Issue 69 20 august, 1998
_______________________________________________________________________________
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 from:
ftp://rtfm.mit.edu/pub/usenet/news.answers/games/corewar-faq.Z
http://www.koth.org/corewar-faq.html
FTP site is: ftp://ftp.csua.berkeley.edu/pub/corewar
Mirrored at: ftp://ftp.inria.fr/INRIA/Projects/para/doligez/cw/mirror
and: ftp://www.koth.org/corewar
pMARS itself is also available from:
Stormking web pages--http://www.koth.org/pmars.html
Terry's web page--http://www.ncs.infi.net/~wtnewton/corewar/
Fechter ftp site--ftp://members.aol.com/ofechner/corewar
Web pages are at:
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!
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
_______________________________________________________________________________
Welcome back.
You can find in this issue three hints written by John Metcalf and Ian Oversby.
So enjoy and happy programming.
-- Christian Schmidt
_______________________________________________________________________________
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
# %W / %L / %T Name Author Score Age
1 44.2/ 35.6/ 20.3 Recycled Bits David Moore 152.7 7
2 35.7/ 25.5/ 38.8 Vain Ian Oversby 145.9 93
3 36.4/ 28.2/ 35.4 Vigor Ken Espiritu 144.6 81
4 39.4/ 34.5/ 26.1 Defender Ian Oversby 144.3 1
5 36.4/ 28.7/ 35.0 Fixed Ken Espiritu 144.1 77
6 35.3/ 26.6/ 38.1 Shadow Christian Schmidt 144.0 0
7 33.5/ 23.5/ 43.0 Newt Ian Oversby 143.4 171
8 27.8/ 14.0/ 58.2 Fugitive David Moore 141.5 47
9 33.5/ 25.5/ 41.0 Tuesday Afternoon John K W 141.5 21
10 40.4/ 40.1/ 19.6 Diamonds and Rust v2 Christian Schmidt 140.7 12
11 36.9/ 33.6/ 29.5 Shape Christian Schmidt 140.2 18
12 31.9/ 25.1/ 43.0 Ultraviolet-B Ken Espiritu 138.6 103
13 27.7/ 16.9/ 55.4 Return Of Return Of The J John K W 138.6 43
14 35.5/ 32.4/ 32.1 Fire and Ice David Moore 138.6 66
15 33.4/ 29.1/ 37.5 Alien Christian Schmidt 137.6 26
16 36.7/ 36.2/ 27.1 Digitalis 4 Christian Schmidt 137.3 65
17 37.2/ 37.2/ 25.5 The Body Guard Ian Oversby 137.3 20
18 39.0/ 42.5/ 18.5 Blurstone M. J. Pihlaja 135.6 17
19 33.0/ 31.0/ 36.0 Nine Seven Six M R bremer 135.0 2
20 38.4/ 42.3/ 19.3 Electric Head 2 Anton Marsden 134.4 21
21 30.0/ 26.1/ 43.9 Gigolo Core Warrior staff 133.9 3
22 34.9/ 36.5/ 28.6 Tornado 4 Beppe Bezzi 133.3 4
23 36.1/ 41.7/ 22.1 Eraser Ken Espiritu 130.6 18
24 32.6/ 34.9/ 32.5 Falcon v0.5 Ian Oversby 130.3 0
25 31.7/ 35.7/ 32.6 C I A Anders Ivner 127.6 7
Age since last issue: 13 ( 7 last issue, 34 the issue before )
New warriors: 9 Turnover/age rate 69%
Average age: 36 ( 34 last issue, 32 the issue before )
Average score: 138 ( 134 last issue, 133 the issue before )
The top 25 warriors are represented by just 9 independent authors: Schmidt and
Oversby with 5, Espiritu with 4, Moore with 3, Wilkinson with 2, and the rest
with one each.
David Moore's Recycled Bits dominates with a big gap the hill.
_______________________________________________________________________________
94 - What's New
# %W / %L / %T Name Author Score Age
2 38.5/ 35.9/ 25.6 Defender Ian Oversby 141.1 0
21 29.4/ 28.6/ 42.0 Faces 0.1 Csaba Biro 130.1 1
1 41.7/ 37.3/ 21.0 Diamonds and Rust v2 Christian Schmidt 146.0 0
3 41.9/ 43.9/ 14.3 Red Carpet David Moore 139.9 1
1 45.6/ 34.5/ 19.8 Recycled Bits David Moore 156.7 0
18 32.8/ 35.8/ 31.4 C I A Anders Ivner 129.8 0
23 30.6/ 36.3/ 33.1 Falcon v0.5 Ian Oversby 124.9 0
4 33.4/ 27.8/ 38.8 Shadow Christian Schmidt 138.9 0
?? ? / ? / ? Nine Seven Six M R bremer ? 1
?? ? / ? / ? Gigolo Core Warrior staff ? 1
?? ? / ? / ? Tornado 4 Beppe Bezzi ? 1
Diamonds and Rust v2 spots the first place until Moore strikes back with his
new version of Recycled Bits.
_______________________________________________________________________________
94 - What's No More
# %W / %L / %T Name Author Score Age
26 29.4/ 33.3/ 37.3 obvious to those who know Robert Macrae 125.4 117
26 1.9/ 2.0/ 0.1 Red Carpet David Moore 5.8 3
26 32.5/ 41.9/ 25.6 He Scans a Little Crooked P.Kline 123.0 7
26 1.2/ 1.6/ 1.2 Shadow Christian Schmidt 4.8 17
The only significant loss here is Macrae's obviouse to those...
_______________________________________________________________________________
94 - What's Old
# %W / %L / %T Name Author Score Age
7 33.5/ 23.5/ 43.0 Newt Ian Oversby 143.4 171
12 31.9/ 25.1/ 43.0 Ultraviolet-B Ken Espiritu 138.6 103
2 35.7/ 25.5/ 38.8 Vain Ian Oversby 145.9 93
Newt keeps on going followed by Ultraviolet-B and Ian's next stone/imp Vain.
_______________________________________________________________________________
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 Damage Incorporated Anton Marsden 373 Q^2 -> Bomber
4 Return Of The Jedimp John K W 357 Q^2 -> Stone/imp
5 unrequited love kafka 346 Q^2 -> Paper
6 Impish v0.2 Ian Oversby 345 Stone/imp
7 Gigolo Core Warrior staff 332 Q^2 -> Stone/imp
8 Falcon v0.3 Ian Oversby 275 P-warrior
9 Nine Seven Six M R Bremer 232 Q^2 -> Stone/imp
10 Rosebud Beppe 218 Stone/imp
11 Q^2 Miro Anders Ivner 214 Q^2 -> Scanner/bomber
12 Instant Wolf 3.4 Edgar 205 P-warrior
13 Goldfinch P.Kline 201 P-warrior
14 Simple v0.4b Ian Oversby 197 QScan -> Stone/imp
15 Trident^2 John K W 195 Q^2 -> Stone/imp
16 ompega Steven Morrell 189 Stone/imp
17 Frogz Franz 172 Q^2 -> Paper
18 Newt Ian Oversby 171 * Q^2 -> Stone/imp
19 The Machine Anton Marsden 164 Scanner
20 Memories Beppe 152 Scanner
21 Head or Tail Christian Schmidt 142 Q^2 -> Paper
22 Electric Head Anton Marsden 140 P-warrior
23 Tiberius 3.1 Franz 130 Q^2 -> Paper
24 obvious to those who k Robert Macrae 117 Q^2 -> Paper
25 Solomon v0.8 Ian Oversby 116 Stone and scanner
Obviouse stops at 24th, while Newt climbs another place to 18th.
_______________________________________________________________________________
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: Mon Aug 17 15:15:29 PDT 1998
# %W / %L / %T Name Author Score Age
1 60.3/ 26.9/ 12.9 Hostile Takeover Ryan Coleman 193.7 25
2 49.0/ 31.9/ 19.2 Cut 'n Paste Mark Howard 166.0 19
3 48.1/ 39.6/ 12.3 Have Pity John Metcalf 156.6 4
4 46.8/ 39.2/ 14.0 OXY-10 Mark Howard 154.4 17
5 45.9/ 39.8/ 14.2 Snowman v4.2 John Metcalf 152.1 3
6 42.4/ 33.5/ 24.1 Jam Major Zick 3 Ryan Coleman 151.2 31
7 44.7/ 39.1/ 16.2 Look or Drop?? Ryan Coleman 150.3 27
8 37.6/ 30.5/ 32.0 StiZ v0.1 Stajp 144.6 38
9 32.4/ 21.6/ 46.0 SpiImp Sergio Camarena 143.3 54
10 32.4/ 21.6/ 46.0 SpiImp Sergio Camarena 143.2 55
11 36.4/ 30.5/ 33.1 Stranger John Metcalf 142.2 9
12 33.7/ 26.7/ 39.6 Weasel WFB 140.7 18
13 40.8/ 41.8/ 17.5 stargate derivative 1 P._V._K., Beppe Bezz 139.7 21
14 33.2/ 27.0/ 39.8 StiY v0.1 Stajp 139.3 39
15 38.1/ 37.2/ 24.7 Arsonic B P._V._K. 139.1 33
16 32.3/ 25.7/ 42.0 Multiply or Drop?? Ryan Coleman 138.8 32
17 31.8/ 25.4/ 42.8 More Snotty CSnots WFB 138.2 35
18 39.5/ 40.9/ 19.6 Silly Evil Quack Ryan Coleman 138.0 30
19 27.2/ 18.1/ 54.7 Unknown Quantity John Metcalf 136.4 2
20 37.0/ 38.5/ 24.5 Big Guy John Lewis 135.4 64
21 39.7/ 44.0/ 16.3 Neverland John Metcalf 135.4 5
22 30.7/ 26.7/ 42.6 More Snotty Snots WFB 134.7 36
23 35.3/ 42.9/ 21.9 One Step Beyond John Metcalf 127.6 79
24 5.6/ 79.5/ 14.8 Whatisit? John Metcalf 31.7 1
25 7.7/ 0.2/ 0.1 Multipack John Metcalf 23.3 11
Ryan dominates with a very, very big gap the Beginner's Hill. Mmmmhhhh, he
gets 100% wins against his own warriors. Mmmhhh, looks like hostile takeover?!
_______________________________________________________________________________
The Hint - Code snippets by John Metcalf
For warriors which contain a multi-process paper, or a vector
launched imp, it is neccessary to create a number of processes
running in parallel. The usual method of achieving this is to
use a combination of spl 1 and mov -1,0 instructions.
So says Beppe Bezzi in Corewarrior #1:
: The simpler way to generate an exact number of parallel
: processes is converting the number required in binary 3 -> 11,
: subtract one -> 10, use a spl 1 for every one and a mov -1,0
: for every zero.
E.g. 5 decimal = 101 binary, take away one = 100 binary. This
becomes the following code,
spl 1 ;
mov -1,0 ; Generate 5 processes
mov -1,0 ;
However, by using these long-overlooked snippets of code I have
found, it is possible to do this slightly faster, and perhaps
even gain an extra b-field or two into the bargain, which can
be used for storing data, or decrementing locations in core.
To date, I have managed to find snippets which product 3,5,7 and
9 processes. If required, it is possible to add a number of
spl 1,<nnn instructions to the end of the snippet. Each one will
double the number of processes.
N OLD CODE NEW CODE COMMENTS
2 spl 1, <xxx spl 1, <xxx Here only for completeness
3 spl 1, <xxx spl 2, <xxx One cycle faster, and one
mov -1, 0 spl 1, <yyy extra b-field
5 spl 1, <xxx spl 2, <xxx Faster by two cycles, and
mov -1, 0 spl 2, <yyy two extra b-fields
mov -1, 0 spl 1, <zzz
7 spl 1, <xxx spl 1, <xxx One cycle faster, but no
spl 1, <yyy spl 1, }0 extra b-fields, and won't
mov -1, 0 spl 1, <yyy work under ICWS'88 and is
self-modifying.
9 spl 1, <xxx spl 2, <xxx Three cycles faster, plus
mov -1, 0 spl 2, <yyy two extra b-fields, but
mov -1, 0 spl 1, }0 won't work under ICWS'88
mov -1, 0 spl 1, <zzz and is self-modifying.
While you will be lucky if this method gains your paper a
fraction of a point on the hill, it is still clearly better.
Of course, the code for 2^n processes remains identical, and
cannot be improved.
Finally, I'm certain there are some other interesting code
snippets just waiting to be found. Its always interesting
to see that sometimes, the way we've always done something
is not neccessarily the best way :-)
John Metcalf
_______________________________________________________________________________
The Hint - Improving a beginner's warrior by Ian Oversby
Once again, this hint is concerned with improving a beginner's
warrior. Thank you to WFB for sharing his warrior Weasel with
us.
Here is the original:
;redcode-b
;name Weasel
;author WFB
;strategy P-spacer...combines Diamond in the Rough and
;strategy More Snotty CSnots.
;assert CORESIZE==8000
_RES equ #0
_STR equ #1 ;Not that obviously
step equ 3044
start
res ldp.ab _RES, #0
str ldp.a _STR, str1 ;load strategy in use
sne.ab #0, res ;check result
lost add.a #1, str1 ;lost change
mod.a #2, str1 ;secure result
win stp.ab str1, _STR
str1 jmp @0, dstart
dat 0, csnots
for 69
dat }-2667,<2667
rof
dstart spl #1234
diamond spl #-step,<step
rough mov >-step,step+1
add diamond,rough
djn.f rough,<diamond-50
dat #0,#0
dat #-7
csnots spl 1,<-200
mov -1,0
spl 1,<-800
mov -1,0
s1 spl @0,}1100 ;experimenting with values-original is 1100
mov }-1,>-1
s2 spl @0,<2620 ;original is 2620
mov }-1,>-1
s3 spl @0,{1870 ;original value is 1870
mov }-1,>-1
mov snot,<88 ;88
s4 mov {-3,<1
spl @0,}-639 ;-639
mov 2,<-4
jmp -1,<-13
snot dat <2667,<2667*2
end start
This is a switch on loss p-spacer combining a paper and a stone.
I can see a few minor modifications that may improve it a little.
Firstly, I boot the stone as otherwise we are easily found and
destroyed by one-shot scanners. I changed the first spl line to
spl #0, #0 for invisibility to f-scans.
I notice the number of processes for the paper does not match the
number of lines. Maybe removing the final launcher would not
matter too much either. At the same time I add another mov line
to the first launcher and use the correct number of processes for
the number of lines in the paper.
The result is:
;redcode-b test
;name Weasel
;author WFB
;strategy P-spacer...combines Diamond in the Rough and
;strategy More Snotty CSnots.
;assert CORESIZE==8000
_RES equ #0
_STR equ #1 ;Not that obviously
step equ 3044
start
res ldp.ab _RES, #0
str ldp.a _STR, str1 ;load strategy in use
sne.ab #0, res ;check result
lost add.a #1, str1 ;lost change
mod.a #2, str1 ;secure result
win stp.ab str1, _STR
str1 jmp @0, sboot
dat 0, csnots
for 40
dat }-2667,<2667
rof
sdist equ 2000
sboot mov.i {loc, <sptr
for 4
mov.i {loc, <sptr
rof
mov.ab #-7, *sptr
spl.b @sptr, <-700
sptr div.f #sdist, #sdist-1
dat.f }300, >200
dstart spl #0, #0
diamond spl #-step,<step
rough mov >-step,step+1
add diamond,rough
djn.f rough,<diamond-50
loc dat #0,#0
csnots spl 1,<-200
spl 1,<-800
spl 1,<-800
s1 spl @0,}1100 ;experimenting with values-original is 1100
mov }-1,>-1
mov }-2,>-2
s2 spl @0,<2620 ;original is 2620
mov }-1,>-1
s3 spl @0,{1870 ;original value is 1870
mov }-1,>-1
mov snot,<88 ;88
mov 2,<-4
jmp -1,<-13
snot dat <2667,<2667*2
end start
15 32.4/ 32.9/ 34.7 Weasel WFB 131.9 1
...
19 29.0/ 30.3/ 40.7 Weasel WFB 127.7 12
This scores about 4 points better than the original. Time for some
analysis.
The score against the hill is as follows:
Hostile Takeover : 38/34/28 P-Spacer
OXY-10 : 32/41/27 Core-clear
Multipack : 53/35/12 Scanner/something
Cut'n'Paste : 24/44/32 Scanner (Oneshot?)
Look or Drop : 63/25/12 Scanner/Stone
Jam Major Zick 3 : 33/48/19 Scanner
Stranger : 36/29/35 Incendiary Bomber
Stranger in town : 44/20/36 ???
StiZ v0.1 : 14/54/32 Paper/Scissors
SpiImp : 22/06/72 Something with imps ?
SpiImp : 22/06/72 Something with imps ?
Arsonic B : 45/37/18 Bomber ?
Multiply or Drop : 02/24/74 Paper/Stone
Stargate... : 35/41/24 Core-clear
Incisor+ : 24/60/16 One-shot
StiY v0.1 : 03/17/80 Paper/Scissors
Weasel : 18/25/57 Paper/Stone
Rusty : 30/51/19 Scanner
More Snotty C... : 14/32/54 Paper
More Snotty Snots : 30/29/41 Paper
One Step Beyond : 55/30/15 Scanner
Silly Evil Quack : 59/35/06 Scanner
Big Guy : 14/29/57 ???
Scarecrow : 64/24/12 Scanner
Sword in the... : 36/46/18 ???
I can see a definite weakness against one-shot scanners and
core-clears. We don't score too well against papers either
as we have no stunning power. I think we will try to counter
the first problem without modifying the warrior too much
away from the original idea.
To kill scanners, we should run the stone more often but
then we are vulnerable to papers. How about adding an imp to
the stone. As it has a double spl we aren't slowed too much
and now we have resistance to papers and other stones. In
addition, maybe the other scanners do not have much imp
resistance and we can improve our score against them.
I took the imp out of Juliet Storm and tacked it on the end.
I also changed the decoy to dat 1,1 to slow djn streams and
moved our own djn stream a little further back to decoy the
enemy one-shots better. I run the imp/stone twice as often
as the paper although subsequent tests showed this was not as
much improvement as I expected.
The code:
;redcode-b test
;name Weasel
;author WFB
;strategy P-spacer...combines Diamond in the Rough and More Snotty
;strategy CSnots.
;strategy More testing (Ian)
;assert CORESIZE==8000
_RES equ #0
_STR equ #121 ;Not that obviously
step equ 3044
start
res ldp.ab _RES, #0
str ldp.a _STR, str1 ;load strategy in use
sne.ab #0, res ;check result
lost add.a #1, str1 ;lost change
mod.a #2, str1 ;secure result
win stp.ab str1, _STR
str1 jmp @0, sboot
dat 0, sboot
dat 0, csnots
for 17
dat.f #1, @1
dat.f #1, *1
dat.f #1, #1
rof
dat.f #1, @1
sdist equ 4620
sboot mov.i {loc, <sptr
for 4
mov.i {loc, <sptr
rof
mov.ab #-7, *sptr
spl.b impboot, <-707
spl.b @sptr, <-700
sptr div.f #sdist, #sdist-1
dat.f }300, >200
dstart spl #0, #0
diamond spl #-step,<step
rough mov >-step,step+1
add diamond,rough
djn.f rough,<diamond-1051
loc dat #0,#0
csnots spl 1,<-200
spl 1,<-800
spl 1,<-800
s1 spl @0,}1100 ;experimenting with values-original is 1100
mov }-1,>-1
mov }-2,>-2
s2 spl @0,<2620 ;original is 2620
mov }-1,>-1
s3 spl @0,{1870 ;original value is 1870
mov }-1,>-1
mov snot,<88 ;88
mov 2,<-4
jmp -1,<-13
snot dat <2667,<2667*2
imp_sz equ 2667
impboot spl 1 ,<-450
spl 1 ,<-458
spl <0 ,#vector+1
djn.a @vector,#0
imp mov.i #imp_sz, *0
jmp imp+imp_sz*7,imp+imp_sz*6
jmp imp+imp_sz*5,imp+imp_sz*4
jmp imp+imp_sz*3,imp+imp_sz*2
vector jmp imp+imp_sz ,imp
end start
The result:
2 37.5/ 21.7/ 40.8 Weasel WFB 153.3 1
...
19 29.0/ 30.3/ 40.7 Weasel WFB 127.7 12
Not bad! We lose very badly against Have Pity and the score is worse
than the original against Multipack. Fixing these problems is left
as an exercise.
_______________________________________________________________________________
The Hint - Longevity, Imp/Stones and Newt by Ian Oversby
In Core War, the true value of a fighter is measured by its age. Beppe Bezzi,
a veteran redcoder who should know a few things about long-lasting warriors,
once pointed out that longevity comes from scoring points not only against the
current hill but also the hill of the future. As we do not know how the hill
of the future is going to look, how can we do this. The answer is simple -
score well against everything that is viable on the hill.
So, how is it possible to future-proof your new program? I believe we must
look at the problem in the correct light. If we take the traditional analogy
of warriors to stone/paper/scissors then we condemn our warrior to death as
the hill balance changes. If we wrote a scanner then when many stones enter
the hill we will surely die. However, this analogy holds very loosely. For
example, it is possible to write a stone that will kill paper or paper that
will kill scissors. That attitude that we _can_ beat a warrior type that
would usually beat us is very important. It is interesting to note that
Vigor, allegedly a paper, suffers its worst defeat at the hands of a stone
- Newt.
I will use Newt to illustrate the points I wish to discuss. Firstly we will
see how the warrior operates and then we will see how he performs against
various enemies.
The stone itself bombs and decrements in a mod-5 pattern with the decrements
offset from the bombs by 2. This gives as good a pattern as I could wish for.
When the bombing run is finished the core should look like this:
*.X..*.X..*.X..*
where * is decrement, X is bomb and . is unbombed. The stone at this point
will look like this:
spl.b #0, <-step+1
stone spl.b #step, <-step
mov.i {0, 2
add.f stone, -1
hit djn.f -2, <stone-step
mov.i cbmb, >gate
djn.f -1, >gate
The mov line executes, decrementing itself and moving the spl at stone over
the hit line. This moves the stone into the core-clear phase. The stone is
quite big and slow at seven lines and 0.33c. However, a correctly placed bomb
is usually fatal to scanner. The decrements are also quite useful, sometimes
wounding opponents. The stone scores well against traditional scanners and
bombers. He Scans Alone scores around even. I believe this is because of the
mod-5 bombing - Tornado scores similarly against HSA. I feel I could have
improved this by leaving the spl above stone as a spl #0, #0 making it
invisible to HSA. Alternatively, the a-field could be changed to (say) 100
so it doesn't point to the rest of the stone. This may have a small effect
against silk a-field bombing.
The spl line above the stone can be hit with either spl or dat with no effect
to the correct operation. The hit line may also be hit with either spl or
dat. A dat will slow the warrior but the clear phase will still be entered.
This helps against both stones and bombers.
The p-spacers are handled by the Q^2 scan and the resistance to imp/stone
enemies clears and one-shots. The pattern of bombing provides this
resistance. There are five possibilities against a standard d-clear.
spl #0
mov bmb, >gate
djn -1, >gate
1. Bomb hits the spl, decrement misses No real harm done
2. Bomb hits the mov, decrement misses Clear disabled
3. Bomb hits the djn, decrement hits spl, Clear 1/3 speed
4. Bomb misses, decrement hits mov. Maybe clear moving dat 0,0
5. Bomb misses, decrement hits djn. Clear slowed
In cases 2 and 3 we have an excellent chance to kill with our own d-clear
which is entered after around 6000 cycles. In case 4 we should tie as the
d-clear stops when the bomb reaches its own gate. In case 5 we have an even
chance to win or lose. Case 1 often results in a loss. We may also hit the
gate or the bomb with our bomb giving more chances to win or draw.
The double spl gives additional protection against d-clear. D-clear often
misses adjacent locations with its bombs. The stone part is cleared later
in the round and the imps have less time to reach the gate often resulting
in a draw. Now, new clears are appearing in p-spacers which perform better
against Newt (although not against Vain :-) A more effective clear may look
like this:
gate dat.f #0, #4000
dat.f #0, last-gate+1
bmb dat.f #0, last-gate+1
spl.b #0, #0
mov.i bmb, >gate
mov.i bmb, >gate
djn.f -1, >gate
Newt is not so effective against this warrior as decrementing the djn and
bombing lines no longer really hamper it. Newt scores about 25/50/25 against
this new clear. Bombing the djn line or the first line slows the clear. Newt
enters its own clear after 6000 cycles and this gives chances to win. This
clear is very tuned against Newt with the a-field decrement.
We could improve Newt against this new clear in a few ways.
1. Shorter bombing run entering clear earlier
2. Heavier imp so our imps do not reach the gate
3. A-field increment to more readily disable the clear.
I tried the last of these which was fairly effective.
stone spl.b #step, <-step # swapped the two spl lines
spl.b #0, <-step-1
mov.i }step-1, hit-step # changed to point at -1
add.f stone, -1
hit djn.f -2, <stone-step
mov.i cbmb, >gate
djn.f -2, >gate # changed to -2
The clear could also optimize against this new form of Newt but I don't
really think it is worth beating just one warrior badly. Besides, we can't
expect to beat everything. I feel these changes weaken Newt against stones as
now both top spls are vulnerable. A spl striking the hit line will no longer
cause Newt to enter the perfect d-clear until part way through the bombing
run. I think the a-field decrement is more effective against paper too.
Consider:
spl @-1, >x
mov }-1, >-1
spl @1, >x
mov }-1, >-1
spl @0, >x
mov }-2, >-1
spl @0, >x
mov }0, >-1
It is just a vague feeling and an untested exercise for the reader.
Specialised p-spacer components such as p-clear can give Newt some problems.
The solution is simple. Be the _best_ imp/stone on the hill. This way, when
the other imp/stones drop off the hill and the specialists lose their prey,
they also drop off the hill and Newt recovers. This just leaves papers and
other imp/stones to worry about. When I wrote Newt, the hill contained seven
papers and five imp/stones. As I write this, the hill has six imp/stones and
seven papers (of sorts). It is important to do well against both of these
classes.
Newt has a similar size/speed ratio to other imp/stones. It has imp-killing
power with the d-clear and resistance to other d-clears and spl/dat bombs and
decrements. The result is reasonable scores against other imp/stones,
particularly those with no gate of their own. I think the d-clear bomb should
have been <2667 as this works better against the medium 3 point imps found in
many imp stones.
Papers are a real problem for imp/stones such as Newt. The main difficulty is
the size of the warrior for the enemy Qscan. The stone and stone boot alone
is 18 instructions. Add this to 12 lines to boot the imp-launch and that
makes 30 lines on top of the QScan. Compare this to a typical paper that is
about 11 lines long or 13 if a boot is included. This 15+ line difference
means that the imp/stone does not score so well in the QScan shootout.
Slimming the launch can reduce this problem a little.
I re-ordered the QScan for the hill version and computer-optimized it against
the Gigolo QScan and Probe QScan. I believe this may have given some good
scores against papers on the hill as they tend to have a very similar QScan
order to these early Q^2 scans. Using a published QScan order in particular
can be very dangerous.
Generally, the two ways for an imp/stone to beat a paper enemy are a long
period of intense bombing like Gigolo or a stone suicide late in heel for a
warrior. It is very dangerous to have the same QScan order as a published
warrior.
While writing this article, I noticed the scores against papers were not so
good as I remember. So saying, I present a new version of Newt with some of
the ideas from the article.
* Imp launch entered earlier for heavier imp
* Imp launch unbooted for slimmer launch -> better QScan performance
* Imp launch decrement decoy further behind full decoy, better against
one-shots
* Stone first spl points after end of stone for [minor] paper protection
and finally...
* QScan scans backwards for hopefully better score on Mt Olympus ;-)
;redcode-94 test
;name Newt v0.2
;author Ian Oversby
;strategy Q^2 -> Imp/Stone
;assert 1
gate1 equ (init-7-dist)
pat equ 3315
sval equ (spin+5220)
dist equ 3
impy equ (imp+sep)
sep equ 1100
st equ 2667
QB EQU (start-550)
QS EQU (QD*2)
QD EQU -100
GAP EQU 12
REP EQU 8
REP2 EQU 2
datz EQU (table-3)
iboot MOV.I imp, impy
spin SPL.B #st+1, >prime
prime MOV.I impy, impy
ADD.F spin, jump
jump JMP.B impy-st-1, <-2200
imp MOV.I #st, *0
for 8
dat 0, 0
rof
dat 10*QS, 2*QS ; can get 21 values from this table
table: dat 4*QS, 1*QS ; and can also use the initial value
dat 23*QS, 3*QS ; of fnd
qinc: spl #GAP,-GAP
tab: add.a table,table
slow: add.a @tab,fnd
fast: add.ba *tab,@slow
which: sne.i datz,*fnd
add.a #QD,fnd
mov.i cbomb,*fnd
add.ab fnd,fnd
fnd: mov.i QB,GAP/2
add.f qinc,fnd
mov.i cbomb,*fnd
djn.b fnd,#REP
jmp boot,}QS*13
start:
; WHICH
seq.i QB+QS*0,QB+QS*0+QD
jmp which,}QB+QS*0+QD/2
; FAST
seq.i QB+QS*1,QB+QS*1+QD
jmp fast,}QB+QS*1+QD/2
seq.i QB+QS*13,QB+QS*13+QD
jmp fast,{fast
seq.i QB+QS*2,QB+QS*2+QD
jmp fast,{tab
seq.i QB+QS*3,QB+QS*3+QD
jmp fast,}tab
; SLOW
seq.i QB+QS*4,QB+QS*4+QD
jmp >fast,}QB+QS*4+QD/2
seq.i QB+QS*5,QB+QS*5+QD
jmp slow,}QB+QS*5+QD/2
seq.i QB+QS*6,QB+QS*6+QD
jmp slow,{tab
seq.i QB+QS*7,QB+QS*7+QD
jmp slow,}tab
seq.i QB+QS*10,QB+QS*10+QD
jmp >fast,<tab
seq.i QB+QS*11,QB+QS*11+QD
jmp slow,<tab
seq..i QB+QS*12,QB+QS*12+QD
djn.f slow,tab
seq.i QB+QS*23,QB+QS*23+QD
jmp >fast,>tab
seq.i QB+QS*24,QB+QS*24+QD
jmp slow,>tab
seq.i QB+QS*17,QB+QS*17+QD
jmp slow,{fast
; TAB
seq.i QB+QS*8,QB+QS*8+QD
jmp <fast,}QB+QS*8+QD/2
seq.i QB+QS*9,QB+QS*9+QD
jmp tab,}QB+QS*9+QD/2
seq.i QB+QS*15,QB+QS*15+QD
jmp tab,<tab
seq.i QB+QS*16,QB+QS*16+QD
jmp tab,{tab
seq.i QB+QS*20,QB+QS*20+QD
djn.f <fast,tab
boot MOV.I cbomb, @sptr
for 3
MOV.I <spos, {sptr
rof
spl.b iboot, <-1000
for 4
MOV.I <spos, {sptr
rof
SPL.B *sptr, <-200
sptr DIV.F #init+sval, #init+sval-7-3
cbomb DAT.F #1, #3+hit-gate1
; 2 DATS
init SPL.B #0, <stone-pat
stone SPL.B #pat, <-pat
loop MOV.I {0+pat, hit-pat
ADD.F stone, loop
hit DJN.F loop, <stone-pat
MOV.I init-dist, >gate1
last DJN.F -1, {gate1
spos DAT.F $0, $0
for 10
DAT.F $0, $0
rof
end start
Ian Oversby
____________________________________________________________________________
Questions? Concerns? Comments? Complaints? Mail them to people who care.
Authors: Beppe Bezzi <giuseppe.bezzi@galactica.it>, Anton Marsden
<amarsden@mcs.vuw.ac.nz>, Christian Schmidt <schmc003@goofy.zdv.Uni-Mainz.de>
and Philip Kendall <pak21@cam.ac.uk>
|
