A midweek review of Corewar
November 4, 1993
-------------------------------------------------------------------------------
I. The Standings:
# %W/ %L/ %T Name Author Score Age
1 47/ 40/ 13 Iron Trap Wayne Sheppard 154 53
2 47/ 42/ 11 Dragon Spear II c w blue 153 11
3 42/ 32/ 26 Winter Werewolf 3 W. Mintardjo 152 678
4 42/ 35/ 22 Keystone t13 P.Kline 149 207
5 35/ 21/ 45 Cannonade P.Kline 148 203
6 43/ 38/ 19 Fast Food v2.1 Brant D. Thomsen 148 17
7 35/ 22/ 43 One Bad Apple v1.2 Mike Nonemacher 147 7
8 43/ 41/ 15 Twilight Pits 8 W. Mintardjo 146 104
9 39/ 31/ 30 Vagabond P.Kline 146 35
10 46/ 46/ 8 Agony 6.0 Stefan Strack 145 312
11 41/ 38/ 22 Distance v8.0 Brant D. Thomsen 144 89
12 42/ 41/ 17 sub-type-cmp c w blue 143 2
13 34/ 26/ 40 Imprimis 7 P.Kline 143 600
14 44/ 45/ 11 Impurge Fredrik Ohrstrom 142 684
15 35/ 30/ 35 test James Layland 140 125
16 32/ 25/ 43 Impact v1.0 Anders Ivner 139 497
17 30/ 23/ 47 pMARS pMARS project 137 409
18 32/ 28/ 40 Night Crawler Wayne Sheppard 136 1360
19 40/ 46/ 14 dproba nandor sieben 135 14
20 35/ 36/ 29 Yop La Boum v2.1 P.E.M & E.C. 133 1
21 2/ 98/ 0 test P.Kline 7 0
-------------------------------------------------------------------------------
II. The Basics:
-Core War Archives, including many helpful articles, warrior source
code, and reliable emulators, are available via anonymous FTP
at soda.berkeley.edu in pub/corewar.
-FAQ for this newsgroup is available via anonymous FTP at
rtfm.mit.edu as pub/usenet/news.answers/games/corewar-faq.z
-------------------------------------------------------------------------------
III. The Scoop:
My oh my! Another seasoned battle-program bit the dust this week, as
N. Sieben's 'ttt' was pushed off at the age of 539. Unpublished, ttt
is suspected to be a lean-and-mean stone-imp that proved an intractable
opponent. Have to give credit to W. Sheppard's Iron Trap, which is a
for-real, take-me-seriously, imp-killer - as his own Night Crawler is
discovering :-)
But Iron Trap is not the only shiny new thing on KotH. B. Thomsen's
Fast Food item is proving tough too. Introducing a very fast
initial scan, ala QuickFreeze, Fast Food's vampire-attack is giving
some players fits.
M. Nonemacher's 'One Bad Apple' is struggling to find its niche on KotH.
With a gate-busting spiral(s), One Bad Apple is trying to shake up
the Hill and see who falls off :-)
And another mystery resolved. (Maybe I need a Mystery-of-the-Week section)
My apologies to Michael Constant. I had 'submitted' several entries to the
'94 Hill at Stormking, but never heard anything back, so gave it up.
Today I discovered that my submittal-program is buggy and nothing
ever left my computer. Hopefully this is fixed, and we can start
including the '94 rankings in _Push Off_.
-------------------------------------------------------------------------------
IV. The Outlook:
1 31/ 16/ 53 One Bad Apple Mike Nonemacher 147 1
1 48/ 40/ 12 Iron Trap Wayne Sheppard 155 1
2 38/ 34/ 28 Vagabond P.Kline 142 1
3 33/ 20/ 46 One Bad Apple v1.2 Mike Nonemacher 146 1
4 43/ 45/ 12 Iron Gate 1.2 Wayne Sheppard 140 14
5 40/ 42/ 18 Distance v8.0 Brant D. Thomsen 139 1
10 42/ 45/ 13 sub-type-cmp c w blue 139 1
-------------------------------------------------------------------------------
V. The Quick Look:
11 42/ 45/ 13 Dragon Spear II c w blue 139 1
12 39/ 41/ 20 Fast Food v2.1 Brant D. Thomsen 137 1
12 40/ 46/ 14 Stoned Again c w blue 135 1
16 31/ 30/ 39 Deck of Many Things c w blue 131 1
16 35/ 37/ 28 Match Stick c w blue 132 1
18 32/ 39/ 29 Yop La Boum v1.0 P.E.M & E.C. 126 1
18 38/ 46/ 16 dproba nandor sieben 130 1
19 30/ 33/ 36 Hydra 2 Stephen Linhart 128 1
19 33/ 37/ 30 Herem VI-test Anders Ivner 129 1
19 38/ 52/ 10 All That Glitters v5.6 Mike Nonemacher 124 1
20 19/ 17/ 65 Imps! Imps! Imps! Steven Morrell 120 1
20 27/ 27/ 46 test Mike Nonemacher 126 1
20 33/ 56/ 11 Nova 2 Jay Han 110 1
20 34/ 51/ 15 Red Baron ][ v2.1 Mike Nonemacher 118 1
20 36/ 52/ 12 Juggernaut v1.5 Anders Ivner 121 1
20 37/ 47/ 15 Dagger v3.1 Michael Constant 128 1
21 0/ 47/ 53 wipe Peter Nyman Hansen 53 0
21 2/ 81/ 17 chaos Rob Weir 23 0
21 6/ 61/ 33 copybomb Peter Nyman Hansen 51 0
21 6/ 68/ 26 SuperImp 1.0 Jonathan Roy 43 0
21 8/ 35/ 57 bangle 4.1 Steve Gunnell 82 0
21 8/ 90/ 2 Trash Mitch Burton 25 0
21 10/ 59/ 31 cloner Peter Nyman Hansen 61 0
21 10/ 70/ 19 m2 program Ray Hann 51 0
21 10/ 86/ 4 Ponderous Mitch Burton 34 0
21 13/ 84/ 3 disrupta 1.0 Fredrik Appelberg 41 0
21 14/ 72/ 14 Mammascan1.0 Fredrik Appelberg 56 0
21 14/ 77/ 9 cascade Steve Gunnell 51 0
21 17/ 61/ 22 Impervious Jay Han 73 0
21 17/ 82/ 1 Unknown Fredrik Appelberg 52 0
21 18/ 81/ 1 fastblitza 1.0 Fredrik Appelberg 55 0
21 19/ 71/ 10 FireWall Bryan Kilian 66 0
21 23/ 59/ 17 K1 Karl Lewin 87 0
21 24/ 67/ 9 Multiblitza 1.0 Fredrik APpelberg 82 0
21 25/ 39/ 37 Incrimination v3.2 Brant D. Thomsen 111 0
21 28/ 52/ 19 Sunburst 40 Jay Han 104 0
21 28/ 55/ 16 death Steven Morrell 101 0
21 28/ 62/ 10 Morannon 4.1 J Kyle Kelso 94 0
21 29/ 56/ 15 test Anders Ivner 103 0
21 29/ 61/ 10 Sling 5 Jay Han 96 0
21 29/ 63/ 8 Paratroops v3.5 Mintardjo W. 95 0
21 32/ 54/ 14 Nova 19 Jay Han 110 0
21 33/ 53/ 14 Permafrost 1 Jay Han 112 0
21 34/ 46/ 20 eproba nandor sieben 122 0
21 34/ 46/ 20 testing a B scanner c w blue 122 0
21 34/ 49/ 17 Early Bird c w blue 120 0
21 35/ 52/ 13 test P.Kline 118 0
21 37/ 52/ 12 sub-type-os c w blue 121 0
-------------------------------------------------------------------------------
VI. The Hint:
(This is a follow-up to P.E.M. and E.C.'s posting of their 2668-imp)
This is great! I'm glad I described Cannonade's spiral in such
a vague fashion, because what you have is very different and
interesting. In fact Cannonade HAS to launch a 2667-imp to follow the
2668 because the standard gate always kills the 2668. Yours
doesn't need the 2667. (Which suggests a way to beat Iron Trap which
is currently whacking Cannonade :-)
Here is a portion of your launch, commented to show the order
of execution, and the mov-destination for each process:
spl 8
spl 4 ; mov-from mov-to
spl 2 ; -------- ------
jmp imp ; N N+2668 a
jmp imp1+1 ; N+2668 N+2668+2668 b
spl 2
jmp imp2 ; N+2667+2667 N+2 c
jmp imp+1 ; N+1 N+1+2668 d
spl 4
spl 2
jmp imp1 ; N+2667 N+2667+2668 e
jmp imp2+1 ; N+2667+2668 N+3 f
spl 2
dat #0
Note that only a and e set up the next process. The rest leave
a gap between moving the imp and executing it. A few tests
showed in some cases there are up to 16 cycles between moving
and executing the imp, which leaves you more vulnerable to
bombing.
If you are combining your imps with a stone, be sure
to split off the stone process before entering the spiral-launch.
Don't just take one of the extra processes out of the end
of the launch code, it may leave huge gaps in your spiral
execution sequence and weaken it greatly.
For the same reason, you might try using SPL 32 and SPL 16 to
start the other two spirals, rather than taking processes out of
the end of each preceding one.
On the other hand, you partly overcome the execution gaps by
interleaving three spirals, creating some redundancy.
On the other-other hand, your launch is so complex, I may be mis-
understanding it, so feel free to rebut :-)
Here's a stripped-down version of Cannonade's 2668-launch:
start mov imp1,imp1+5
mov imp1,<start
mov imp1,<start
mov imp1,<start
mov imp1,<start
lnch2 spl lnch2b
lnch2a spl lnch2ab
lnch2aa spl 2
jmp imp1+(2668*0)
jmp imp1+(2668*1)
lnch2ab spl 2
jmp imp1+(2668*2)
jmp imp1+(2668*3)
lnch2b spl lnch2bb
lnch2ba spl 2
jmp imp1+(2668*4)
jmp imp1+(2668*5)
lnch2bb spl 2
jmp imp1+(2668*6)
jmp imp1+(2668*7)
imp1 mov 0,2668
This is a true spiral where every instruction moves the imp to
the very next location to be executed (except the last one).
It does have a weakness in pushing four unprotected instructions
along ahead of it. I overcome that by 'braiding' a pair of spirals
two instructions apart, making it much stronger and more resistant
to spl-jmp attack. A single spiral needed 4-5 'twitches' to
clear a spl-jmp bomb, but against most scanners there was never
time for that. The braided pair needs only 2-3 'twitches' to
overcome spl-jmp's and survive the opponent's core-clear.
BTW, I advise taking a look at Chimera for a very neat launch-time
trick that is incorporated into Cannonade.
-------------------------------------------------------------------------------
VII. The End:
Paul Kline
pk6811s@acad.drake.edu