corewar newbie

[Guest]

Archived from groups: rec.games.corewar (More info?)

Hi, I'm brand new to CoreWar. I've been reading up on redcode and sample
programs in the tutorials over the past week, and I finally tried
downloading CoreWin and making one of my own warriors. It's not faring
very well yet, but I didn't expect much from my first try.

The thing I wanted to ask is if anyone has tried working with the idea
of purposely jumping into other people's code... sort of a vampire in
reverse. I guess my theory is that if their warrior is so great, I
should be able to jump into an opportune execution point and run as it
myself.

My first attempt used this idea and basically acted like a replicating
bomber which would split into remote addresses rather than drop dats on
them. The results were less than favorable. I got a lot of draws against
Mice, several losses, and maybe one lucky win. From the GUI, it looked
like my processes were frequently becoming mice themselves, but it
wasn't enough to win.

Is this just a stupid idea to begin with?

/joe
--
THE EMPEROR OF THE SOLAR SYSTEM HAS SPOKEN

 

[Guest]

Archived from groups: rec.games.corewar (More info?)

phat_joe wrote:
> Hi, I'm brand new to CoreWar. I've been reading up on redcode and
sample
> programs in the tutorials over the past week, and I finally tried
> downloading CoreWin and making one of my own warriors. It's not
faring
> very well yet, but I didn't expect much from my first try.
>
> The thing I wanted to ask is if anyone has tried working with the
idea
> of purposely jumping into other people's code... sort of a vampire in
> reverse. I guess my theory is that if their warrior is so great, I
> should be able to jump into an opportune execution point and run as
it
> myself.
>
> My first attempt used this idea and basically acted like a
replicating
> bomber which would split into remote addresses rather than drop dats
on
> them. The results were less than favorable. I got a lot of draws
against
> Mice, several losses, and maybe one lucky win. From the GUI, it
looked
> like my processes were frequently becoming mice themselves, but it
> wasn't enough to win.
>
> Is this just a stupid idea to begin with?
>
> /joe
> --
> THE EMPEROR OF THE SOLAR SYSTEM HAS SPOKEN


Hi.
I've had the same idea when i started playing corewar.
Somtimes you might get lucky and win a couple of ties, but generaly
speaking, it's not good. If the other warrior uses spl carpets and
traps, you'll get stuck in them and easily lose. And, as already
mentioned, modern replicators are made in such a way that only one
process is not enough to take control of a replicator copy. Of course,
you are very unlikely to win. Sometimes you can, against very sensitive
warriors if they are made in such a way to get destroyed if their
process number is disrupted (but i don't know if there are any - you
certainly won't find them on the hill).
Try some stones and simple papers instead, they are easy to write.
Also, get yourself a decent benchmark.
You can look for warriors at Koenigstuhl. At first, you might want to
try those from the middle and the lower half of that infinite hill. (if
you try benchmarking against the best in the very beginning, you won't
notice the slight increase in your results). After that, when you make
some warriors to beat your first benchmark, pick harder opponents, etc.
There are some existing benchmark (old and new), and you can find them
at http://www.corewar.info, where you'll find everything else you need.
Whenever you have questions, don't hesitate to post a message here. You
might also go to irc.koth.org, #corewars, if you want to discuss some
of your warriors and get new ideas, or ask for advise.

 

[Guest]

Archived from groups: rec.games.corewar (More info?)

On Tue, 8 Mar 2005 17:53:30 +0000 (UTC), phat_joe wrote:

> The thing I wanted to ask is if anyone has tried working with the idea
> of purposely jumping into other people's code... sort of a vampire in
> reverse. I guess my theory is that if their warrior is so great, I
> should be able to jump into an opportune execution point and run as it
> myself.

I guess I started from the same idea. In a nutshell: it's not worth it.
As you stated, you will get mostly ties which is not that good (you
strive for wins, don't you?). Bear in mind: you should destroy your
opponent, not execute its code hoping for an easy tie.

> Is this just a stupid idea to begin with?

It is not that stupid.

Few points, you may want to consider: Mice is not the best oponnent in
the world, I wouldn't bother benchmarking my warrior against so ancient
code. Get yourself modern papers (gosh, I don't know the names, Roy and
Chris has both made so many of them and check out how your warrior
performs when fighting real ones.

I also think (just a vague feeling) that if you have less processes when
required, you die when you try to execute silk papers code.

Good luck, anyway. Stick to your idea but try to improve on it. My first
warrior to enter the hill was strongly based on my very first attempt to
create viable code.

 

[Guest]

Archived from groups: rec.games.corewar (More info?)

phat_joe <scag@moralminority.org> wrote in message news:<d0koqq$5f0$1@news-int.gatech.edu>...
> Hi, I'm brand new to CoreWar. I've been reading up on redcode and sample
> programs in the tutorials over the past week, and I finally tried
> downloading CoreWin and making one of my own warriors. It's not faring
> very well yet, but I didn't expect much from my first try.

Welcome. First warriors always score poorly. keep trying!

> Is this just a stupid idea to begin with?
>
> /joe

Of course not. Many newbies start in that way.
As other redcoders have said, executing your opponent's code will give
you ties instead of wins, (as best result). Also some warriors need a
certain number of parallel processes to run correctly. This has been
"exploited" by some warriors to die when they are executed by more
processes, (in replicators).

I think the best way to learn is take a look at other people's code
and try to follow the same strategy on your own. Also beginner guides
ofers you nice pieces of code to start with.

Neogryzor

 

[Guest]

Archived from groups: rec.games.corewar (More info?)

> I think the best way to learn is take a look at other people's code
> and try to follow the same strategy on your own. Also beginner guides
> ofers you nice pieces of code to start with.


hi and welcome to corewar!

a few days ago i stated teaching some of my friends how to start
playing corewar. some of the best ways to start it is analysing other
peoples code. but in the beginning its sometimes hard to understand
current state-of-art warriors.
therefore i started a kind of tournament with my friends: someone is
writing a warrior. it doesnt matter wheater it is a good one or. the
others a wanted to analyse the warrior and write a "counter warrior".
this counter warrior should beat the other by atleast 60% wins.

you can take part in our little tournament if you want to. here you
find all you need:


http://home.t-online.de/home/familie.kersten/johannes/beginner/


el kauka