Sections
Hills
Infinite Hills
Tournaments
Software
Evolving
Optimizer
Community
Newsletter
Discussion
History
Sections
 
For Beginners
First Steps
FAQ
Guides
Lexicon
Benchmarks
For Beginners
> Home > Tournaments > Corewar Single Elimination Championship 2005> Semifinal

Semifinal


Some comments will be added later

Deadline: Saturday, November 19, 2005, 9pm CET


The Rules

This time the coresize is larger than usual. But that's not all. Your warrior must start with the following sequenze at pGo:
pGo   spl   1,  0
      spl   1,  0
      spl   1,  0
      spl   1,  0
      spl   1,  0
      mov.i #0, {0
      
;your warrior code starts here:
You'll have 32 processes for the battle then. As the spl-instruction is forbidden in your code that are all processes you will have ;-)

This time p-space is allowed without any restrictions.

The parameter setting are:

Coresize16384
max. Cycles150000
max. Processessee special rules
max. Warrior Length200
min. Distance200
p-spaceallowed
Restrictionsspl is forbidden, see special rules
Rounds1000


The commands for pmars is:

pmars -s 16384 -c 150000 -p 32 -l 200 -d 200 -r 1000


Fixtures and Results

Roy van Rijn vs. S. Fernandes


Fizmo's Comment


Again Roy is the clear favourite in this match. Nevertheless it will be pretty hard for him, because S. Fernandes has a very good run in this tournament as well.

The Comments


Roy van Rijn:
Ok, 32 processes and no way to get more...
Not even a split, so we can't make a paper :-(

I don't know SF very well, but I think he makes a lot of scanners. 
This oneshot should kill scanners...

This round its easier to make a clear
oneshots are good against clear too.

Did I make the right decision? Will I really get in the finals!?? 
Now there is nothing left to do but hope for the best! 
May the best coder & anticipator win!
S. Fernandes:
Roy seems to instinctively place numerous processes into a 
variety of components scattered through core, with deadly 
efficiency. Roy shouldn't have any problems adapting to 
this round's restrictions and will probably come up with a 
stone/clear/imp. 
I've never written a multi-component warrior before. I'm 
going to have to learn quickly to stand a chance against 
Roy. I'll try a stone/clear/imp. 
processes - 13 stone, 11 clear, 8 imp 
The Results


Roy van Rijn sent a oneshot, while S. Fernandes a stone/clear/imp. It was a very close match with a slight advantage for the oneshot.
#%W%L%TName AuthorScore%
148.447.4 4.2 AlacrityRoy van Rijn149.40100.0
247.448.4 4.2 SandstoneS.Fernandes146.40 98.0


Battle SF1Result
Roy van Rijn149.4
S. Fernandes146.4


Chip Wendell vs. Zul Nadzri


Fizmo's Comment


Chip is the clear favourite in this match after defeating John Metcalf in the previous round. Could he defeat now also Zul Nadzri?

The Comments


Chip Wendell:
32 processes but no SPL? Ooh, this is gonna be interesting. And fun.

First off, I can rule out several strategies:
- No papers. Hoppers are possible, but they're too vulnerable to stones.
- No scanners, because they won't have any papers to prey on.
- No qscans, without throwing away all but one process.
- P-space? Again, hard to do when you're starting with 32 processes.
- Vampires? Probably not, because prisoners can't be stunned.

That just leaves stones and imps. Therefore, a stone/imp will probably be 
best. Unfortunately, this conclusion is pretty obvious, so I'll need to 
make sure that my entry can beat other S/I's.

By now, Elven King has proven itself as a very versatile and deadly 
warrior, so I'll again start with that. I need to make lots of adjustments 
this time, though: new coresize, no SPL, a booting scheme that can redirect 
a lump of 32 processes. Then, I start sifting through variations: 3-point 
or 5-point imps, A- or B-imps, various anti-imp bombs, different process 
balance between stone and imp. I'm sure learning a lot about imps and how 
to kill them!

Finally, I run a big round-robin among all the variations, and pick the 
one that beats all the others. After some final tweaking, and filling in my 
extra operands with core-coloring, it's ready. OK, Zul, do your worst!
Zul Nadzri:
Just got out from the hospital after minor surgery, virtually 2 days to prepare.
Pick up my recently kicked off the hill warrior 'The 9th Process'.
Changed the step and add a 3-line bomber.
Get a pspacer, believe whatever in there, and sent to Fizmo.
No time for coding in J-factor (John) but surely could give a good name.
BTW, no valid excuse for losing this round except Chip is more skillful than me.
The Results


Zul Nadzri sent a p-warrior switching between dwarf and scanner, while Chip Wendell sent a stone/imp. The Zul's scanner in his warrior can't defeat the stone/imp which results in a victory for Chip Wendell.

#%W%L%TName AuthorScore%
161.633.6 4.8 Goblin PatrolChip Wendell189.60100.0
233.661.6 4.8 ~~ The Relaxing Process ~~Zul Nadzri105.60 55.7


Battle SF2Result
Chip Wendell189.6
Zul Nadzri105.6



The Second Chance Tournament Tree Fixtures

There is a second chance for all players who wants to continue the CSEC2005:

As there are just three player left I decided to let them run round-robin, while the best two will play in the Second Chance finals for the 5th place in the CSEC2005.

Nenad Tomasev vs. datagram vs. inversed


The Comments


inversed:
I was deciding between submitting this, 
pspacer and dodger. This thing seem to be
most balanced. It consists of 3 stones, 2 clears
and imp spiral. My biggest fear are dodgers.
Nenad Tomasev:
the start of the code, regulated by the rules of this round, is a little 
tricky ... noone needs 32 procs in the same spot, so some separation is 
necessary. I think that I've found a nice and quick way to do it. Of 
course, one can choose to kill 31 porcs and to make a one process scanner... 
or maybe leave 2 procs and make a double scanner - better against stones?
since there is no spl - papers are impossible - and some pseudopapers, like 
hoppers, completely useless and expencive in terms of code size and speed. 
since there won't be any papers, the strength of scanners drops, because 
they've lost their primary source of food. stones can achieve greater speed 
with smaller code. thus, stones are, IMHO better here.
 
since the coresize allows imps, it would be a shame not to use them :). 
Especially since imps are very hard to kill in core of this size, without 
previous stun by spl's or a really long bombing phase - which is hard to 
accomplish, since you can't allow a selfhit by a dat, like in most good modern 
stones. I think that it is a little better to use jmp bombs, since opponent 
may be using several attack modules, and such bombs would slow down the 
remaining functional parts. ... of course, one can attempt to make a bomb 
dodger here, it could score well. one could also stick to coreclears... I 
decided not to do that. I've made a stone imp, whose stone uses jmp #1, >1 
(in case someone is crazy to use an unprotected djn stream) and jmp #step, >1 bombs. 
and operates in 0.6c speed, which is, I think, fast enough. I also give some 
procs to imps - 12. the stone gets 20. I use a vector table launch to set 
everything up and running and to make sure procs in the stone are well placed. 
The stone uses a mod-8 pattern with the best mod-8 step according to optima 
score. I think that it will be hard to kill this one... I don't think that it 
can lose much, although I'm certain that it can lose - after all, it isn't 
optimaxed (of course).
 
I spent 15 mins per warrior in the last few rounds. 45 mins this time. So, I 
expect a nice score - even if it loses, I expect to see a tough fight.
 
a thought: maybe I should've airbagged the stone
a thought: I think that using pspace is bad, because you need to use it before 
you kill some procs, since you don't know if you'll need them and that slows 
down the "thinking" process... boot and everything else wastes enough time 
already, so no need to give your opponent the edge if you don't really have to
The Results


Nenad and inversed clearly reaching the finals of the Second Chance defeating datagram who have played a great tournament with his final 7th place in the CSEC2005.

#%W%L%TName AuthorScore%
164.6 1.633.9 HalfOfMoonNenad Tomasev227.55100.0
227.836.835.5 No time to make up a good nameinversed118.70 52.2
321.775.7 2.7 Charlie Giteaudatagram 67.75 29.8


BattleResult
Nenad Tomasev227.55
inversed118.70
datagram67.75


The Off-Topic Round-Robin Results

#%W%L%TName AuthorScore%
150.311.138.6 HalfOfMoonNenad Tomasev189.48100.0
235.327.137.6 Goblin PatrolChip Wendell143.47 75.7
334.827.737.6 No time to make up a good nameinversed141.87 74.9
446.051.5 2.5 AlacrityRoy van Rijn140.53 74.2
530.832.536.7 SandstoneS.Fernandes129.07 68.1
639.053.9 7.1 ~~ The Relaxing Process ~~Zul Nadzri124.18 65.5
732.865.2 2.0 Charlie Giteaudatagram100.33 53.0
2002-2005 corewar.info. All Rights Reserved. Logo C. Schmidt
CSEC 2005
Home
Schedule
Qualification
Round 2
Quarterfinal
Semi Final
Final
CSEC 2005
 
Player Profiles
Jens Gutzeit
X-man
el kauka
fiveop
Nenad Tomašev
John Metcalf
LAchi
Zul Nadzri
S. Fernandes
Roy van Rijn
German Labarga
Michal Janeczek
brx
Mizcu
Sascha Zapf
Will Varfar
Chip Wendell
inversed
datagram
John K. Lewis
Simon Wainwright
Player Profiles