Issue 8 December 4, 1995
______________________________________________________________________________
Core_Warrior_ is a weekly 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:
http://www.stormking.com/~koth ;Stormking
http://www.ecst.csuchico.edu/~pizza/koth ;Pizza
http://pauillac.inria.fr/~doligez/corewar/ ;Planar
______________________________________________________________________________
Greetings.
BIG changes this week. The Pizza server has some new additions. The
'Holding Pen' lists the queue of warriors awaiting battle much like the
StormKing page. Beppe grilled Thos on the meanings of the numbers at the
bottom labeled load average and he had this to say:
3:39pm load average: 3.49, 2.75, 2.68
The numbers are the load average for the last 1, 5, and 15 minutes.
1 / (load average) = % CPU pmars is getting. During the day here load is
about 2.5, at night it drops to 1.8 or less. When the pen finally empties,
the load average should go to less than 1.0.
Going by the "last challenge" times of the hill, it looks like it takes about
35 minutes to run the 94 hill with a load of around 2.5. It looks like it
takes about 45 minutes to run the 94x hill, but that may just be standard
deviation. Looks like a 94 + 94x challenge will take around 1:15 or so
during the day here, and a little under an hour at night.
Those are about as accurate as I can give you.
-Thos
____
Pizza has also opened a new hill, the 5*40. Also notice that five new slots
have opened up on the pizza hills and the number of battles was reduced to
200. Makes the division easier for slow people like me.
We've said it once, and we'll say it again--the big (CORESIZE 55440) hill has
been reopened at the Stormking server, address: <koth@stormking.com>.
Access the hill with ;redcode-94x
In addition to Planar's Corner, we have the staffs'--M R Bremer's and Beppe
Bezzi's--tournament entires for you to enjoy along with comments by the
authors. Check out Extra, Extra for all the goods.
The hints section is the last in the line in presenting basic warrior
strategies. Maybe one more. The question is: where do we go from here?
Your input is always appreciated. Also included in the hint: a version of
myVamp by Magnus Paulsson.
______________________________________________________________________________
Tournament Time:
The results of round 7 are in. To recap, the challenge was to beat your up to
500 line long opponent loaded at a fixed position halfway through core. This
was an exciting round with huge score ranges. Robert Macrae's Last Laugh won
by a land slide, with 90% wins. We saw some intriguingly new P-space
strategies, and we'd like to ask the top 3 contenders, Robert Macrae,
Myer Bremer and Beppe Bezzi to post their warriors with discussion of their
strategy. (Editior's Note: Macrae has already posted his warrior to the
newsgroup. Bremer's and Bezzi's warrior are in Extra Extra.)
The results in detail:
Battle 1 of 66 (2%) is finished:
Back Propagation by M R Bremer scores 471 (93 1 6)
BlackMoor by P.Kline scores 11 (1 93 6)
Battle 2 of 66 (3%) is finished:
Back Propagation by M R Bremer scores 484 (96 0 4)
Burger Chef & Jeff by Karl Lewin scores 4 (0 96 4)
Battle 3 of 66 (5%) is finished:
Back Propagation by M R Bremer scores 500 (100 0 0)
Frontwards by Steven Morrell scores 0 (0 100 0)
Battle 4 of 66 (6%) is finished:
Back Propagation by M R Bremer scores 15 (3 97 0)
Last Laugh by Robert Macrae scores 485 (97 3 0)
Battle 5 of 66 (8%) is finished:
Back Propagation by M R Bremer scores 176 (19 0 81)
Mister Bond by Derek Ross scores 81 (0 19 81)
Battle 6 of 66 (9%) is finished:
Back Propagation by M R Bremer scores 0 (0 100 0)
myVamp v3.7 by Paulsson scores 500 (100 0 0)
Battle 7 of 66 (11%) is finished:
Back Propagation by M R Bremer scores 50 (8 82 10)
run and hit by Beppe Bezzi scores 420 (82 8 10)
Battle 8 of 66 (12%) is finished:
Back Propagation by M R Bremer scores 240 (38 12 50)
Settete by Maurizio Vittuari scores 110 (12 38 50)
Battle 9 of 66 (14%) is finished:
Back Propagation by M R Bremer scores 483 (96 1 3)
Slayer by Randy Graham scores 8 (1 96 3)
Battle 10 of 66 (15%) is finished:
Back Propagation by M R Bremer scores 444 (86 0 14)
Tim by G. Eadon scores 14 (0 86 14)
Battle 11 of 66 (17%) is finished:
Back Propagation by M R Bremer scores 443 (88 9 3)
Victory or death by Anders Ivner scores 48 (9 88 3)
Battle 12 of 66 (18%) is finished:
BlackMoor by P.Kline scores 0 (0 100 0)
Burger Chef & Jeff by Karl Lewin scores 500 (100 0 0)
Battle 13 of 66 (20%) is finished:
BlackMoor by P.Kline scores 485 (97 3 0)
Frontwards by Steven Morrell scores 15 (3 97 0)
Battle 14 of 66 (21%) is finished:
BlackMoor by P.Kline scores 5 (1 99 0)
Last Laugh by Robert Macrae scores 495 (99 1 0)
Battle 15 of 66 (23%) is finished:
BlackMoor by P.Kline scores 425 (84 11 5)
Mister Bond by Derek Ross scores 60 (11 84 5)
Battle 16 of 66 (24%) is finished:
BlackMoor by P.Kline scores 500 (100 0 0)
myVamp v3.7 by Paulsson scores 0 (0 100 0)
Battle 17 of 66 (26%) is finished:
BlackMoor by P.Kline scores 93 (15 67 18)
run and hit by Beppe Bezzi scores 353 (67 15 18)
Battle 18 of 66 (27%) is finished:
BlackMoor by P.Kline scores 113 (22 75 3)
Settete by Maurizio Vittuari scores 378 (75 22 3)
Battle 19 of 66 (29%) is finished:
BlackMoor by P.Kline scores 96 (18 76 6)
Slayer by Randy Graham scores 386 (76 18 6)
Battle 20 of 66 (30%) is finished:
BlackMoor by P.Kline scores 50 (10 90 0)
Tim by G. Eadon scores 450 (90 10 0)
Battle 21 of 66 (32%) is finished:
BlackMoor by P.Kline scores 75 (15 85 0)
Victory or death by Anders Ivner scores 425 (85 15 0)
Battle 22 of 66 (33%) is finished:
Burger Chef & Jeff by Karl Lewin scores 100 (0 0 100)
Frontwards by Steven Morrell scores 100 (0 0 100)
Battle 23 of 66 (35%) is finished:
Burger Chef & Jeff by Karl Lewin scores 97 (19 79 2)
Last Laugh by Robert Macrae scores 397 (79 19 2)
Battle 24 of 66 (36%) is finished:
Burger Chef & Jeff by Karl Lewin scores 140 (13 12 75)
Mister Bond by Derek Ross scores 135 (12 13 75)
Battle 25 of 66 (38%) is finished:
Burger Chef & Jeff by Karl Lewin scores 0 (0 100 0)
myVamp v3.7 by Paulsson scores 500 (100 0 0)
Battle 26 of 66 (39%) is finished:
Burger Chef & Jeff by Karl Lewin scores 8 (1 96 3)
run and hit by Beppe Bezzi scores 483 (96 1 3)
Battle 27 of 66 (41%) is finished:
Burger Chef & Jeff by Karl Lewin scores 100 (0 0 100)
Settete by Maurizio Vittuari scores 100 (0 0 100)
Battle 28 of 66 (42%) is finished:
Burger Chef & Jeff by Karl Lewin scores 0 (0 100 0)
Slayer by Randy Graham scores 500 (100 0 0)
Battle 29 of 66 (44%) is finished:
Burger Chef & Jeff by Karl Lewin scores 32 (6 92 2)
Tim by G. Eadon scores 462 (92 6 2)
Battle 30 of 66 (45%) is finished:
Burger Chef & Jeff by Karl Lewin scores 225 (45 55 0)
Victory or death by Anders Ivner scores 275 (55 45 0)
Battle 31 of 66 (47%) is finished:
Frontwards by Steven Morrell scores 0 (0 100 0)
Last Laugh by Robert Macrae scores 500 (100 0 0)
Battle 32 of 66 (48%) is finished:
Frontwards by Steven Morrell scores 500 (100 0 0)
Mister Bond by Derek Ross scores 0 (0 100 0)
Battle 33 of 66 (50%) is finished:
Frontwards by Steven Morrell scores 0 (0 100 0)
myVamp v3.7 by Paulsson scores 500 (100 0 0)
Battle 34 of 66 (52%) is finished:
Frontwards by Steven Morrell scores 496 (99 0 1)
run and hit by Beppe Bezzi scores 1 (0 99 1)
Battle 35 of 66 (53%) is finished:
Frontwards by Steven Morrell scores 500 (100 0 0)
Settete by Maurizio Vittuari scores 0 (0 100 0)
Battle 36 of 66 (55%) is finished:
Frontwards by Steven Morrell scores 10 (2 98 0)
Slayer by Randy Graham scores 490 (98 2 0)
Battle 37 of 66 (56%) is finished:
Frontwards by Steven Morrell scores 204 (26 0 74)
Tim by G. Eadon scores 74 (0 26 74)
Battle 38 of 66 (58%) is finished:
Frontwards by Steven Morrell scores 35 (7 93 0)
Victory or death by Anders Ivner scores 465 (93 7 0)
Battle 39 of 66 (59%) is finished:
Last Laugh by Robert Macrae scores 496 (99 0 1)
Mister Bond by Derek Ross scores 1 (0 99 1)
Battle 40 of 66 (61%) is finished:
Last Laugh by Robert Macrae scores 486 (97 2 1)
myVamp v3.7 by Paulsson scores 11 (2 97 1)
Battle 41 of 66 (62%) is finished:
Last Laugh by Robert Macrae scores 500 (100 0 0)
run and hit by Beppe Bezzi scores 0 (0 100 0)
Battle 42 of 66 (64%) is finished:
Last Laugh by Robert Macrae scores 500 (100 0 0)
Settete by Maurizio Vittuari scores 0 (0 100 0)
Battle 43 of 66 (65%) is finished:
Last Laugh by Robert Macrae scores 500 (100 0 0)
Slayer by Randy Graham scores 0 (0 100 0)
Battle 44 of 66 (67%) is finished:
Last Laugh by Robert Macrae scores 104 (19 72 9)
Tim by G. Eadon scores 369 (72 19 9)
Battle 45 of 66 (68%) is finished:
Last Laugh by Robert Macrae scores 500 (100 0 0)
Victory or death by Anders Ivner scores 0 (0 100 0)
Battle 46 of 66 (70%) is finished:
Mister Bond by Derek Ross scores 23 (0 77 23)
myVamp v3.7 by Paulsson scores 408 (77 0 23)
Battle 47 of 66 (71%) is finished:
Mister Bond by Derek Ross scores 53 (6 71 23)
run and hit by Beppe Bezzi scores 378 (71 6 23)
Battle 48 of 66 (73%) is finished:
Mister Bond by Derek Ross scores 194 (32 34 34)
Settete by Maurizio Vittuari scores 204 (34 32 34)
Battle 49 of 66 (74%) is finished:
Mister Bond by Derek Ross scores 276 (50 24 26)
Slayer by Randy Graham scores 146 (24 50 26)
Battle 50 of 66 (76%) is finished:
Mister Bond by Derek Ross scores 211 (28 1 71)
Tim by G. Eadon scores 76 (1 28 71)
Battle 51 of 66 (77%) is finished:
Mister Bond by Derek Ross scores 130 (26 74 0)
Victory or death by Anders Ivner scores 370 (74 26 0)
Battle 52 of 66 (79%) is finished:
myVamp v3.7 by Paulsson scores 10 (2 98 0)
run and hit by Beppe Bezzi scores 490 (98 2 0)
Battle 53 of 66 (80%) is finished:
myVamp v3.7 by Paulsson scores 100 (0 0 100)
Settete by Maurizio Vittuari scores 100 (0 0 100)
Battle 54 of 66 (82%) is finished:
myVamp v3.7 by Paulsson scores 0 (0 100 0)
Slayer by Randy Graham scores 500 (100 0 0)
Battle 55 of 66 (83%) is finished:
myVamp v3.7 by Paulsson scores 87 (9 49 42)
Tim by G. Eadon scores 287 (49 9 42)
Battle 56 of 66 (85%) is finished:
myVamp v3.7 by Paulsson scores 85 (17 83 0)
Victory or death by Anders Ivner scores 415 (83 17 0)
Battle 57 of 66 (86%) is finished:
run and hit by Beppe Bezzi scores 96 (16 68 16)
Settete by Maurizio Vittuari scores 356 (68 16 16)
Battle 58 of 66 (88%) is finished:
run and hit by Beppe Bezzi scores 481 (96 3 1)
Slayer by Randy Graham scores 16 (3 96 1)
Battle 59 of 66 (89%) is finished:
run and hit by Beppe Bezzi scores 319 (57 9 34)
Tim by G. Eadon scores 79 (9 57 34)
Battle 60 of 66 (91%) is finished:
run and hit by Beppe Bezzi scores 75 (15 85 0)
Victory or death by Anders Ivner scores 425 (85 15 0)
Battle 61 of 66 (92%) is finished:
Settete by Maurizio Vittuari scores 363 (66 1 33)
Slayer by Randy Graham scores 38 (1 66 33)
Battle 62 of 66 (94%) is finished:
Settete by Maurizio Vittuari scores 314 (58 18 24)
Tim by G. Eadon scores 114 (18 58 24)
Battle 63 of 66 (95%) is finished:
Settete by Maurizio Vittuari scores 85 (17 83 0)
Victory or death by Anders Ivner scores 415 (83 17 0)
Battle 64 of 66 (97%) is finished:
Slayer by Randy Graham scores 88 (1 16 83)
Tim by G. Eadon scores 163 (16 1 83)
Battle 65 of 66 (98%) is finished:
Slayer by Randy Graham scores 480 (96 4 0)
Victory or death by Anders Ivner scores 20 (4 96 0)
Battle 66 of 66 (100%) is finished:
Tim by G. Eadon scores 491 (98 1 1)
Victory or death by Anders Ivner scores 6 (1 98 1)
Elapsed time: 5286 seconds (01:28:06)
Rank Name Author %W %L %T Score
___________________________________________________________________________
1 Last Laugh Robert Macrae 90 9 1 4963
2 Back Propagation M R Bremer 57 27 16 3306
3 run and hit Beppe Bezzi 54 36 10 3096
4 Victory or death Anders Ivner 52 48 0 2864
5 Slayer Randy Graham 45 41 14 2652
6 Tim G. Eadon 40 27 32 2579
7 myVamp v3.7 Paulsson 37 48 15 2201
8 Settete Maurizio Vittuari 30 37 33 2010
9 Frontwards Steven Morrell 31 53 16 1860
10 BlackMoor P.Kline 33 64 3 1853
11 Burger Chef & Jeff Karl Lewin 17 57 26 1206
12 Mister Bond Derek Ross 15 54 31 1164
The scores so far:
Name 1 2 3 4 5 6 7 total
Steven Morrell 5 10 9 13 14 10 4 65
P.Kline 7.5 9 7 11 11 12 3 60.5
Paulsson 7.5 11 11 9 2 5 6 51.5
Anders Ivner 5.5 8 4 0 10 14 9 50.5
Beppe Bezzi 7 7 13 2 8 3 10 50
Robert Macrae 0 0 0 12 12 13 12 49
M R Bremer 7 4 3.6 5 7 11 11 48.6
Maurizio Vittuari 6.5 5 6 3 9 8 5 42.5
Randy Graham 0 0 8 10 4 7 8 37
John K. Wilkinson 4 6 12 0 13 2 0 37
Karl Lewin 0 0 10 4 6 10 2 32
Derek Ross 3.5 3 3.3 7 3 6 1 26.8
G. Eadon 1.5 2 5 6 1 4 7 26.5
Kurt Franke 0 0 0 8 0 0 0 8
Michael Constant 0 0 0 0 5 0 0 5
Anders Scholl 0 1 2 1 0 0 0 4
John Lewis 0 0 3 0 0 0 0 3
Calvin Loh 0 0 1 0 0 1 0 2
As this tournament draws to a close -- only 2 more rounds to go, it looks like
we're up for some up for some final stretch excitement. The lead of Steven
Morrell and Paul Kline has dwindled to a mere 9 points and the midfield is
densely packed with 5 players scoring between 48 and 51 points. The game is
still wide open. Just as a reminder, the GIFs of winner and two runners-up
will be featured in the upcoming Core War Hall of Fame at stormking -- how's
that for an incentive?
Good luck for the remaining rounds,
Nandor & Stefan
______________________________________________________________________________
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
Last challenge: Sun Dec 3 10:16:49 PST 1995
# %W/ %L/ %T Name Author Score Age
1 44/ 37/ 19 quiz Schitzo 151 137
2 38/ 36/ 25 Derision M R Bremer 140 156
3 42/ 44/ 14 TestTTB Maurizio Vittuari 139 6
4 35/ 31/ 34 Tornado 2.0 Beppe Bezzi 138 1
5 40/ 42/ 18 Porch Swing 2 JKW 138 45
6 37/ 37/ 26 myVamp v3.7 Paulsson 137 484
7 20/ 5/ 75 Die Hard P.Kline 135 38
8 33/ 30/ 36 Torch t18 P.Kline 135 516
9 30/ 25/ 44 La Bomba Beppe Bezzi 134 132
10 27/ 20/ 53 juliet and paper M R Bremer, B. Bezzi 133 133
11 29/ 27/ 43 Jack in the box Beppe Bezzi 131 504
12 19/ 7/ 74 Night Train Karl Lewin 130 24
13 38/ 46/ 17 Taking Names P.Kline 130 25
14 37/ 46/ 16 ltest Anders Ivner 128 53
15 33/ 37/ 30 endpoint . M R Bremer 128 161
16 24/ 20/ 56 Hector 2 Kurt Franke 128 52
17 21/ 15/ 64 Impfinity v3i Planar 127 66
18 34/ 40/ 26 Test z JKW 127 44
19 26/ 25/ 49 test9q Kurt Franke 126 11
20 31/ 36/ 33 Saveme! Maurizio Vittuari 126 51
21 22/ 19/ 59 Evolve-- v.1 JKW 124 33
22 36/ 47/ 17 Leprechaun on speed Anders Ivner 124 312
23 30/ 36/ 34 Son & Father Maurizio Vittuari 123 167
24 22/ 23/ 55 test9p Kurt Franke 120 8
25 31/ 45/ 24 The Inferno Kurt Franke 116 2
Two behemoths get pushed off this week. Armory has been suffering for a
awhile and was finally pushed off by JKW's own test. Phq has had similar
problems. His last ditch defense (Saveme!) wasn't quite enough to keep it on
the hill. Torch t18 and Jack in the box move up to claim the vacated spots.
Torch is in very good health, as high as third, as low as 11th, but overall
hanging around 5th or 6th. Jack in the box is a super elastic yo-yo. Bezzi
thought it may get pushed off soon, but it has hung out in the top 5 this
week and is currently in 11th place. You'll have to see next week's issue
for the updated Hall of Fame. Sorry.
quiz continues to dominate, usually enjoying a >10 point advantage. Someone
_really_ needs to smash that thing. Who would be king then? ;-) La Bomba
RULED and is now sliding down the hill. Are qscanners becoming ineffective?
Hector 2 is a qscan --> replicator. Smart Bomb is a qscan --> p-warrior.
We'll see how these three fare in the weeks ahead.
The five additional slots on the hill means lots of new boys. P.Kline
decided to increase his presence on the hill with the addition of Taking
Names (which pushed off Phq) and Die Hard. Maurizio Vitturari has a warrior
under wraps that looks to be a worthy successor to Phq.
______________________________________________________________________________
94 - What's New
3 42/ 44/ 14 TestTTB Maurizio Vittuari 139 6
4 35/ 31/ 34 Tornado 2.0 Beppe Bezzi 138 1
5 40/ 42/ 18 Porch Swing 2 JKW 138 45
7 20/ 5/ 75 Die Hard P.Kline 135 38
12 19/ 7/ 74 Night Train Karl Lewin 130 24
13 38/ 46/ 17 Taking Names P.Kline 130 25
14 37/ 46/ 16 ltest Anders Ivner 128 53
16 24/ 20/ 56 Hector 2 Kurt Franke 128 52
18 34/ 40/ 26 Test z JKW 127 44
19 26/ 25/ 49 test9q Kurt Franke 126 11
20 31/ 36/ 33 Saveme! Maurizio Vittuari 126 51
21 22/ 19/ 59 Evolve-- v.1 JKW 124 33
24 22/ 23/ 55 test9p Kurt Franke 120 8
25 31/ 45/ 24 The Inferno Kurt Franke 116 2
______________________________________________________________________________
94 - What's No More
21 22/ 21/ 57 test4 Kurt Franke 122 2
21 31/ 35/ 33 Father & Son Maurizio Vittuari 127 110
21 36/ 45/ 19 myZizzor v2.0 Paulsson 127 12
21 35/ 49/ 16 Frontwards Steven Morrell 121 10
21 36/ 47/ 17 Porch Swing + Randy Graham 124 166
21 32/ 40/ 28 Armory - A5 Wilkinson 123 609
26 25/ 27/ 47 Fly Fishing w/Plastic Wor Karl Lewin 122 83
26 31/ 40/ 29 Phq Maurizio Vittuari 121 589
26 1/ 2/ 1 test lp 2 Beppe Bezzi 4 36
______________________________________________________________________________
What's Old
8 33/ 30/ 36 Torch t18 P.Kline 135 516
11 29/ 27/ 43 Jack in the box Beppe Bezzi 131 504
6 37/ 37/ 26 myVamp v3.7 Paulsson 137 484
22 36/ 47/ 17 Leprechaun on speed Anders Ivner 124 312
No other warrior is even above 200.
______________________________________________________________________________
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: Sun Dec 3 15:11:21 PST 1995
# %W/ %L/ %T Name Author Score Age
1 54/ 29/ 16 Fire Master v1.5 JS Pulido 179 26
2 52/ 32/ 15 Clear Sighted v1 JS Pulido 172 23
3 41/ 15/ 44 Paper8-IV 1.1 G. Eadon 167 6
4 47/ 34/ 18 dualstrat v.0.2 Brian Haskin 160 22
5 36/ 14/ 49 Schizo J. Doster 158 73
6 38/ 28/ 34 Hint Test v6 M R Bremer 148 9
7 42/ 37/ 21 Look J. Doster 147 70
8 29/ 14/ 57 Impfinity v3c11 Planar 144 85
9 28/ 13/ 59 Sheet 1.0 J. Doster 143 83
10 27/ 17/ 56 RingWorm_v2.4 Calvin Loh 136 71
11 21/ 9/ 69 RingWorm_v2.5 Calvin Loh 133 56
12 34/ 40/ 26 Good old DJN Test Karl Lewin 127 36
13 33/ 43/ 24 Test-G G. Eadon 124 1
14 26/ 28/ 46 Loh_tst_1.3 Calvin Loh 123 46
15 23/ 30/ 47 Serendipity A. Nevermind 115 10
16 31/ 48/ 21 Headache Andy Nevermind 113 7
17 27/ 50/ 23 Enlightenment III Scott Manley 104 12
18 10/ 16/ 74 Fortitude A. Nevermind 103 8
19 21/ 55/ 24 Joy! Kevin Deitchman 87 11
20 21/ 8/ 13 test L 9 Beppe Bezzi 78 24
21 22/ 11/ 10 Lurker 1.1 Kurt Franke 76 41
22 14/ 56/ 29 Frau Langmeier v2.4 Ian Oversby 73 4
23 17/ 5/ 20 Trapper 1.1 Kurt Franke 70 21
24 1/ 46/ 53 PlinyScan 1.1 G. Eadon 57 3
25 15/ 17/ 11 Paper Shredder 2.1 Kurt Franke 55 40
Twenty new entries on the hill this week.
Pulido has had solid lock on the hill lately. Fire Master was seen on the
'94 draft hill for awhile. The top 5 warriors are the same this week except
the absence of two of Franke's--Lurker and teamwork. Franke has been trying
to gain a foothold on the '94 hill. Hector 2 is in 16th place there. His
test program is at the _top_ of the '94x hill. Lewin has also claimed a
spot on the '94 hill with Night Train riding in 12th place. From the looks
of his score, it seems to be some kind of Die Hard derivative.
Bezzi's test L 9 is in 20th place. It was riding high on the '94 draft hill
if you can recall. If your program dominates the b hill and fares poorly on
the '94 draft hill, don't lose confidence. The composition of the hill has a
huge effect on how your warrior scores. Keep trying.
______________________________________________________________________________
Planar's Corner
The mathematics of imp steps
If you find this too mathematical, there is some redcode and cdb
macros in section 4.
0. Notations
------------
We'll talk about an imp number N and its corresponding imp step S in a
core of size C. This will give us an N-point ring or spiral with this
instruction:
MOV 0, S
N, S, and C must be such that, for some integer k,
N*S = k*C + 1 (1)
Given C and N, the game is to find an S that satisfies equation (1).
We're not really interested in the value of k. We'll call (N,S) an
imp pair in C. When C is implicit from the context, we'll simply
say that (N,S) is an imp pair.
We can see that equation (1) is unchanged when we exchange N and S:
if (N,S) is an imp pair, then (S,N) is also an imp pair.
1. Existence
------------
Let me state here the most important theorem of mathematics (as far as
imp rings are concerned, anyway), Bezout's theorem:
For any integers x, y, and g, there exists integers u and v such that
u*x + v*y = g (2)
if and only if g is a multiple of the Greatest Common Divisor of x and y.
I won't explicitely prove Bezout's theorem here. One of the
implications is easy, and the programs below are proof of the other.
Trust me, Bezout's theorem is true.
Let us take x = N, y = C, and g = 1, and equation (2) is almost the
same as equation (1): u*N + v*C = 1
If we can find u and v that satisfy this relation, then (u,N) is an imp
pair. Conversely, if (N,S) is an imp pair, then S*N + (-k)*C = 1, so 1
must be a multiple of GCD(N,C). This means that 1 = GCD(N,C), so N
and C have no common factors.
Our first result is:
In a core of size C, there is an imp step for N if and only if N and
C have no common factor.
Examples:
In a core of size 8192 all odd numbers are imp numbers.
In a core of size 8000 all numbers ending in 1, 3, 7, or 9 are imp numbers.
In any core size, (1,1) is an imp pair.
In any core size C, (C-1,C-1) is an imp pair.
In any odd core size C, ((C-1)/2,C-2) is an imp pair.
2. Unicity
----------
Assume that (N,S) and (N,T) are both imp pairs. We have:
(a) N*S = k*C + 1
(b) N*T = i*C + 1
Subtracting (b) from (a), we get:
(c) N*(S-T) = (k-i)*C
But N and C have no common factors, so S-T must be a multiple of C.
Thus, S = T + j*C: S and T are congruent modulo C. In a redcode
program they are equal, since all redcode arithmetic is done modulo C.
Our second result is:
In a core of size C, if both (N,S) and (N,T) are imp pairs, then S
and T must be congruent modulo C, and thus equal, as far as redcode
programs are concerned.
3. Computation with recursion
-----------------------------
Computing the u and v parameters of equation (1) can be done with
Euclid's extended algorithm. This is a simple extension of Euclid's
algorithm for computing the GCD. Given x and y, Euclid's extended
algorithm will compute g = GCD(x,y) and values for u and v that make
equation (1) hold.
The algorithm is extremely simple. You want to solve
u*x + v*y = g
if x > y, solve u'*y + v'*x = g and take u = v', v = u'.
if x = 0, take g = y, u = 0, and v = 1.
else compute q and r such that y = q*x + r and 0 <= r < x;
solve u'*r + v'*x = g;
take u = v'-q*u' and v = u'.
This is the same as Euclid's algorithm except that you keep track of the
quotients of the divisions and use them to compute u and v.
4. Computation without recursion
--------------------------------
There is a version of Euclid's extended algorithm that uses a loop
instead of the recursion. Instead of boring you with a description
of how it works, I'll simply give the code:
;redcode
;name Euclid's extended
;author Planar
;assert MAXLENGTH >= 15 && CORESIZE >= 16
; To compute the imp step for a given imp number,
; enter the imp number in the following line:
n equ 3
; The program terminates with gcd (n,CORESIZE) in the a-value at
; address 0. If this is 1, there is a spiral of size n, and the
; b-value at address 1 is the corresponding step.
; This program works in any core size, and it computes imp steps
; for that core size.
; Use the following cdb macro to display all the (N,S) imp pairs
; with N <= 100 in the current core size.
; imps=@ca z=0~!!~@ca z=z+1~@ed 0~dat z~@g~@0~if a==1~!!~@1~ca z,b~!1~@s~!100
; (and launch pMARS with -r 100)
org start
yx dat 0, n ; y = CORESIZE = 0
ab dat 1, 0
done equ ab
start sub.b yx, q ; (CORESIZE-n)
div.b yx, q ; ((CORESIZE-n)/n)
add.ab #1, q ; q = (CORESIZE-n)/n+1 = CORESIZE/n
loop mul.ab ab, q ; (a*q)
mov.x ab, ab ; (b); b' = a
sub.ba q, ab ; a' = b - a*q
mov.x yx, yx ; (y); y' = x
sub.ab yx, yx ; (y - x)
mod.ab yx, yx ; x' = (y-x) % x = y % x
jmz.b done, yx ; We're done if x = 0
mov.ab yx, q ; (y)
div.b yx, q ; q' = y' / x'
jmp loop
q end
This program is quite fast for every imp number N and core size C:
the number of loop iterations is bounded by 2*ceil(log N/log 2).
And this is a version entirely written in cdb macros:
euclid=@ca c=1,d=0~m eucloop
eucloop=!!~@ca t=d-y/x*c,d=c~@ca c=t,t=y%x~@ca y=x,x=t~if x!=0~!
imp-pairs=@ca s=$~@ca z=0~!!~@ca x=z=z+1,y=$+1~m euclid~if y==1~ca z,d~!s
(Do not try to tune your modem, the above is not line noise.)
This macro is much faster than Stefan's "imp-numbers", and it really
finds all imp pairs, whereas Stefan's misses a few.
To use it, save it in a file named "planar.mac", launch pMARS with the
core size you're interested in and type "m imp-pairs,planar.mac".
pMARS will list all (N,S) imp pairs. When N = S, for some reason
pMARS will only display one of them. (Bug or feature ?)
If you dislike negative numbers, just add CORESIZE to them. This is
an easy exercise in cdb programming.
If you only want the reasonable imp numbers, change the "s" at the end
of the last line into 100, and it will display only the imp numbers up
to 100.
5. Future work
--------------
There is a relation between imp numbers, Euclid's algorithm, and
bombing patterns. I'll try to work it out.
Now that I've been showing off my new cdb skills (two months ago I was
reading the docs for the first time), I guess I'll have to write the cdb
tutorial I promised to Beppe and Myer.
<Damien.Doligez@inria.fr>
______________________________________________________________________________
Extra Extra
You ask; we deliver. As promised the second and third place entries for
the tournament--round 7.
_____
run and hit by BEPPE BEZZI
I pswitch 5 attack modules:
A very fast static, non booted, vector bomber mov <x,y followed by a quick
boot of Tornado
A booted away multiprocess vector vampire to a large pit, self clearing and
brainwashing. This is followed by a forward running continual dat clear.
shot1 mov >0, @vector1-A0
This is the module that, hanging your warrior, cused my victory against you
after a few wins yours and ties.
A Tornado bomber with multipass clear.
A stone--juliet storm.
A paper module, in case someone uses a single pass clear.
I'm too surprised of how good I did. In fact, having nothing to test it
against I tried to mix as many thing I was able to.
-Beppe
;redcode-94
;name run and hit
;author Beppe Bezzi
;contact bezzi@iol.it
;NSFCWT round 7 entry
;assert CORESIZE == 8000 && MINDIST == 4000
;strategy flee then shot (not always :-)
org start
RES equ #0
STR equ #299
vstep equ 8
proc equ 8
disp1 equ 0
disp2 equ 70
disp3 equ 140
disp4 equ 210
disp5 equ 280
disp6 equ 350
disp7 equ 420
disp8 equ -100
spacer equ 0
A0 equ 1800
A1 equ 3740
A2 equ -2024
;Tornado equ
step equ 95
count equ 530
away equ 4456
djnoff equ -2000
;juliet equs
ptr EQU -1333
;Paper equ
dest0 equ 2200
dest1 equ 3740
dest2 equ -1278
range equ 933
;vector bomber equ
target equ (zero+4000+20)
zero
q for 20
dat q%7, 1
rof
point1 dat vbomb, vbomb+A0
vboot
spl 1, <pit
spl 1, <-10
spl bclear
spl 1, <-10
mov.i {point1,<point1 ;copy
jmp @point1
;break
shot1 mov >0, @vector1-A0
mov >0, @vector2-A0
mov >0, @vector3-A0
mov >0, @vector4-A0
mov >0, @vector5-A0
mov >0, @vector6-A0
mov >0, @vector7-A0 ;vector bombing
;break
jmp clear1+A2-A0-5
vbomb dat 10, 10
point2 dat clear1+4,A2
bclear mov {point2,<point2
clear1 dat #10, #10
spl #0, #100
mov -2, >-5
djn.f -1, <-100
vector1
s for proc
jmp pit-zero+4000-vstep-disp1,zero+4000+s*vstep+disp1
rof
vector2
s for proc
jmp pit-zero+4000-vstep-disp2,zero+4000+s*vstep+disp2
rof
vector3
s for proc
jmp pit-zero+4000-vstep-disp3,zero+4000+s*vstep+disp3
rof
vector4
s for proc
jmp pit-zero+4000-vstep-disp4,zero+4000+s*vstep+disp4
rof
vector5
s for proc
jmp pit-zero+4000-vstep-disp5,zero+4000+s*vstep+disp5
rof
vector6
s for proc
jmp pit-zero+4000-vstep-disp6,zero+4000+s*vstep+disp6
rof
vector7
s for proc
jmp pit-zero+4000-vstep-disp7,zero+4000+s*vstep+disp7
rof
vector8
s for proc
jmp pit-zero+4000-vstep-disp8,zero+4000+s*vstep+disp8
rof
pit mov -1000, <zero+3500
stp #0, #0
for 30
spl pit, <pit+1
spl pit, {pit+1
rof
for 50
dat zero, zero+500
rof
;****** START *********
start
res ldp.ab RES, #0 ;load result last match
str ldp.a STR, str1 ;load strategy in use
seq.ab #1, res ;check result, win OK
lost add.a #1, str1 ;lost change
mod.a #5, str1 ;secure jump
win stp.ab str1, STR ;save strategy
str1 jmp @0, vboot ;booted vector bomber + clear
dat 0, vector ;static vector bomber
dat 0, tornado ;
dat 0, juliet
dat 0, paper
for 20
dat 1,1
rof
tornado
mov bombm, away
for 12
mov {tornado,<tornado
rof
jmp @tornado,<tornado
warr
gate
nop split, 100 ;(clear start)
incr spl #3*step,#3*step
start1 add incr ,stone
stone mov step+jump, *(2*step)+jump
mov bombm, @stone
mov bombs, *stone
jump djn.f start1, <djnoff
clr mov *gate, >gate
djmp djn.f clr, {gate-20
wip dat <2667, <20
split spl #wip-gate, #wip-gate+5
bombs spl #step, #1
;break
bombm mov step, 1
for 50
dat zero, zero+500
rof
;juliet storm (sorry Myer I'm very short of time :-)
jgate dat <-445, <-446
s spl #445, <-445
spl #0, <-446
mov {445-1, -445+2
add -3, -1
djn.f -2, <-2667-500
mov 32, <-20
go dat #0, #ptr
juliet mov {-1, <-1
mov {-2, <-2
mov {-3, <-3
mov {-4, <-4
mov {-5, <-5
mov {-6, <-6
mov jgate, ptr+24
mov jgate, ptr+24
jmp @go, <4000
for 50
dat 1,1
rof
paper
spl 1, <300 ;\
spl 1, <400 ;-> generate 8 consecutive processes
spl 1, <500
silk spl @0, {dest0
mov.i }-1, >-1
silk1 spl @0, <dest1 ;split to new copy
mov.i }-1, >-1 ;copy self to new location
mov.i bomba, }range
mov {silk1, <silk2
silk2 jmp @0, >dest2
bomba dat <2667, <1
;linear vector bomber
vector
mov <target+50, target+100
mov <target+450,target+300
mov <target+150,target+200
mov <target+250,target+400
mov <target+350,target
mov <target+130,target+330
mov <target+230,target+430
mov <target+30, target+370
mov <target+270,target+70
mov <target+170,target+220
;now it's sure escaped
step1 equ 85
count1 equ 533
away1 equ -3750
mov bm, djmp1+away1+31
mov bd, djmp1+away1+32
mov *tptr, @tptr
mov {tptr, <tptr
mov {tptr, <tptr
mov {tptr, <tptr
mov {tptr, <tptr
mov {tptr, <tptr
mov {tptr, <tptr
mov {tptr, <tptr
jmp @tptr, <tptr
tptr dat djmp1, away1+djmp1
bomber1 mov bd+30, *stone1
mov bm+30, @stone1
stone1 mov *step1+2,*(2*step1)+2
add incr1, stone1
jump1 djn.b bomber1, #count1
incr1 spl #3*step1,#3*step1
clr1 mov -12, }bomber1+1
djmp1 djn clr1, <bomber1-5
bm mov.i #step1, 1
bd dat #step1, #1
a for (MAXLENGTH-CURLINE-1) /3
dat zero, a
dat 1,1
dat fine, a
rof
fine jmp zero
end
_____
Back Propagation by M R BREMER
No one knew how they would fare this round, myself included--nothing to test
against, only vague ideas of what to expect. Seeing Ivner's program after
the fact had me laughing. I have lots of respect for an author who will
intentionally cause his program to die. Anyways. I operated on two lines of
thought:
(1) everybody is going to vamp to a brainwashing pit.
(2) everybody is going to use switching.
After all, that's what I was doing 8). My program contains ten totally
independent vector vamping modules. Each module throws 10 fangs of the type
jmp 50 (50 addresses later) jmp 50 (+50 spaces) . . . jmp pit. I only need
two fangs per module instead of 10. Is this better? I have no idea.
A switching routine ( out of Jack in the box ) selects one of 5 strategies.
Each strategy executes module x and module x+5 yielding twenty fangs, 25
addresses apart. At the end of module x is code to cleanse module x+5 by
using a-indirect bombing on it. If I have been targeted by a fang, hopefully
I'll eliminate the pit with a mov #1, }1 bomb. I think it should have been
a mov#1, }0. Error #1. Next the code totally eliminates the module, over-
writing it with dat 1, 1s. Hopefully this will eliminate any stun attacks.
Of course module x+5 does the same of x.
Next launch two copies of Phoenix by Jippo. Plenty of a-indirect bombing.
Hard on imps. Of course I also have a backup continuous imp ring launcher
acquiried from Keystone.
By filling my code with pointers to locations 4000-4500, any bombing done
by the paper on my code will land on the enemy.
My pit is actually a jumper to two different pits. One brainwashes and the
other pit just keeps growing. Check it out at the very end of my source
listing. Neither of my pits core clear. Error #2.
It seems to have worked though. Unbelievable. Sorry it's so long. I'm not
good with macros.
-M R Bremer
;name Back Propagation
;author M R Bremer
;nsfcwt round 7
;contact <bremermr@ecn.purdue.edu>
;strategy create 10 independent vector fang bombing modules targeting
;strategy different parts of enemy code.
;strategy use that module from now until start to lose
;strategy launch Phoenix (from Jippo) as backup to all
;strategy i take _way_ too much time. i expect to get creamed
org start
STEP EQU 50
PSTEP equ 3520
NSTEP equ 1520
ESTEP equ 830
_STR EQU #336
_RES EQU #0
spl mimp+5334
launch2 spl -1 ; continuous imp-ring launcher
spl mimp+2667 ; activated by stone if needed
mimp mov.i #0,2667 ;
for 23
jmp 4000, start-5
rof
start
res ldp.ab _RES, #0
str ldp.a _STR, str1 ;load strategy in use
seq.ab #1, res ;check result
lost add.a #1, str1 ;lost or tie --> change
mod.a #5, str1 ;secure result
win stp.ab str1, _STR
str1 jmp @0, mod1
dat 0, mod2
dat 0, mod3
dat 0, mod4
dat 0, mod5
for 17
jmp 4075, start
rof
spl p1, 0
pit spl -1, 0
jmp p2, 0
for 11
jmp 4050, start
rof
p2 spl #0
spl -1
spl -2, 0
stp.b >-1, -1
jmp pit
for 19
jmp 4400, start
rof
mod1 spl mod6
mov.i m1fang, start+4000+9*STEP
mov.i fang1, start+4000+8*STEP
mov.i fang1, start+4000+7*STEP
mov.i fang1, start+4000+6*STEP
mov.i fang1, start+4000+5*STEP
mov.i fang1, start+4000+4*STEP
mov.i fang1, start+4000+3*STEP
mov.i fang1, start+4000+2*STEP
mov.i fang1, start+4000+STEP
mov.i fang1, start+4000
mov.i av1, *mod6
mov.i dat1, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang1 jmp STEP
m1fang jmp -1*(start+4000+9*STEP)+pit
av1 mov.i #1, }1
dat1 dat 1, 1
for 12
jmp 4250, 4350
rof
mod2 spl mod7
mov.i m2fang, start+4005+9*STEP
mov.i fang2, start+4005+8*STEP
mov.i fang2, start+4005+7*STEP
mov.i fang2, start+4005+6*STEP
mov.i fang2, start+4005+5*STEP
mov.i fang2, start+4005+4*STEP
mov.i fang2, start+4005+3*STEP
mov.i fang2, start+4005+2*STEP
mov.i fang2, start+4005+STEP
mov.i fang2, start+4005
mov.i av2, *mod7
mov.i dat2, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang2 jmp STEP
m2fang jmp -1*(start+4005+9*STEP)+pit
av2 mov.i #1, }1
dat2 dat 1, 1
for 19
jmp 4400, 4090
rof
mod3 spl mod8
mov.i m3fang, start+4010+9*STEP
mov.i fang3, start+4010+8*STEP
mov.i fang3, start+4010+7*STEP
mov.i fang3, start+4010+6*STEP
mov.i fang3, start+4010+5*STEP
mov.i fang3, start+4010+4*STEP
mov.i fang3, start+4010+3*STEP
mov.i fang3, start+4010+2*STEP
mov.i fang3, start+4010+STEP
mov.i fang3, start+4010
mov.i av3, *mod8
mov.i dat3, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang3 jmp STEP
m3fang jmp -1*(start+4010+9*STEP)+pit
av3 mov.i #1, }1
dat3 dat 1, 1
for 19
jmp 4400, 4090
rof
mod4 spl mod9
mov.i m4fang, start+4015+9*STEP
mov.i fang4, start+4015+8*STEP
mov.i fang4, start+4015+7*STEP
mov.i fang4, start+4015+6*STEP
mov.i fang4, start+4015+5*STEP
mov.i fang4, start+4015+4*STEP
mov.i fang4, start+4015+3*STEP
mov.i fang4, start+4015+2*STEP
mov.i fang4, start+4015+STEP
mov.i fang4, start+4015
mov.i av4, *mod9
mov.i dat4, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang4 jmp STEP
m4fang jmp -1*(start+4015+9*STEP)+pit
av4 mov.i #1, }1
dat4 dat 1, 1
for 19
jmp 4400, 4090
rof
mod5 spl mod10
mov.i m5fang, start+4020+9*STEP
mov.i fang5, start+4020+8*STEP
mov.i fang5, start+4020+7*STEP
mov.i fang5, start+4020+6*STEP
mov.i fang5, start+4020+5*STEP
mov.i fang5, start+4020+4*STEP
mov.i fang5, start+4020+3*STEP
mov.i fang5, start+4020+2*STEP
mov.i fang5, start+4020+STEP
mov.i fang5, start+4020
mov.i av5, *mod10
mov.i dat5, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang5 jmp STEP
m5fang jmp -1*(start+4020+9*STEP)+pit
av5 mov.i #1, }1
dat5 dat 1, 1
for 19
jmp 4400, 4090
rof
mod6 mov.i m6fang, start+4025+9*STEP
mov.i fang6, start+4025+8*STEP
mov.i fang6, start+4025+7*STEP
mov.i fang6, start+4025+6*STEP
mov.i fang6, start+4025+5*STEP
mov.i fang6, start+4025+4*STEP
mov.i fang6, start+4025+3*STEP
mov.i fang6, start+4025+2*STEP
mov.i fang6, start+4025+STEP
mov.i fang6, start+4025
mov.i av6, *mod1+1
mov.i dat6, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang6 jmp STEP
m6fang jmp -1*(start+4025+9*STEP)+pit
av6 mov.i #1, }1
dat6 dat 1, 1
for 12
jmp 4400, 4090
rof
c2 spl 1, <start+4000+123
mov.i -1, 0
mov.i -1, 0
mov.i -1, 0
phoenix spl @phoenix, PSTEP
mov.i }phoenix, >phoenix
nest spl @nest, NSTEP
mov.i }nest, >nest
egg spl @egg, ESTEP
mov.i }egg, >egg
mov.i ash, {3999
mov.i ash, }4001
ash dat.f <2667, <2*2667
dat.f <2667, <2*2667
dat.f <2667, <2*2667
dat.f <2667, <2*2667
dat.f <2667, <2*2667
for 12
jmp 4400, 4090
rof
mod7 mov.i m7fang, start+4030+9*STEP
mov.i fang7, start+4030+8*STEP
mov.i fang7, start+4030+7*STEP
mov.i fang7, start+4030+6*STEP
mov.i fang7, start+4030+5*STEP
mov.i fang7, start+4030+4*STEP
mov.i fang7, start+4030+3*STEP
mov.i fang7, start+4030+2*STEP
mov.i fang7, start+4030+STEP
mov.i fang7, start+4030
mov.i av7, *mod2+1
mov.i dat7, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang7 jmp STEP
m7fang jmp -1*(start+4030+9*STEP)+pit
av7 mov.i #1, }1
dat7 dat 1, 1
for 12
jmp 4400, 4090
rof
mod8 mov.i m8fang, start+4035+9*STEP
mov.i fang8, start+4035+8*STEP
mov.i fang8, start+4035+7*STEP
mov.i fang8, start+4035+6*STEP
mov.i fang8, start+4035+5*STEP
mov.i fang8, start+4035+4*STEP
mov.i fang8, start+4035+3*STEP
mov.i fang8, start+4035+2*STEP
mov.i fang8, start+4035+STEP
mov.i fang8, start+4035
mov.i av8, *mod3+1
mov.i dat8, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang8 jmp STEP
m8fang jmp -1*(start+4035+9*STEP)+pit
av8 mov.i #1, }1
dat8 dat 1, 1
for 12
jmp 4400, 4090
rof
mod9 mov.i m9fang, start+4040+9*STEP
mov.i fang9, start+4040+8*STEP
mov.i fang9, start+4040+7*STEP
mov.i fang9, start+4040+6*STEP
mov.i fang9, start+4040+5*STEP
mov.i fang9, start+4040+4*STEP
mov.i fang9, start+4040+3*STEP
mov.i fang9, start+4040+2*STEP
mov.i fang9, start+4040+STEP
mov.i fang9, start+4040
mov.i av9, *mod4+1
mov.i dat9, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang9 jmp STEP
m9fang jmp -1*(start+4040+9*STEP)+pit
av9 mov.i #1, }1
dat9 dat 1, 1
for 12
jmp 4400, 4090
rof
mod10 mov.i m10fang, start+4045+9*STEP
mov.i fang10, start+4045+8*STEP
mov.i fang10, start+4045+7*STEP
mov.i fang10, start+4045+6*STEP
mov.i fang10, start+4045+5*STEP
mov.i fang10, start+4045+4*STEP
mov.i fang10, start+4045+3*STEP
mov.i fang10, start+4045+2*STEP
mov.i fang10, start+4045+STEP
mov.i fang10, start+4045
mov.i av10, *mod5+1
mov.i dat10, >-1
djn.b -1, #10
spl launch1
jmp launch2
fang10 jmp STEP
m10fang jmp -1*(start+4045+9*STEP)+pit
av10 mov.i #1, }1
dat10 dat 1, 1
for 12
jmp 4400, 4090
rof
launch1
spl c2
spl 1, <start+4000+123
mov.i -1, 0
mov.i -1, 0
mov.i -1, 0
phoenix2 spl @phoenix2, PSTEP
mov.i }phoenix2, >phoenix2
nest2 spl @nest2, NSTEP
mov.i }nest2, >nest2
egg2 spl @egg2, ESTEP
mov.i }egg2, >egg2
mov.i ash2, {3999
mov.i ash2, }4001
ash2 dat.f <2667, <2*2667
dat.f <2667, <2*2667
dat.f <2667, <2*2667
dat.f <2667, <2*2667
dat.f <2667, <2*2667
for 12
jmp 4400, 4090
rof
spl #0, 0
p1 spl -1, 2
mov -1, >-1
______________________________________________________________________________
The Hint
This will be the last hint in Core_ Warrior_'s attempt to introduce basic
warrior strategies. And we've saved the best for last, at least in my
opinion. Vampires such as Sucker, Twilight Pits, Request, Snake, Pyramid,
and probably a host of others, used to rule the hills of old. But then came
A-indirect addressing making anti-vamping much easier to do. Silk Warrior
by Jippo incorporated a-indirect bombing in his replicator, which greatly
improved a vampire resistance. Soon the slavers were dropping of the hills
in droves. Vampires were thought to be obselete, but Paulsson's myVamp
continues to hold a top spot on the '94 hill so vamping could be your next
effective strategy. Or has Paulsson come up with something unexpected? Code
for a version ( perhaps the one on the hill ) is included for your review.
Magnus is very busy and didn't have time to offer his comments. I would if
I could, but someone has my corewar disk (I don't own a computer), plus the
server for my university account can't access the internet right now. *sigh*
The following comments were taken from a _PUSH OFF_ by P.Kline. Hope he
doesn't mind. Edited for clarity ( just kidding, sheesh ).
Basically a pit-trapper works by making your processes jump into a rapid-
splitting, core-clearing, self-destroying, nasty bit of code. You are forced
to split many many processes, thereby robbing your healthy processes of time.
You are forced to erase the core, destroying any surviving components. And
you finally erase the pit itself, killing your remaining processes.
Vampire strategy goes back to one of the standard battle programs, 'Vampire',
and was used successfully by Eugene P. Lilitko's Cowboy program to win the
1988 tournament. Here's a modern version of Vampire:
fang jmp trap,4
inc spl #3364, #-3364 ; start here
add inc,fang
mov fang,@fang
jmp -2
trap mov 10,<-10 ; Of course this wouldn't have a prayer in stunning
spl -1 ; today's paper. How about mov, spl -1, spl -1,
jmp -2 ; spl -1, jmp?
All vampires eventually cause their own processes to jump into the trap to
speed up the core clear (and to guarantee one).
Vampire's strengths include small size, rapid bombing (with fangs), and
complete core-clear. Their most important feature, however is enslaving the
opponent's processes to do their dirty work.
Variations include booting the code to separate the bomber from the trap,
putting the fang in different locations, multiple spl's to more rapidly
overpower the opponent, and 'gates' to stop imps.
Areas to explore:
What else can the pit do?
Using multiple fang-bombers to better deal with stones.
Evolving the fang-bomber into a repeating core-clear.
_____
myVamp by MAGNUS PAULSSON
;redcode-94
;name myVamp v3.7
;author Paulsson
;strategy Throw jmp's around at 2/3c.
;strategy Then a spl/spl/dat and djn coreclear.
;strategy All in the tiny space of 14 bytes!
;strategy Submitted: 30 september 95
;assert CORESIZE > 1
;kill myVamp
;The vamp bomb is the best bomb, unless it fights a special oponent with
;anti vamp, most programs on the hill cant affod anti vamp. Normal vamp
;adds +step to a field, -step to b field of jmp then uses as bomb (once, to
;b field) and pointer, myVamp also scans the a-field and bombs if not zero.
step equ 3024 ;good step, mod 16
boot equ 8*200 ;offset so that bomb/scan will be mod 8
org start
trap jmp 11 ;use a as pointer to the pit.
i for 10
dat.f 0, 0
rof
spl #0,<trap ;real trap a bit away from
spl 5,0 ;trap line. note that b-field
spl #0,0 ;is none zero as the enemy is
spl #0,0 ;trapped
spl #0,0
spl #0,0
spl #0,0
spl #0,0
spl #0,0
jmp -4,0
i for 39-16
dat.i #1, i
rof
start mov.i }ptr,>ptr ;boot away
i for 13
mov.i }ptr,>ptr
rof
jmp st+boot,1 ;start fighting
ptr dat.f c2,boot+c2
c2 spl.f #b1, b2+2 ;pointer for coreclear, spl incase of djn'ed
st add.f s, c
mov.i c, @c ;bomb
jmz.a st, *c ;scan other place
mov.i @0,*c ;bomb
mov.x *c,*c ;make sure bomb points right
exit jmz.b st, trap-boot ;start clear if trap is hit or enemy trapped
b1 spl #b2-c2, b2-c2+2 ;coreclear, same as myZizzor
mov.i *c2, >c2
djn.f -1, {c2-b2+c2-2
b3 dat.f <-2666, <2667 ;a little something for Die Hard
;most imps wont survive anyway.
b2 spl #b3-c2, b2-c2+2
c jmp *trap-boot, 0 ;bomb and counter, standard vamp
s dat.f -step, step ;step
i for 13
dat.i #1,1 ;filling it up
dat.f 1,#1
rof
We will be taking your suggestions for hints. If any of you beginner's ( or
veterans for that matter ) would like to see a topic covered in more detail,
just ask.
______________________________________________________________________________
Questions? Concerns? Comments? Complaints? Mail them to:
Beppe Bezzi <bezzi@iol.it> or Myer R Bremer <bremermr@ecn.purdue.edu>
|
