>"Sam K. Rushing, IV" wrote:
>
>  >
>  > We haven't actually hooked it up to the vehicle yet, because the team
>  > that was responsible for building that didn't finish in time.
>
>Hmm...  I would like to see this vehicle- sounds pretty dangerous.
>
>"energy nuke' store"
>
>Damn, there goes my carpet...
>
>
>And I'm also curious as to the mechanism used for output from the "movex" and
>"movey" registers.
>


Amazing.  If actually completed, I believe this will be the fist
teleportation-competent nuclear device produced by an academic group.
Very impressive.

But seriously, it does sound like an interesting project.  I'll have
to check it out.

Paul

"Sam K. Rushing, IV" wrote:

>
> We haven't actually hooked it up to the vehicle yet, because the team
> that was responsible for building that didn't finish in time.

Hmm...  I would like to see this vehicle- sounds pretty dangerous.

"energy nuke' store"

Damn, there goes my carpet...


And I'm also curious as to the mechanism used for output from the "movex" and
"movey" registers.

I have been working with a group of my classmates this past semester
to design a microprocessor that implements the architecture implied
by the robots' interpreters in RoboWar. The course in now complete
and we prepared a final report that describes the design in detail.

We've tested the processor and it functions properly, as far as we can tell.

We haven't actually hooked it up to the vehicle yet, because the team
that was responsible for building that didn't finish in time.

I have included the code for the compiler I built in the report, and
that code may be adaptable for future versions of RoboWar.

I have posted the report as a PDF file to the project web site:
http://www.prism.gatech.edu/~gt6911c/RWCPU/

I welcome any questions, comments, criticisms, kudos, or anything
else you want to say regarding the processor. Just reply to me via
email.

Thanks,
--
Sam Rushing
Computer Engineering
Georgia Institute of Technology
gt6911c@...

Does anyone know when the next tournament will be held?

Well, as far as I can tell, there have been a few days
   where nobody sent anything in... I wonder why? Maybe
   it's that time of the year where everybody starts
   slaving over their Macs to create the ultimate warrior
   robot, or something like that.

Mmmm.... Finals!.... RoboWar CPU nearing completion....

--
   _____ ___ __ __
  /  ___) . \  \  \  "If I ever meet myself, I'll hit myself so hard
/\____  \   \ __/ \   I won't know what hit me." - Zaphod Beeblebrox
\_______/\__/__/__/

> From: "Paul K. Olson"
<polson@...>
> Subject: Just checking
> Just checking to see if I get this message.  I
haven't seen any
> activity on the list lately, so I got to
wondering.....Paul

Well, as far as I can tell, there have been a few days
where nobody sent anything in... I wonder why? Maybe
it's that time of the year where everybody starts
slaving over their Macs to create the ultimate warrior
robot, or something like that.

=====
"It's as easy as 3.1415926535897932384626433832795028841."
-Anon

__________________________________________________
Do You Yahoo!?
Send online invitations with Yahoo! Invites.
http://invites.yahoo.com

Just checking to see if I get this message.  I haven't seen any
activity on the list lately, so I got to wondering.....

Paul

[The first KOTH results ever to be sent out to the new mailing list
address!]

MKOTH 4 results:

Added: JEST.ALT

No robots updated

Booted: Dislexix

Robot Name          Solo Score  Group Score Total    WC Solo  WC Group Final
CHANGER             1174        919         2093     577      368      3038
JEST.ALT            1121        702         1823     501      415      2739
SYCTHE              1015        844         1859     370      416      2645
Mudshark            1014        828         1842     297      447      2586
Munik               1072        725         1797     371      275      2443
Return of the Tree  1058        604         1662     340      266      2268
Long Dark Tree-Time 950         674         1624
Half-CensoredUpgrag 750         813         1563
Blot                720         836         1556
Stupidity 2.0b2     804         727         1531
Kilgore Sprout      820         692         1512
Snowflake           767         739         1506
Hyoga               930         561         1491
Dave Harris Shaves  703         743         1446
Lives in a tree v3  748         675         1423
Anarchy             783         610         1393
Blood 5.2           603         712         1315
Dave Harris, Age 13 439         866         1305
#####Booted#####
Dislexix            429         848         1277

Happy Hacking,

****  *   *   *   *   *  *
*     *   *   *   *   * *
*     *****   *   *   **
*     *   *   *   *   * *
****  *   *   *****   *  *
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com

MKOTH 3 results:

Added: freezerburn

No robots updated

Booted: freezerburn

Robot Name          Solo Score  Group Score Total    WC Solo  WC Group Final
CHANGER             1250        934         2184     570      365      3119
Mudshark            1057        941         1998     413      462      2873
SYCTHE              1024        883         1907     437      469      2813
Munik               1035        745         1780     392      356      2528
Return of the Tree  1077        717         1794     351      317      2462
Long Dark Tree-Time 1032        677         1709     230      243      2182
Stupidity 2.0b2     901         763         1664
Half-CensoredUpgrag 765         848         1613
Snowflake           818         772         1590
Kilgore Sprout      803         724         1527
Lives in a tree v3  888         638         1526
Blot                678         836         1514
Hyoga               926         558         1484
Anarchy             822         612         1434
Dave Harris Shaves  674         756         1430
Dislexix            486         865         1351
Dave Harris, Age 13 446         877         1323
Blood 5.2           619         673         1292
#####Dropped#####
freezerburn         548         337         885

Happy Hacking,

****  *   *   *   *   *  *
*     *   *   *   *   * *
*     *****   *   *   **
*     *   *   *   *   * *
****  *   *   *****   *  *
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com

> -I'm trying to do this without using much of other
> people's code, or the disassembler--open source comes
> to the Mac! ;)--Hopefully I can bring in a fresh idea
> or two.

Good luck!

My thoughts on strategy:

   * Make sure you have an aggressive attack. This means either look'
tracking or large bullets.
   * Always use some sort of 2x weapon (missiles or explosive bullets)
   * Mines are useless.
   * TacNukes are hard to use.
   * It's easier to program a defensive movement pattern than a good
active defense.

> -I get lots of "Can't move and shoot" messages with
> certain sample bots.

The sample bots were written before the move/shoot limitation was
implemented. This mostly doesn't matter, except for the Dasher.

> -Don't leave the radar interrupt on when firing.

I don't understand this. By "when firing" do you mean while inside the
range interrupt, or while you have bullets on the screen?

If the first, then interrupts are automatically turned off during an
interrupt handler. If the second, then you need to make your radar
interrupt recognize whether a shot is going out or coming in, like this:

radar SYNC radar -
# leaves in-out speed of detected projectile on stack
# if positive, then it's coming closer

> -Learn some trig. I haven't yet dusted off the old
> memories but I can see the need.

Here's something to get you started; this makes your bot move with speed
5 in the direction of your turret.

aim DUP  5 SIN speedX' STORE
         -5 COS speedX' STORE

--
  Kevin Reid: |    Macintosh:
   "I'm me."  | Think different.

>One piece of advice: when you are writing time critical
>sections of your code, do not use this "sugared" notation.
>Actually type in "x' recall y' recall > ifeg". It really
>helps when you have to count instructions; otherwise you
>will often count "x" as one instruction and screw up.

I usually never use the recall instruction unless I am actually
using a variable name in a strange way (in Pyrozot I, for instance.)

>Always run a new robot through the debugger: it gives you
>a much more accurate picture of how RoboWar sees your
>code, and what exactly is happening on the stack.

I'm more experienced at writing robots, so I generally don't.

News flash: Arachnee and Jade are still good robots! Really! I took them,
disassembled them, and updated Jade a lot to get around leap-kills, and
they placed first and second in a large tournament that included the entire
T18 winner's circle (although D9, Music, and Manne did badly).

-- Prfnoff

Edge wrote:
>
> Hello again,
>
> I've a question about counting instructions. The idea
> is to keep rigid control of when instructions are
> processed, in hopes of minimizing code length and
> perfecting the timing.
>
> For example:
>
> main: --no instructions, just a label
> 0 aim' store --would be 3 instructions
> ...
> main jump --just 1 instruction here?

That's 2 instructions.

>
> return --just 1
> RTI --same as return?

They are both one instruction. (But RTI is not the
same as RETURN: it is "denotationally" equivalent
to INTON RETURN, but "operationally" distinct, since
it only costs one instruction. (RETURN is identical
to JUMP, on the other hand. They are equivalent all
the way down to the byte code.))

>
> run collision' setint --3 instructions to define, but
> how many when the interrupt occurs (to skip to the
> defined subroutine), 3? 0? 1?

The beauty of interrupts is that the actual transfer
of control to the interrupt handler is cost-free.
That includes the pushing of the return address
onto the stack. So 0.

>
> sub1 sub2 x y > ifeg --gibberish, but 5 or 6
> instructions?

8 instructions, actually. "sub1" and "sub2" are labels,
so there is no trickery there. But "x" is syntactic sugar
for "x' recall". That is, the RoboTalk compiler/assembler
will actually compile the expression "x" into the code
for the register and the code for the RECALL instruction.

One piece of advice: when you are writing time critical
sections of your code, do not use this "sugared" notation.
Actually type in "x' recall y' recall > ifeg". It really
helps when you have to count instructions; otherwise you
will often count "x" as one instruction and screw up.

Always run a new robot through the debugger: it gives you
a much more accurate picture of how RoboWar sees your
code, and what exactly is happening on the stack.

-Tom

> For example:
>
> main: --no instructions, just a label
> 0 aim' store --would be 3 instructions
> ...
> main jump --just 1 instruction here?

Two.  One for pushing the label "main" to the stack and a second for
"jump".

> return --just 1
> RTI --same as return?

rti is one instruction, and will turn interrupts on while you're doing it.
This is why sometimes it is known to be used synonymously with a "jump"
instruction to save the extra instruction for an "inton".

> run collision' setint --3 instructions to define, but
> how many when the interrupt occurs (to skip to the
> defined subroutine), 3? 0? 1?

Zero to head to the interrupt handler.

> sub1 sub2 x y > ifeg --gibberish, but 5 or 6
> instructions?

One for pushing sub1.
One for pushing sub2.
Two for dereferencing x.  ("x' recall")
Two for dereferencing y.
One for pushing the >.
One to execute the ifeg.

Total of eight.
--
=-=-=-=-=-=-=-=
  The Stilt Man                stiltman@...
    http://www.teleport.com/~stiltman/stiltman.html
    < We are Microsoft Borg '98.  Lower your expectations and    >
    < surrender your money.  Antitrust law is irrelevant.        >
    < Competition is irrelevant.  We will add your financial and >
    < technological distinctiveness to our own.  Your software   >
    < will adapt to service ours.  Resistance is futile.         >

Hello again,

I've a question about counting instructions. The idea
is to keep rigid control of when instructions are
processed, in hopes of minimizing code length and
perfecting the timing.

For example:

main: --no instructions, just a label
0 aim' store --would be 3 instructions
...
main jump --just 1 instruction here?

return --just 1
RTI --same as return?

run collision' setint --3 instructions to define, but
how many when the interrupt occurs (to skip to the
defined subroutine), 3? 0? 1?

sub1 sub2 x y > ifeg --gibberish, but 5 or 6
instructions?

I'm also considering a CGI interface (perl) to aid in
construction if there's interest, and the more
specific information (and ideas) I can get the better.

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

I am going away to Windsor for sometime (approximatly
to Saturday the 6th of April) so I won't be able to
post or update my website. When I get back, I'll have
to work until the 12th on an ISU. After that, I'll be
updating the page again.

=====
"It's as easy as 3.1415926535897932384626433832795028841."
-Anon

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

Thanks to Peter Ström for showing the way to the real
robowar mailing list.

Thanks also to everyone who's offered help/advice/code
samples.

Thoughts so far:

-I'm trying to do this without using much of other
people's code, or the disassembler--open source comes
to the Mac! ;)--Hopefully I can bring in a fresh idea
or two.

-Everything is vulnerable in some way.

-I get lots of "Can't move and shoot" messages with
certain sample bots.

-Don't leave the radar interrupt on when firing.

-Learn some trig. I haven't yet dusted off the old
memories but I can see the need.

-Don't drink and battle! The beer causes short
circuits in your keyboard!! ;)

Just another bot hacker...
E



__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

I am starting up a RoboWar website at:
http://www.macrobo.homestead.com/

but it is still under construction and all it
currently has is a brief description of the game and a
link to Lucas's Hindquarters. In the future, I will
have downloadable robots for newbies, etc. but I can't
see into the future and I don't know if there will be
any delays in the postings.

Thanks for reading this.

=====
"It's as easy as 3.1415926535897932384626433832795028841."
-Anon

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

Edge wrote:
> I'm missing something simple with the stack.. Well,
> and I need to study up on the trig too..
>
> doppler -12 arctan aim +
>
> ....would leave nothing on the stack?

That expression leaves one item on the stack. The trace (by hand, not by
RoboWar) is (assuming aim=90 and doppler=4):

Instruction: doppler'
Stack: DOPPLER
Instruction: recall
Stack: 4
Instruction: -12
Stack: 4 -12
Instruction: arctan
Stack: -18
Instruction: aim'
Stack: -18 AIM
Instruction: recall
Stack: -18 90
Instruction: +
Stack: 72

..which is a good angle to fire bullets at. I think what's confusing you is
that the code does not change aim; you'll need to store the result into aim
afterward.

BTW, the original message was sent to robowar@.... Remember to use
robowar@onelist.com in the future.

- Devon


____________________________________
Email services provided by ApexMail
http://www.apexmail.com

I figured out how to get the tournament 18 robots onto my computer in
a form I can use.  This was particularly difficult since I'm using
the demo version of the Fusion emulator on a PC for running RoboWar.

1) Download the tournament robots using Netscape running under
Windows.
2) Reboot into Linux and mount both the Windows hard disk containing
the binhex file and the Macintosh disk image used by the emulator.
3) Copy the binhex file from the hard disk to the disk image.
4) Reboot again, this time into the emulator.
5) Un-binhex and unstuff the robots.

Mark

> From: "Kevin McNish" <SirKevinMc@...>
> I've been working on a stun streamer and I need a bit of help. Which
> is better for a killshot? HellMissiles or HellBullets? My KillShot
> code currently looks like this:

Well, it depends on a lot of things.

If you're building a titan, you can efficiently use hellmissiles and
hellexplosives at varying range.  At closer ranges (roughly 120 on in or
so, I did the math years ago for Nightshade and have since forgotten it),
the smaller cost of the hellbore is less expensive than the added cost of
stunning the opponent longer so that the killshot behind the stunner can
impact before they come out of stun.  (Not to mention that at extremely
close ranges it's downright dangerous.)  At longer ranges, this stunner gets
more expensive for hellmissiles and the hellbore strength of 12 becomes
more cost effective.  In either case, you should use a probe on the enemy
since they're free for titans.

For a mortal, I don't know if I'd suggest using probes, but there's also the
factor that stunner bots are far and away the most lethal users of probes
since they don't have to show a negative energy to their victim to kill them,
EVER.  They can hit them with a stunner, keep in positive energy, and no
matter whether you kill them with a probe snipe or a stun kill, either way
they never need see you in negative energy when the lethal blow is on its
way.  It's a factor to keep in mind, but I don't know if it's worth a
hardware point for it for a stunner bot.  In most cases stunner bots can
get away with not being able to probe snipe, because their stunner offense
can just as effectively take advantage of a helpless opponent as an instant
probe snipe could in a lot of cases.

However, for a mortal, missiles are clearly better than bullets for most
stunner bots.  The efficiency of the kill shot is paramount here.  If you
want to use hellbores, neat, that makes it a little better, but there's still
some very real possibilities of overload there.  With the missiles, you'll
have to stay closer, but a solid stunner bot should stay within a range of
about 180 to its enemy anyway, so that's not a huge issue.  If you're using
probes, don't bother with hellbores, and vice versa.  You can possibly afford
the hardware for one or the other but most certainly not both.
--
=-=-=-=-=-=-=-=
  The Stilt Man                stiltman@...
    http://www.teleport.com/~stiltman/stiltman.html
    < We are Microsoft Borg '98.  Lower your expectations and    >
    < surrender your money.  Antitrust law is irrelevant.        >
    < Competition is irrelevant.  We will add your financial and >
    < technological distinctiveness to our own.  Your software   >
    < will adapt to service ours.  Resistance is futile.         >

> I've been working on a stun streamer and I need a bit of help. Which
> is better for a killshot? HellMissiles or HellBullets? My KillShot
> code currently looks like this:
>
> KillShot:
> 60 stunner' store
> 5 hellbore' store
> 80 missile' store


The amount of energy for the stunner should be:

(time for missile to hit)-(time for stunner to hit)*4

e.g

range 5 / range 14 / - 4 * stunner' sto


This can be roughly simplified to:

range 2 / stunner' sto


Also if a hellmissile is used then the energy expenditure for the missile
only needs to be 75 so your killshot routine should be:


KILLSHOT:
range 2 / stunner' sto
5 hell' sto
75 missile' sto


I'd suggest that you forget about probes unless you also have a high
processor and you're confident that you'll use them enough to make them
worthwhile. With a kill shot like this it doesn't really matter if you use
only the exact amount of energy as:
a) the above killshot will kill any target if it hits, and
b) your robot will go into negative energy anyway so if your kill shot
misses (which it shouldn't) then the enemy should be able to kill you anyway

I currently working on a stun-streamer myself for T19 :o)

Hope this helps,

Ed

Whoa! I didn't do that! I never even read anything about it doing that! I'll
get on it immediately and see if I can quell the problem.

Austin


-----Original Message-----
From: Devon Schudy <dms@...>
To: robowar@onelist.com <robowar@onelist.com>
Date: Thursday, March 23, 2000 12:44 PM
Subject: [robowar] Re: New poll for robowar


>From: "Devon Schudy"<dms@...>
>
>Uh oh. Polls.
>
>I'd think it obvious, but this is *not* a feature we should use. It's
>annoying and generates data that looks meaningful but isn't. If we actually
>had any issues we cared about, democracy would not be useful even if we had
a
>good sample. Please, don't create polls. (That is restricted to the
>list-owner, isn't it?)
>
>- Devon
>
>
>
>____________________________________
>Get your free full featured email @
>http://www.apexmail.com
>
>
>
>
>------------------------------------------------------------------------
>Free Cool Black Virgin T-shirt with ANY order at virginmega.com!
>http://click.egroups.com/1/2625/1/_/_/_/953844189/
>------------------------------------------------------------------------
>
>To unsubscribe from this group, send an email to:
>robowar-unsubscribe@onelist.com
>
>
>
>

Uh oh. Polls.

I'd think it obvious, but this is *not* a feature we should use. It's
annoying and generates data that looks meaningful but isn't. If we actually
had any issues we cared about, democracy would not be useful even if we had a
good sample. Please, don't create polls. (That is restricted to the
list-owner, isn't it?)

- Devon



____________________________________
Get your free full featured email @
http://www.apexmail.com

Enter your vote today!  Check out the new poll for the robowar
group:


What is your opinion of the new mailing list?

   o Superb!
   o OK.
   o Bad.
   o Haven't seen it yet.
   o No opinion.


To vote, please visit the following web page:

http://www.onelist.com/polls/robowar

Note: Please do not reply to this message. Poll votes are
not collected via email. To vote, you must go to the ONElist
web site listed above.

Thanks!

SirKevinMc@... wrote:
> I've been working on a stun streamer and I need a bit of help.
> Which is better for a killshot? HellMissiles or HellBullets?

Missiles are much better than bullets for killshots, since they cost half the
energy. Hellbores are questionable here, because of the hardware cost - after
buying stunners and missiles, you get painfully short on hardware points.

> Or should I use probes to fire the exact amount of damage to
> finish th enemy?

If you've got probes, by all means use them. Unless you can make good use of
them elsewhere, though, it's probably not worth buying them for this.

Titans, of course, should use hellbores, and probes wherever they can, and
should use xbullets for long-range killshots.

We still need a newbie guide... Veterans should post bits of
advice/explanation as the spirit moves them; we can extract them from the
list archives later to build a newbie guide.

- Devon



____________________________________
Get your free full featured email @
http://www.apexmail.com

I've been working on a stun streamer and I need a bit of help. Which
is better for a killshot? HellMissiles or HellBullets? My KillShot
code currently looks like this:

KillShot:
60 stunner' store
5 hellbore' store
80 missile' store

Should it be this?

KillShot:
32 stunner' store
12 hellbore' store
100 fire' store

Or should I use probes to fire the exact amount of damage to finish
th enemy?

Thanks

Kevin

This isn't a particularly useful message, I'm just determining how the mail
to the new RoboWar list is going to come to me so I can file it as mailing
list fare the same way as the old list.  Pay it no mind.  :)
--
=-=-=-=-=-=-=-=
  The Stilt Man                stiltman@...
    http://www.teleport.com/~stiltman/stiltman.html
    < We are Microsoft Borg '98.  Lower your expectations and    >
    < surrender your money.  Antitrust law is irrelevant.        >
    < Competition is irrelevant.  We will add your financial and >
    < technological distinctiveness to our own.  Your software   >
    < will adapt to service ours.  Resistance is futile.         >

So far only you and me on! Stiltman won't send me a list of current members!
I can't add them up. No one else will voice opinion.

Some interesting thoughts on robots. If you want help on figuring out
problems, I love designing solo  modes, so send em to me and I'll see what I
can come up with!

Thanks,
Austin

-----Original Message-----
From: prfnoff@... <prfnoff@...>
To: robowar@onelist.com <robowar@onelist.com>
Date: Monday, March 20, 2000 12:34 PM
Subject: [robowar] I've arrived!


>From: prfnoff@...
>
>So, here I am on the new RoboWar list!
>
>Here's a RoboWar note: the Mortal entries were surprisingly devoid of
>the mock-dashers that appeared in T15-T17, and Munik was designed to
>take maximal advantage of them. Perhaps that's why they weren't
>entered, but Dave Harris Shaves could have taken a good chunk out of
>the hesitators' scores, even against Mudshark. Of course, then you
>have the whole stun-streamer problem, which is what I'll get to in
>a moment.
>
>Now, Munik-set has only two real enemies out in the world: high-energy
>(usually 100), high-processor (usually 30) robots that have been
>expected to outmaneuver it. The others, hesitators with lots of
>damage (usually 150) and lots of processor (usually 50), have been an
>annoyance, and it's not as if they're unbeatable (as Tom Cornell
>says that they can be beaten, and that's the whole trick). The real
>problem is Munik: how to defeat its streamer/gunner mix without using
>hesitation or high energy. How can this major problem be fixed?
>
>-- Prfnoff
>
>
>------------------------------------------------------------------------
>Now's your chance to get organized! Get the Web's lowest prices on
>Palm PDAs and accessories at Accompany, the leader in group buying
>Visit us today at start saving!
>http://click.egroups.com/1/2513/1/_/_/_/953584420/
>------------------------------------------------------------------------
>
>To unsubscribe from this group, send an email to:
>robowar-unsubscribe@onelist.com
>
>
>
>

So, here I am on the new RoboWar list!

Here's a RoboWar note: the Mortal entries were surprisingly devoid of
the mock-dashers that appeared in T15-T17, and Munik was designed to
take maximal advantage of them. Perhaps that's why they weren't
entered, but Dave Harris Shaves could have taken a good chunk out of
the hesitators' scores, even against Mudshark. Of course, then you
have the whole stun-streamer problem, which is what I'll get to in
a moment.

Now, Munik-set has only two real enemies out in the world: high-energy
(usually 100), high-processor (usually 30) robots that have been
expected to outmaneuver it. The others, hesitators with lots of
damage (usually 150) and lots of processor (usually 50), have been an
annoyance, and it's not as if they're unbeatable (as Tom Cornell
says that they can be beaten, and that's the whole trick). The real
problem is Munik: how to defeat its streamer/gunner mix without using
hesitation or high energy. How can this major problem be fixed?

-- Prfnoff