrus1__________ 7.5
The Turk_________ 6 (+1)
Psi Chess________0.5
Summary:
Is it all over? Is Cyrus2 already 'the Winner'? 3 points is a huge
gap... Is battle for 3th position only 'unknown' left on Tour? No more
'magic' for Turk vs. CP Software products, but, SURPRICE happened
thanks to Cyrus1. Remember that cyrus1 has the fastest response time
on this tournament! My story that developers of chess programs write
their products 'to kill' enemy no.1 in their mind, gains new
support... Cyrus2 in 1986 became undisputed champion when chess on
Speccy is in question!
In the next round... 'silence before storm'? Cyrus-Clock, Cyrus2-
Colos4, Super-Psi.
Daniel Mandic
> Speccy is in question!
jebemtiboga
You can't say Czrus II wins over Chess Master Three point Five.
(get over it, one more time, ED)
--
Daniel Mandic
Alex_Lux
Cyrus1 will seek its chance vs. Clock'89 as white... 6 years is
software design is, I guess like 30 years in car design... imagine
such race... poor Cyrus1. Colossus4 lost from 'Super', Cyrus2 killed
'Super', so one more boring match in front of us? 3rd pair are Super
Chess and Psi... if Psi win today on one board, I'll jump through my
window (on 6th floor)! ;-) Boring 'day'? It seems... Yawn! ZZZzzz...
Cyrus2 - Colossus 4.2:
1st match, Cyrus2 white, 60 moves, Cyrus2 time 2:27:42, Colossus 4.2
time 2:57:21, Cyrus2 wins.
2nd match, Cyrus2 white, 85 moves, Cyrus2 time 3:12:40, Colossus 4.2
time 4:34:23, draw!
3rd match, Colossus4 white, 42 moves, Cyrus2 time 1:34:35, Colossus
4.2 time 2:05:14, Colossus4.2 wins.
4th match, Colossus4 white, 70 moves, Cyrus2 time 2:21:23, Colossus
4.2 time 3:37:27, draw!
Cyrus1 - Clock'89:
1st match, Cyrus1 white, 38 moves, Cyrus1 time ???, Clock'89 time
1:32:21, Clock'89 wins.
2nd match, Cyrus1 white, 36 moves, Cyrus1 time ???, Clock'89 time
1:41:00, Cyrus1 wins. Hats down for 'old boy'!!
3rd match, Clock'89 white, 64 moves, Cyrus1 time ???, Clock'89 time
3:09:06, Clock'89 wins.
4th match, Clock'89 white, 55 moves, Cyrus1 time ???, Clock'89 time
2:35:00, Clock'89 wins.
Super Chess 3.5 - Psi Chess:
1st match, Sup.Chess 3.5 white, 20 moves, Sup.Chess 3.5 time 0:50:28,
Psi time 1:56:15, Sup.ch.3.5 wins. (express)
2nd match, Sup.Chess 3.5 white, ?? moves, Sup.Chess 3.5 time ???, Psi
time ???, Sup.ch.3.5 wins.
3rd match, Psi ch. white, 33 moves, Sup.Chess 3.5 time 1:45:38, Psi
time 1:36:11, Sup.ch.3.5 wins.
4th match, Psi ch. white, 54 moves, Sup.Chess 3.5 time 2:34:29, Psi
time 3:14:20, Sup.ch.3.5 wins.
Cyrus2 - Colossus 4.2 2:2
Cyrus1 - Clock'89 1:3
Sup.Ch.3.5 - Psi 4:0
Standings - after 6.round
=========
Cyrus2_________ 17.5
Clock '89 ______15.5
Super Ch.3.5___ 13.5 (+1 match)
Colossus 4.2 ___10.5
Cyrus1__________ 8.5
The Turk _______ 6
Psi Chess_______ 0.5
Summary:
In a momment, it looked that Cyrus2's advantage to Clock'89 will be
melted down, but Cyrus1 with its 'miracle' saved own successor. It's
noticeable, Cyrus2 'thinks' 25 to 30 % less then Colossus 4.2, so
setting Colossus to 160 or 165 seconds (instead 180) probably be more
fair... in only similar Tour (4 years ago), Cyrus2 was set one level
higher, while Colossus4 had the same setting as now. Yep, my intention
was to make hard challenge to Cyrus2 and proove is it a true
'champion' here. OK, back Cy.1-Clk89, light, small 'overlook' in 18th
move for Clk.89 plus bright 'sacrifice' in 32th move by Cy.1, made
"Cyrus1 highlight of Tour" :-)
In final round, we'll see battle of oldies (Turk-Cyrus1), Psi's
attempt to save 'bronze' medal for Super Chess 3.5 (Colossus needs 3.5
points) and the duel we wait, the duel which we missed on previous
tournament... CHALLENGER - Clock'89 (program which beats 'everybody'
with 10:0 or 9:1) vs. TITLE DEFENDER - Cyrus2 ... taA-dAAA !!
Alex_Lux
Final Round:
Clock'89 needs win of 3:1 over Cyrus2 in order to win Tournament.
What we know from previous rounds:
- Clock'89 is vulnerable as white, Cyrus2 plays excellent in 'both
colours' (--> clear 2 points for Cyrus2 ?!?);
- CP Software knows 'to do their job', Super Chess 3.5 proove that,
after 2 years of domination among chess programs, Cyrus1 was swept
from throne...;
- CP Software now had 3 years to prepare answer on Cyrus2... (--> in
this light, we don't see "clear 2 points for Cyrus2" anymore !) :-)
As 'supporting roles' we have duels, for 5th position, Cyrus1-
Turk, and Psi-Colossus4.
Cyrus - Turk:
1st match, Cyrus1 white, 56 moves, Cyrus1 time ???, Turk time 2:55:56,
Cyrus1 wins.
2nd match, Cyrus1 white, 56 moves, Cyrus1 time ???, Turk time 4:27:47,
Cyrus1 wins.
3rd match, Turk white, 55 moves, Cyrus1 time ???, Turk time 3:00:20,
Cyrus1 wins.
4th match, Turk white, 78 moves, Cyrus1 time ???, Turk time 5:34:49,
draw!
Psi Chess - Colossus 4.2:
1st match, Psi ch. white, 49 moves, Psi time 3:58:58, Colossus 4.2
time 2:39:50, Colossus 4 wins.
2nd match, Psi ch. white, 17 moves, Psi time 1:10:33, Colossus 4.2
time 0:43:26, Colossus 4 wins. (express!)
3rd match, Colossus 4 white, 33 moves, Psi time 2:42:08, Colossus 4.2
time 1:53:16, Colossus 4 wins.
4th match, Colossus 4 white, 60 moves, Psi time 2:53:14, Colossus 4.2
time 4:53:25, Colossus 4 wins.
Clock'89 - Cyrus2:
1st match, Clock'89 white, 66 moves, Clock'89 time 3:18:34, Cyrus2
time 2:37:39, draw!
2nd match, Clock'89 white, 53 moves, Clock'89 time 2:37:00, Cyrus2
time 2:01:04, Clock'89 wins.
3rd match, Cyrus2 white, 61 moves, Clock'89 time 2:54:15, Cyrus2 time
2:33:23, Cyrus2 wins.
4th match, Cyrus2 white, 92 moves, Clock'89 time 4:36:51, Cyrus2 time
3:54:30, draw!
Cyrus - The Turk 3.5:0.5
Psi - Colossus 4 0:4
Clock'89 - Cyrus2 2:2
FINAL LEADERBOARD
================= TOTAL score (score as "Duel"
_____________________________________'white') W-D-L
1. Cyrus 2___________19.5 : 4.5_________(11:1) 4-2-0
2. Clock Chess'89____ 17.5 : 6.5_________(11:1) 5-1-0
3. Colossus 4.2______ 14.5 : 9.5__________(9:3) 2-1-3
4. Super Chess 3.5___ 13.5 : 10.5_________ (6:6) !! 3-1-2
5. Cyrus IS___________ 12 : 12__________(7:5) 3-0-3
6. The Turk___________ 6.5 : 17.5_______ (2.5:9.5) 1-1-4
7. Psi Chess__________0.5 : 23.5_________(0:12) 0-0-6
Summary:
Who's better? If you're fan of Clock'89, fact is Clock'89 showed
as 'stronger' in duel with Cyrus2, in long run (e.g. 10 matches)
Clock'89 would win (6:4 or 5.5:4.5 ?) as we had (when Clock played
white) clear, dominant win for Clock and draw where Cyrus2 barely
'survive' - serious material deficit with giving check eternally. When
playing black, Clock'89 in one match blocked every attempt of Cyrus2
(Clock was in 'safe zone' whole match), in other, Cyrus2 won but not
dominant, in manner of Clock'89.
On the other side, Cyrus2 won the most points, didn't make
'mistakes' with weaker opponents, and use less time (except Cyrus1),
in duel with Clock'89 is was average 20% less 'thinking time' as black
and 14% as white. Enough arguments for me to declare winner!
Super Chess show dissapointing performance as white! Only 6:6!
Very inovative, 1st chess program with 'flow of thought'!
Colossus 4.2 lost (minimal!) both from Super Ch.3.5 and Cyrus1,
but outrun them on leaderboard! It shows as fine, balanced (black/
white) product with many options. All these with 3D graphics under
48kB... maybe the best choice for customers!
Cyrus1, nice surprise, playing equaly with 'modern' programs,
briliant 'ballance'.
Turk, first nice looking program, with many options, made some
dammage on Tour, too ;-)
Psi, is better then Chessmaster 2000 and Master Chess
(Mastertronic), his role on Tour was to show us how good Cyrus1 was
(and still is) ;-)
In next message, maybe more statistics, maybe link to page with RZX
recordings from this tournament, and board screenshots, and definitly
some ideas "What to do next?" (with chess programs).
OwenBot
> In next message, maybe more statistics, maybe link to page with RZX
> recordings from this tournament, and board screenshots, and definitly
> some ideas "What to do next?" (with chess programs).
Play off Cyrus2 against GNU chess on a modern OS with both at maximum
strength to see if modern Chess programs are any good. :)
Chris Young
to MC OwenBot:
and DanSolo's C64 chess competition idea...
Alex_Lux
If we focus on 4 'main chess programs' (on Speccy), we have these
tournament leaderboard:
1. Cyrus 2 ---------- 8 pts.
2. Clock '89 -------- 7 pts.
3. Colossus 4.2 ----- 5 pts.
4. Super Chess 3.5 -- 4 pts.
In 12 matches, Sup.Chess 3.5 took 715 moves, and used 34 hours, 44
minutes and 6 seconds (34:44:06), Colossus 4.2 took 759 moves with
40:42:33 time, Clock '89 made 723 moves with time 35:07:48, Cyrus 2
played 765 moves with agregate time 29:33:38, more then 5 hours less
then 2nd fastest program.
Average response time per move, in minutes are:
Colossus 4.2 = 3.22;
Clock '89 = 2.92;
Sup.Ch.3.5 = 2.91;
Cyrus 2 = 2.32
Remember, except Cyrus2, all programs were set on 180 seconds
response! One more argument in favor of Cyrus 2 :-)
Importance of playing such a exhaustive tournament can be seen if we
pick just Cyrus 1, Sup.Ch.3.5 and Colossus 4.2, we've already seen
that last one was 3rd in general LB, and also 3rd in 'Top4 head-to-
head LB'... Cyrus 1 isn't member of Top4, right? So, Colossus 4.2
should be 1st in this company?
Lets remind results: 'super'-'colos4' 2.5:1.5; 'super'-cyrus1 3.5:0.5;
and 'colos4'-cyrus1 1.5:2.5! So, table be: 1. Super Chess 3.5 -- 6
pts,
2. Cyrus 1 ---------- 3 pts,
3. Colossus 4.2 ----- 3 pts. Hahaha, Colossus 4.2 is 3rd again!
Some 'conclusions':
* Tournament show that "Round Robin" is a must if we want solid, true
results. If "program A" beats "priogram B", and "program B" do the
same to "program C", that doesn't necessarily means "program A" will
beat "program C" !!!
* Some programs are written to play better as black (e.g. Turk and
Super Chess 3.5), some as white, but 'winners' are well 'balanced' :-)
this shows that playing both sides is a must;
* In many matches on this Tour, we had different outcomes when 2
programs played more then one match on same sides - e.g. in 2 matches
among Super Ch.3.5 and Clock'89, when 'Super' was white, once winner
was Clock'89 and once 'Super' !!!
Now, in all these messages, I think you have a lot of facts, about
Speccy's chess programs, so you can pull out your own conclusions and
choices. My work about that topic finishes here.
Alex_Lux
> On Thu, 4 Mar 2010 07:02:27 -0800 (PST) da kidz on comp.sys.sinclair were rappin'
> to MC OwenBot:
>
> > strength to see if modern Chess programs are any good. :)
>
> and DanSolo's C64 chess competition idea...
What to do next?
a) add 1 or even 2 programs to Tournament - Spectrum chess 2 or Chess
Player were 'close' to qualify, Chess (Psion) or 16k Super Chess can
count on invitation from honour... IMO, many matches, nothing
happens... important. Only it be more obvious how good (for its time)
Cyrus 1 was :-) it won't be anymore 12 of 24 score, it be 16 of 28 or
20 of 32!
b) try 'funny' stuff - to try matches with same program on different
levels, e.g. Cyrus2 on level 8 vs. Cyrus2 on level 9 :-) Hmm, 16kB
Cyrus1 vs. its 48kB incarnation (on same level!) fits here too! ;-)
c) try different kind of program settings - this Tournament I call
'optimal' i.e. it was 3 min/move, we can play 'burst-mode' with 30 or
45 seconds per move, and 'marathon' settings - more then 5 min/move!
Both variants has their interesting bits... e.g. 'burst' will show us
how 'smart' algorithms are!
d) the most interesting one IMO, play similar tournament with Comodore
64 programs, pick best two and challenge them with 'best 2' on Speccy,
Cyrus2 and Clock'89 !!! Yep, 4 matches, round robin :-)
After reading all above posts, if you're inspired to run similar
tournament on your own, let us know about it :-) As you see, there's
still left a lot of work to be done.
It's time to close this thread, I hope that you enjoyed in reading it!
Have a nice day! Bye!
link to RZX recordings, screenshots etc. you'll find on:
****** http://alexlux.orgfree.com/chess-tour/index.htm *******
Daniel Mandic
> In next message, maybe more statistics, maybe link to page with RZX
> recordings from this tournament, and board screenshots, and definitly
> some ideas "What to do next?" (with chess programs).
Hi!
Firstly, thanks a lot for the Chess Computer Tournament!
It seems that Cyrus II is unbeatable. Your recommendation over Colossus
Chess IV sounds good to me (my fav).
What to do next!??
I have heard that actual PC Chess Software plays very strong.
A tournament vs. Speccy Chess would be interesting for me.
Let's say with some advanteges for the Speccy, as "infinite" time for
example (well, e.g. 14h speccy time vs. 5min PC 1GHz....)
Kind regards
--
Daniel Mandic
Daniel Mandic
> good idea (Danijele, hvala Ti na dobrom komentaru, i finoj ideji !!),
Nema na cemu, .....hvala i tebi!
> not necessarily this dis-balance in thinking time... I guess Cyrus2 on
> 180s (avg. per move) be good opponent to 30s PC Chess. Actually, I'd
> go for 3 kind of matches, both progs on 30s, then both 'on optimal'
> level (180s), and finally 'handicap' for PC... (like Daniel said!)
> e.g. PC on 30s and Clock'89 on 500s :-)
I have started a couple of games with Cyrus II vs. Arasan. Interesting
was that Cyrus have had a quiet better time by playing black.
--
Daniel Mandic
DanSolo
> and DanSolo's C64 chess competition idea...
Which sadly I've had no time to select "our" champion for. Luckily
someone else has already done this!
http://www.spacious-mind.com/html/commodore_8_bit_open_chess_cha.html
Cyrus II, by Alligata software from 1986 looks like the pick of the
bunch.
I'll drop him a mail to let him know about the goings on here.
DanSolo
> On Mar 4, 5:09 pm, "Chris Young" <chris.use...@mail-filter.com> wrote:
>
> > and DanSolo's C64 chess competition idea...
>
> Which sadly I've had no time to select "our" champion for. Luckily
> Cyrus II, by Alligata software from 1986 looks like the pick of the
> bunch.
> I'll drop him a mail to let him know about the goings on here.
Of course, this may not be quite so interesting as the same program,
with most likely the same algorithms, is the winner. Then again it
might be *more* interesting for us as it's more C64 vs ZX than
anything else then. Hehe.
DanSolo
***********************************************************************************************************************
To compare Spectrum against C64 is not exactly fair because of the
different speeds. C64 1.02 MHz (NTSC) and Spectrum 3.6 Mhz which
means the Spectrum Programs get to calculate 3.6 times deeper /faster
into positions. Same as comparing ST or Amiga against Spectrum. They
would be twice as fast as the Spectrum therefore the Amiga and ST
programms will more likely be better.
Overall C64 has better programs than the Spectrum for Chess. Spectrum
only has Clockchess from Chris Whittington, Colossus 4 (Martin
Bryant) and Cyrus II (Richard Lang) and possibly Superchess 3.5
(Chris Whittington) which are decent chess games. Chris Whittington
wrote all the Superchess programs I think but he obviously improved
with later versions like Clockchess.
The C64 has Mychess II (David Kittinger), Colossus 4 (Bryant),
Chessmaster 2100 (Spracklen) Sargon III (Spracklen), Cyrus II (Lang),
Chessmaster 2000 (Kittinger), Chess 7.5 (Larry Atkin) and Chess 7.0
(Larry Atkin) These are all great programs as well as Coffeehouse
Chess Monster (Julio Kaplan)
The chess programmers that I am showing above were all from brilliant
chess programmers in their day. For example Dan & Kathe Spracklen 4
times World Chess Computer champions. Richard Lang 7 times World
Champion with his programs that are also in dedicated chess
computers. Larry Atkin 3 or 4 times world champion.
As you can see above C64 has the better programs but Colossus 4 on C64
will lose to Colossus 4 on Spectrum because of the speed difference as
well as a 6502 processor is a little different to a z80 processor.
Try playing Colossus 4 at say 30 seconds per move or 60 moves in 30
seconds (same thing) against each other on both machines.
Then play somer games with Spectrum at 30 seconds per move and C64 at
108 seconds per move (3.6 times factor for speed difference) and you
should find that the games between the machines are really close.
Best regards
Nick
***********************************************************************************************************************
As I suspected, it's not the computer it's the programmer that makes
the it a good chess machine. A lot like other games I guess. Cyrus II
it is.
Fred
> To compare Spectrum against C64 is not exactly fair because of the
> different speeds. C64 1.02 MHz (NTSC) and Spectrum 3.6 Mhz which
> means the Spectrum Programs get to calculate 3.6 times deeper /faster
> into positions.
Whenever the csc64 and css flamewars have descended to technical
details, I believe the C64 bods have claimed that due to reduced cycle
counts etc., that the 6502 is not that much slower than the Z80. They
are either wrong or Nick above is wrong.
Fred