Issue 61 4 August, 1997
______________________________________________________________________________
Core Warrior is a newsletter promoting the game of corewar. Emphasis is
placed on the most active hills--currently the '94 draft hill and the beginner
hill. Coverage will follow where ever the action is. If you have no clue what
I'm talking about then check out these five-star internet locals for more
information:
FAQs are available by anonymous FTP from rtfm.mit.edu as
pub/usenet/news.answers/games/corewar-faq.Z
FTP site is: ftp.csua.berkeley.edu /pub/corewar
Web pages are at: (Please note Stormking's new address)
http://www.koth.org/ ;Stormking
http://www.ecst.csuchico.edu/~pizza/koth/ ;Pizza
http://pauillac.inria.fr/~doligez/corewar/ ;Planar
Newbies should check the stormking page for the FAQ, language specification,
guides, and tutorials. Post questions to rec.games.corewar. All new players
are infinitely welcome!
If ftp.csua.berkeley.edu is unreachable, you can download pMARS at:
Terry's web page--http://www.nc5.infi.net/~wtnewton/corewar/
Planar ftp site--ftp://ftp.inria.fr/INRIA/Projects/para/doligez/cw/pmars
Fechter ftp site--ftp://members.aol.com/ofechner/corewar
A collection of Bezzi's hints in the first issues is available at:
ftp://ftp.volftp.vol.it/pub/pc/msdos/games/solutions/bbhints.zip
Beppe Bezzi web page - http://www.aspide.it/freeweb/Bezzi
______________________________________________________________________________
Welcome to my first issue of Core Warrior, following Christian's two weeks
back. I hope at least someone finds it vaguely useful :-)
The Hills are still quite, with very little action. Is everyone just off on
their summer/winter break, or have we exhasted the possibilities of the '94
Draft? It was a year ago last Tuesday that Anton Marsden published Probe,
with its Q^2 scan, which is still seen on many of the warriors on the Hill. Who
will come with something similar this year?
STOP PRESS -- First and exclusive: the results of the second round of Anton's
Core War Tournament -- see below -- STOP PRESS ENDS
Phil Kendall
______________________________________________________________________________
Anton's Tournament Round 2 Results
The aim of round 2 was to write a warrior designed to beat Electric Head,
currently on top of the '94 Hill. (See _Core Warrior 59_ for the source)
Rank Name Author Score Points
_________________________________________________________________
1 Shelter 1f3 Ilmari Karonen 2986 7
2 Head Basher Brian Haskin 2940 6
3 PBeat v1.0 Philipp Offermann 2903 5
4 Short Circuit Philip Kendall 2875 4
5 P^(1/2) Christian Schmidt 2389 3
6 Nautilus Mole IX 2098 2
7 Anti Head Ian Oversby 1533 1
Congratulations to Ilmari Karonen for dropping only 14 points over the 1000
round fight.
_______________________________________________________________________
Current Status of the Internet Pizza Server ICWS '94 Draft Hill:
Hill Specs:
coresize: 8000
max. processes: 8000
duration: after 80,000 cycles, a tie is declared.
max. entry length: 100
minimum distance: 100
rounds fought: 200
instruction set: ICWS '94 Draft
The current ICWS '94 Draft hill:
# %W / %L / %T Name Author Score Age
1 43.5/ 38.3/ 18.3 Electric Head Anton Marsden 148.7 24
2 45.6/ 45.6/ 8.9 He Scans Again P.Kline 145.6 12
3 35.1/ 25.1/ 39.9 Newt Ian Oversby 145.1 15
4 30.5/ 19.0/ 50.5 Return Of The Jedimp John K W 142.0 269
5 33.8/ 26.8/ 39.4 RetroQ P.Kline 140.9 69
6 35.2/ 29.6/ 35.2 Head or Tail Christian Schmidt 140.8 57
7 35.5/ 30.8/ 33.7 Nine Seven Six M R Bremer 140.2 133
8 41.6/ 43.3/ 15.1 C^2 Christian Schmidt 139.9 19
9 37.4/ 36.2/ 26.4 PAN-TAU-RA Christian Schmidt 138.6 4
10 34.1/ 30.9/ 35.0 Ultraviolet Ken Espiritu 137.3 6
11 40.2/ 43.8/ 16.0 Blain Nimon 136.6 35
12 31.2/ 26.6/ 42.2 unrequited love kafka 135.8 16
13 36.7/ 38.0/ 25.3 First Strike Nimon 135.5 30
14 31.8/ 28.3/ 39.9 Ultra-RQ^2 Ken Espiritu 135.3 7
15 38.0/ 41.3/ 20.7 NCC-1701-A Philip Kendall 134.8 72
16 37.7/ 41.0/ 21.3 Guessed v0.1 Ian 134.4 9
17 26.5/ 18.9/ 54.6 Impish v0.2 Ian Oversby 134.1 308
18 37.2/ 40.7/ 22.1 Damage Incorporated Anton Marsden 133.6 310
19 40.1/ 47.0/ 12.9 Overdrive Christian Schmidt 133.1 1
20 30.9/ 30.1/ 39.0 Gigolo Core Warrior staff 131.6 323
21 36.8/ 42.0/ 21.2 Solomon v0.8 Ian Oversby 131.5 115
22 36.4/ 42.1/ 21.5 Guru Robert Hale 130.7 2
23 31.3/ 32.3/ 36.5 Terkonit 3 Christian Schmidt 130.2 12
24 37.4/ 45.7/ 16.9 Fawkes v0.2 Ian Oversby 129.2 7
25 37.6/ 46.5/ 15.9 myVamp5.4 Paulsson 128.7 19
Age since last issue: 2 ( 2 last issue, 13 the issue before )
New warriors: 2 Turnover/age rate 100%
Average age: 74 ( 74 last issue, 72 the issue before )
Average score: 142 ( 139 last issue, 142 the issue before )
The top 25 warriors are represented by 13 independent authors: Schmidt with 6;
Oversby with 5; Espiritu, Kline, Marsden and Nimon with 2. All others with one
warrior each.
______________________________________________________________________________
94 - What's New
# %W / %L / %T Name Author Score Age
19 36.9/ 41.9/ 21.2 Guru Robert Hale 131.9 1
19 40.1/ 47.0/ 12.9 Overdrive Christian Schmidt 133.1 1
We welcome Robert Hale to the '94 Hill.
______________________________________________________________________________
94 - What's No More
# %W / %L / %T Name Author Score Age
26 11.4/ 45.4/ 43.1 Pure Genius Compudemon 77.4 0
26 40.3/ 52.5/ 7.1 Obsidian Nimon 128.2 29
______________________________________________________________________________
94 - What's Old
# %W / %L / %T Name Author Score Age
20 30.9/ 30.1/ 39.0 Gigolo Core Warrior staff 131.6 323
18 37.2/ 40.7/ 22.1 Damage Incorporated Anton Marsden 133.6 310
17 26.5/ 18.9/ 54.6 Impish v0.2 Ian Oversby 134.1 308
4 30.5/ 19.0/ 50.5 Return Of The Jedimp John K W 142.0 269
The top three here have all looked in trouble in recent weeks, never leaving
the lower reaches of the Hill. Jedimp, on the other hand, goes from strength
to strength.
______________________________________________________________________________
OLD HALL OF FAME
* means the warrior is still active.
Pos Name Author Age Strategy
1 Thermite II Robert Macrae 2262 Qscan -> bomber
2 Impfinity v4g1 Planar 1993 Stone/ imp
3 Jack in the box Beppe Bezzi 1620 P-warrior
4 Tornado 3.0 Beppe Bezzi 1567 Bomber
5 Torch t18 P.Kline 1539 Bomber
6 Chameleon Myer R Bremer 1437 P-warrior
7 Frontwards v2 Steven Morrell 1420 One shot scanner
8 Evol Cap 6.6 John Wilkinson 1299 Imp / stone
9 quiz Schitzo 1262 Scanner/ bomber
10 T.N.T. Maurizio Vittuari 1204 Bomber
11 Grilled Octopus v0.5 David Boeren 1154 P-warrior
12 Hazy Shade II John Wilkinson 1102 P-warrior
13 Stepping Stone Kurt Franke 1049 Qscan -> Vampire
14 Rosebud Beppe Bezzi 993 Stone/ imp
15 Iron Gate 1.5 Wayne Sheppard 926 CMP scanner
16 T.N.T. pro Maurizio Vittuari 925 Bomber
17 Agony II Stefan Strack 912 CMP scanner
18 Barrage Anton Marsden 876 Qscan -> replicator
19 Blue Funk Steven Morrell 869 Stone/ imp
20 Flurry Anton Marsden 835 Qscan -> pwarrior
21 Thermite 1.0 Robert Macrae 802 Qscan -> bomber
22 Blue Funk 3 Steven Morrell 766 Stone/ imp
23 Night Train Karl Lewin 755 Replicator
24 Mirage 1.5 Anton Marsden 736 Scanner/ bomber
25 Blizzard Anton Marsden 713 Qscan -> replicator
______________________________________________________________________________
NEW HALL OF FAME
* means the warrior is still active.
Pos Name Author Age Strategy
1 Probe Anton Marsden 403 Q^2 -> Bomber
2 Blur 2 Anton Marsden 396 Scanner
3 unrequited love kafka 346 Q^2 -> Paper
4 Gigolo Core Warrior staff 323 * Q^2 -> Stone/imp
5 Damage Incorporated Anton Marsden 310 * Q^2 -> Bomber
6 Impish v0.2 Ian Oversby 308 * Stone/imp
7 Falcon v0.3 Ian Oversby 275 P-warrior
8 Return Of The Jedimp John K W 269 * Q^2 -> Stone/imp
9 Rosebud Beppe 218 Stone/imp
10 Q^2 Miro Anders Ivner 214 Q^2 -> Scanner/bomber
11 Instant Wolf 3.4 Edgar 205 P-warrior
12 Goldfinch P.Kline 201 P-warrior
13 Simple v0.4b Ian Oversby 197 QScan -> Stone/imp
14 Trident^2 John K W 195 Q^2 -> Stone/imp
15 ompega Steven Morrell 189 Stone/imp
16 Frogz Franz 172 Q^2 -> Paper
17 The Machine Anton Marsden 164 Scanner
18 Memories Beppe 152 Scanner
19 Nine Seven Six M R Bremer 133 * Q^2 -> Stone/imp
20 Tiberius 3.1 Franz 130 Q^2 -> Paper
21 Solomon v0.8 Ian Oversby 115 * Stone and scanner
22 CC Paper 3.3 Franz 107 Q^2 -> Paper
23 mrb-test m r bremer 106 ?
24 T.N.T. pro Maurizio Vittuari 105 Bomber
25 Jack in the box II Beppe Bezzi 100 P-warrior
______________________________________________________________________________
Current Status of the Internet Pizza Server Beginner's Hill:
Hill Specs:
coresize: 8000
max. processes: 8000
duration: after 80,000 cycles, a tie is declared.
max. entry length: 100
minimum distance: 100
maximum age: At age 100, warriors are retired.
rounds fought: 200
instruction set: ICWS '94 Draft
Last challenge: Sat Aug 2 20:56:55 PDT 1997
# %W / %L / %T Name Author Score Age
1 51.6/ 37.1/ 11.4 Pentagram 2.a J.A.Denny 166.1 94
2 47.2/ 33.9/ 18.9 Hexagram J.A.Denny 160.5 92
3 47.7/ 38.7/ 13.6 Guessed v0.1 Ian 156.8 32
4 45.6/ 37.0/ 17.5 guru v0.1 Robert Hale 154.1 8
5 39.9/ 25.7/ 34.5 Sections Zul Nadzri 154.1 52
6 45.8/ 40.9/ 13.3 Nautilus Mole IX 150.6 11
7 44.4/ 41.9/ 13.7 qDeath v .169 Ryan Coleman 146.8 39
8 41.5/ 38.4/ 20.1 qbomber v .469 Ryan Coleman 144.5 43
9 32.4/ 20.7/ 46.9 Dust 7.0 Justin Kao 144.1 72
10 34.1/ 24.4/ 41.6 Flimsy v0.6 Ian Oversby 143.8 93
11 41.8/ 41.0/ 17.3 guru v0.3 Robert Hale 142.6 2
12 32.2/ 22.4/ 45.4 A Blaise Affair Robert J. Street 142.1 12
13 43.5/ 45.6/ 10.9 Short Sword 4 JS Pulido 141.4 74
14 31.2/ 21.2/ 47.5 Scanitator 3.0 Christian Schmid 141.3 99
15 38.8/ 37.3/ 23.9 Pain and Panic v0.1 Robert Hale 140.3 1
16 31.0/ 23.5/ 45.5 Six-Finger Discount From Ryan Coleman 138.5 46
17 34.1/ 33.5/ 32.4 If I only had a brain? Ryan Coleman 134.6 44
18 39.4/ 47.4/ 13.2 Dwa Michaly b Waldemar Bartolik 131.4 80
19 26.0/ 21.3/ 52.7 4-Finger Discount From th Ryan Coleman 130.7 47
20 23.6/ 19.4/ 56.9 Quantum Christian Schmidt 127.9 96
21 28.0/ 31.6/ 40.4 Zorm-B Anonymous 124.4 77
22 23.7/ 23.3/ 53.0 It's a Harmless Addition Robert Hale 124.1 30
23 25.3/ 28.7/ 46.0 It's a Harmless Addiction Robert Hale 121.9 33
24 24.6/ 29.3/ 46.2 Kohonenian Dream Robert J. Street 119.9 66
25 1.1/ 54.6/ 44.4 PruebaX5 Eleazar Martinez Can 47.5 3
35.0/ 32.7/ 32.3 Top 25 Averages 137.2 49
______________________________________________________________________________
The Hint
A new way of avoiding self-bombing
[Note: the warriors mentioned in this article are available from Planar's Web
page: http://pauillac.inria.fr/~doligez/corewar/]
When writing a stone, if all else is equal (which it never is :-) ), a stone
which bombs with a smaller mod will beat the one which bombs with a larger
mod. However, if a stone has a small mod, then it will very often drop a bomb
on itself. Sometimes this can be used to mutate the warrior (for example, see
Beppe Bezzi's _Tornado 3.0_, which uses a spl bomb to change from a bomber to
a coreclear), but often it is a thing to be avoided.
This article gives a (hopefully) new way which means that the mod of a (self-
splitting) bomber can be reduced. The actual method is rather complicated, so
instead of doing a general analysis, I'll do a specific example instead:
Consider this stone, taken from John K Wilkinson's _evoltmp_:
(The actual stone is between the lines labelled p and l, with the d-clear bomb
at jclr; the rest of the warrior is a boot routine and a dat 1,1 decoy)
;redcode-94
;name evoltmp's stone
;author John K W & Philip Kendall
;assert CORESIZE == 8000
org EvolCap
BOOT_DIST equ (2626+24*66)
S equ 24
jclr: dat >2667, S+10
evol
EvolCap mov.i jclr, evol+BOOT_DIST+S+4
for 6
mov.i {pptr, <pptr
rof
jmp @pptr
for 84
dat.f 1,1
rof
p spl #0, {-S+1
a add.f 3, 1
m mov }S-2, -S+1
d djn.f -2, <-1160
c mov S, >-S
l djn.f -1, >-S-1
pptr: nop 0, evol+BOOT_DIST+6
end
(Note that this method will work only for self-splitting stones which have the
add line _before_ one of the bombing lines)
This stone is a stone -> d-clear, which mutates when the mov line looks like
mov }-2,1
which moves the spl at p on top of the first djn.f. Currently, it uses a mod-8
step for bombing, with a-field increments offset from this by 3. If we trace
the execution of the warrior once it is booted, the instructions execute in
this order:
[p]
[a p] (*)
[m a p]
[d m a p]
[a d m a p] (*)
[m a d m a p]
[d m a d m a p]
[a d m a d m a p] (*)
etc...
If we look at the lines marked with * (those where the number of processes is
congruent to 2 mod 3), we notice that there are more adds (a) executed than
movs (m). Therefore there must be some of the mod-8 locations which we thought
were going to be bombed which will not be, at least before the mutation occurs
(a small amount of bombing still occurs after this due to the spl instruction
at p still feeding processes into the loop). In fact, the locations which are
missed are those which occur after (3/2)*n*(n+1) additions, where n=0,1,2,...
This fact enables us to reduce the mod to 4 and, so long as we syncronise our
stone correctly, still avoid self-bombing. As an example, let us consider a
step size, S, of 28 (it is important to keep the step size small with this
warrior as the region from (c-S) to (c+S) is not cleared). The mov instruction
we are trying to miss is mov }2,-3, as this would increment our step size.
Each time a is executed, 28 is added to the initial value of -2 of the a-field
of m, so we require 28k mod 8000=4 (k=0,1,2,...) to get to mov.i }2,-3. This
has a solution of k=1143. There is no value of n such that (3/2)*n*(n+1)=1143,
but if we take n=27, this gives 1134. Then, if we start the warrior with m as
mov }(-2+(9*S)),(1-(9*S))
after 1134 additions (the point at which m is skipped), m will be
mov }(-2+(1143*S)),(1-(1143*S))
which is equivalent to
mov }2,-3
the instruction we are trying to miss. This only leaves one very small
problem: before the mutation to d-clear occurs, at one stage m looks like
mov }30,-31
which increments the a-field of the bomb, so we must change the jclr
instruction to
jclr: dat >2666, S+10
(A better bomb to use would be dat >5335,S+10, which would have to start off
as dat >5334,S+10 - see _Core Warrior 52_ for details)
So, here's the complete source to the mod 4 version of evoltmp's stone:
;redcode-94
;name evoltmp's stone
;author John K W & Philip Kendall
;assert CORESIZE == 8000
;see Core Warrior 61 for details of this warrior
org EvolCap
BOOT_DIST equ (2626+24*66)
S equ 28
magic equ 9 ; to avoid self-bombing
jclr: dat >2666, S+10 ; changed to >2667 by the stone
evol
EvolCap mov.i jclr, evol+BOOT_DIST+S+4
for 6
mov.i {pptr, <pptr
rof
jmp @pptr
for 84
dat.f 1,1
rof
spl #0, {-S+1
add.f 3, 1
mov }(-2+(magic*S)),(1-(magic*S))
djn.f -2, <-1160
mov S, >-S
djn.f -1, >-S-1
pptr: nop 0, evol+BOOT_DIST+6
end
At present, if this warrior is run off against some other warriors, its
performance is actually worse than that of the mod-8 version; for example,
versus Planar's _Impfinity v4g1_, the mod-8 version scores 61/86/53
(236 points), whilst the mod-4 version scores 37/99/64 (175 points), and
against Paul Kline's _Torch t18_, mod-8 scores 95/63/42 (327), but mod-4
scores only 77/68/55 (286). One of the principal reasons for this change in
performance is that the mod-4 version is in its bombing mode for
(approximately) twice as long as the mod-8 version; this means that a)
Impfinity has twice as long to build up processes in its imp spirals and b)
the d-clear is running for less time, both of which make Impfinity harder to
kill. This is shown by the fact that the change versus Impfinity is mostly
from wins to ties. Similar considerations apply versus Torch, and many other
warriors.
The above is only an example of a technique, rather than an attempt to create
a warrior which will do well on the Hills, so the fact that performance is
actually degraded by this is not of prime importance; however, it does show
that reducing the mod of a stone is not always a good point.
______________________________________________________________________________
Questions? Concerns? Comments? Complaints? Mail them to people who care.
authors: Beppe Bezzi <bezzi@nemo.it>, Anton Marsden <amarsden@mcs.vuw.ac.nz>,
Christian Schmidt <schmc003@goofy.zdv.Uni-Mainz.de> and Philip Kendall
<pak21@cam.ac.uk>
|
