Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

The Corewar DynaHill

5 views
Skip to first unread message

fizmo

unread,
Jan 9, 2009, 2:13:36 PM1/9/09
to
Hello all,

I'm planning to open a new hill called The Corewar DynaHill. It uses a
different (dynamic) hill concept, which has some similarities to the
ranking system used in Japanese professional sumo sport.

http://www.corewar.info/dhill/


The major differences are:


Unlimited hill size:
--------------------
Every new warrior will enter the DynaHill at the bottom. No warrior
will be pushed off. A player can retire his own warriors if he wishes.
A replacement is not allowed.


Battle concept and Scores:
--------------------
Every warrior fights 15 neighbouring warriors on the hill in a 250
round battle (-r 250). Depending on the scores of the 250 rounds a
victory or a defeat is declared. In case of a tie a victory is
declared randomly. This means after fighting the 15 opponents the
overall score can range between 15 victories (15 - 0) and 15 defeats
(0 - 15). The higher the number of victories, the higher the warrior
climbs up the hill for the next run. The higher the number of defeats,
the further the warrior drops down the hill. No matter what the
result, it is most likely that on the next hill run the warriors have
to face different opponents.

And as the DynaHill will run in certain time intervals - no matter if
new warriors enters the hill or not - you can watch your warriors
climbing up the hill from hill run to hill run until they hit the wall
or become the King of the DynaHill.


94draft AND 94LP warriors can be used:
--------------------------------------
Another major difference to the existing hills is, that both warrior
types - 94draft warriors and lp warriors - can co-exist on a single
hill, as the DynaHill sets the max. processes dynamically depending
who fights who.

94draft vs. 94draft: -p 8000
94LP vs. 94LP: -p 8
94draft vs. 94LP: random -p between 40 and 120

For that reason every warrior entry must start their code either with

;redcode-94

or

;redcode-lp

to determine under which rules the warrior will fight.

Restrictions:
-------------
PIN and MAXPROCESSES are disallowed.

Hints:
------
From my experiments the nop warriors (warriors which didn't take
advantage of the p-space) tends to form clusters on such a hill. And
as the hill becomes larger, the more difficult it is to climb up to
the top with a 'dumb' 94 nop warrior. P-warrior should have an
advantage, especially if they combine draft and lp components.

digital.w...@googlemail.com

unread,
Jan 9, 2009, 3:20:40 PM1/9/09
to
On Jan 9, 7:13 pm, fizmo <fizmo_mas...@yahoo.com> wrote:
> I'm planning to open a new hill called The Corewar DynaHill. It uses a
> different (dynamic) hill concept, which has some similarities to the
> ranking system used in Japanese professional sumo sport.

I'm just figuring out what to enter :-) Sounds like it could be fun.

John

--
http://corewar.co.uk

Skybuck Flying

unread,
Jan 9, 2009, 3:28:02 PM1/9/09
to
> Sounds like it could be fun.

Yup sounds like fun.

Me all for it :)

Bye,
Skybuck =D


neog...@gmail.com

unread,
Jan 9, 2009, 4:25:39 PM1/9/09
to
Wow, very good news!

I just have a few questions...
Do you have to move warriors from yor mail to the hill and then update
the web page with the results? Maybe it is a bit nasty for you...
What is the run frequency?
from the website:"...P-warrior should have an advantage, especially if


they combine draft and lp components."

I suppouse having ;redcode-94 or ;redcode-lp on top doesn't let a p-
warrior to change the settings, right?
And the most important, when it is going to be up? :D

I'm figuring what to do.

digital.w...@googlemail.com

unread,
Jan 9, 2009, 5:54:25 PM1/9/09
to
On Jan 9, 9:25 pm, neogry...@gmail.com wrote:
> from the website:"...P-warrior should have an advantage, especially if
> they combine draft and lp components."
>  I suppouse having  ;redcode-94 or ;redcode-lp on top  doesn't let a p-
> warrior to change the settings, right?
> And the most important, when it is going to be up? :D
>
> I'm figuring what to do.

I guess you'd have to specify one or the other.

Either lp components with components for 40-120 processes, or '94draft
components with components for 40-120 processes.

John
--
http://corewar.co.uk

saye...@gmail.com

unread,
Jan 9, 2009, 11:26:34 PM1/9/09
to
I assume this is 8000 cores. Full command specs is appreciated.

fizmo

unread,
Jan 10, 2009, 8:23:33 AM1/10/09
to
On 9 Jan., 22:25, neogry...@gmail.com wrote:
> Wow, very good news!
>
> I just have a few questions...
> Do you have to move warriors from yor mail to the hill and then update
> the web page with the results? Maybe it is a bit nasty for you...

Yes, so far I have to copy the entry into a certain directory. The
script will do the hill run update the ranking and generate a result
html and a hmtl with the new ranking.

> What is the run frequency?

I am thinking of run the hill once per day. But depending on the
amount of entries I let it run in the beginning more often

> from the website:"...P-warrior should have an advantage, especially if
> they combine draft and lp components."
>  I suppouse having  ;redcode-94 or ;redcode-lp on top  doesn't let a p-
> warrior to change the settings, right?

Hehe, that is the tricky thing of it, you never know which strategy
yor're facing.

For example you have a ;redcode-94 warrior which is a defensive 94
paper and let it then faceing an opponent which is a LP paper or LP
scanner running then under max. processes somewhere between 40 and
120 ;-)

> And the most important, when it is going to be up? :D

It's up already, I just need to wait for enough entries to fill the
hill before I can start the first hill run. 16 warriors are minimum
for that

fizmo

unread,
Jan 10, 2009, 8:40:36 AM1/10/09
to

The max. process of a battle between two warriors depends on the which
setting you used for the warrior ;redcode-94 or ;redcode-lp

The following max. process are used for the battles:

warrior A vs. warrior B max. process
;redcode-94 vs. ;redcode-94 8000
;redcode-lp vs. ;redcode-lp 8
;redcode-94 vs. ;redcode-lp randomly between 40 and
120

The reason that I make the max. process of 94 vs. lp battles randomly
is that I try to keep the hill more dynamic. Well, let's think of a
slightly varying daily fighting condition of a warrior.

Btw, the max. processes is kept unchanged during the 250 round battle.
It didn't change from round to round but from battle to battle.

The reason why I choosed 40-120 is because my experiments showed that
this is the max. process area where the two strategies - ;redcode-94
and ;redcode-lp - have comparable strengths. Depending on the various
strategies it is sometimes more to the lower processes and sometimes
more to the higher processes. The average is somewhere around 80-90

> John
> --http://corewar.co.uk

fizmo

unread,
Jan 10, 2009, 8:42:48 AM1/10/09
to
On 10 Jan., 05:26, sayemb...@gmail.com wrote:
> I assume this is 8000 cores. Full command specs is appreciated.

Specs are:

Rounds 250
Core Size -s 8000
Cycles To Completion -c 80000
Max Processes -p (8 or 8000 or randomly between 40-120)
Max Warrior Length -l 200
Min Distance -d 200


You cand find also here:

http://www.corewar.info/dhill/

dhill...@netscape.net

unread,
Jan 10, 2009, 2:46:01 PM1/10/09
to
This looks like a lot of fun. I'm guessing that the scoring system
will lead to quite a few tie battles. Time will tell. I don't have
time to make a new warrior, so I sent in a couple of my old ones, just
to help get the ball rolling. I can retire them later if the hill
starts to get cluttered.

- Dave Hillis

fizmo

unread,
Jan 10, 2009, 5:01:34 PM1/10/09
to
I received the first submissions. Just some more and I can activate
the DynaHill.

All submissions are welcome. Beginners, Evolvers, Pros...

saye...@gmail.com

unread,
Jan 11, 2009, 2:59:44 PM1/11/09
to

I resubmit Sunset (as champion of pspace hill) and also Digital Rain
(as champion of LP-hill) on behalf of the authors...without permission
though. I really believe these two warriors are good candidates to
populate the dynahill. Hope you already receive them :)

fizmo

unread,
Jan 12, 2009, 12:07:42 PM1/12/09
to

Received

0 new messages