nano hill

Here you go, including some currently on the hill.
(Other than the fact that it's always easier to get on a hill if you
know what's on it, I don't think the nano hill has gotten any stronger
since early in its history.)

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Slug Stampede
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.i # -30, > 21
mov.i * -12, { 0
mov.i > -29, { -2
djn.x $ -2, $ -3
spl.i < -1, # 35
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Cosy Eruption
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.b # -36, > 15
mov.i * -14, { 0
mov.i > -32, { -2
djn.f $ -2, $ -3
mov.i @ -4, { -38
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Strychnine Banshee
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.i # 36, > -17
mov.i } -6, { -1
mov.i } -31, { -1
mov.i > -27, { -1
mov.i # 2, } 1
end 0


;redcode-nano
;assert CORESIZE == 80
;name rdrc: Revere Gimmickry
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.i # 39, > 15
mov.i } -20, { 0
mov.i > -18, { -2
djn.f $ -2, $ -3
mov.i $ -4, < 24
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Silhouette Ulcer
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
mov.i > -21, $ -37
spl.b # 34, > 14
mov.i } -5, { 0
mov.i > -29, { -2
djn.i @ -2, $ -3
end 0

;redcode-nano
;assert 1
;name h1_38.red
;author Dave Hillis
;strategy - Created with redrace, 1999.
;strategy - for Ilmari's Mini-Tournament #2.
mov.i { 34, > 29
mov.i { 9, } -1
mov.i > 21, } -1
djn.i $ -2, } -3
end 0

;redcode-nano
;assert 1
;name h1_36
;author Dave Hillis
;strategy - Created with redrace.
;strategy - for Ilmari's Mini-Tournament #2.
spl.i # -30, > 21
mov.i * -12, { 0
mov.i > -29, { -2
djn.x $ -2, $ -3
end 0

;redcode-nano
;assert 1
;name h1_3.red
;author Dave Hillis
;strategy - Created with redrace, 1999.
;strategy - for Ilmari's Mini-Tournament #2.
spl.i # -38, > 18
mov.i } -9, { 0
mov.i > -31, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Surpass Stricture
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.i # 36, > -32
mov.i } -6, { -1
mov.i > -17, { -1
mov.i > 18, { -1
mov.i # 10, $ 1
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Delicate Crowbait
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
mov.i > -37, $ 32
spl.i # -28, > 24
mov.i } -6, { 0
mov.i < -24, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Hexadecimal Bunny
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.i # -38, > 18
mov.i } -9, { 0
mov.i > -31, { -2
djn.i $ -2, $ -3
mov.i # 6, < -15
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Chiefdom Monogamy
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
mov.i $ 39, $ 31
mov.i { 9, } -1
mov.i } 25, } -1
djn.i $ -2, } -3
slt.a < 4, > 9
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Butterball Acetate
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
;strategy - new batch.
spl.b # -15, > 16
mov.i } -5, < 36
mov.i > 8, { -2
mov.i < 34, < -34
mov.i > 36, { -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Mesozoic Bronco
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
;strategy - new batch.
spl.ab # 1, > -22
spl.ab # -33, > -36
mov.i } -17, { 0
mov.i # -12, } 1
jmp.a { 18, < -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Promote Credulity
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.b # -6, > -35
spl.b # -33, < -39
mov.i } -17, { 0
mov.i # -12, } 1
djn.i $ 18, < -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Sanctimonious Burden
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
mov.i < 32, $ 13
spl.i # -27, > 24
mov.i * -6, { 0
mov.i < -23, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Effluvia Verbosity
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 147.028564
mov.i > -22, $ 13
spl.f # -26, > 35
mov.i } -7, { 0
mov.i > -20, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Borneo Birdie
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 145.899994
mov.i > 29, $ 13
spl.i # -35, > -28
mov.i } -11, { 0
mov.i > -23, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Repent Linemen
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 146.828568
mov.i > -4, $ 30
spl.ab # -29, < 24
mov.i } -6, { 0
mov.i < -25, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Sportsmen Momentary
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 144.800003
mov.i > 35, $ 30
spl.i # -26, > 22
mov.i } -5, { 0
mov.i > -23, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Blanch Autoclave
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 146.592850
spl.i # -38, > 18
mov.i * -14, { 0
mov.i > -32, { -2
djn.f $ -2, $ -3
mov.i $ -4, < 19
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Alcoholism Malt
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 146.964279
mov.i < -35, $ 13
spl.ab # 38, > -23
mov.i * -8, { 0
mov.i > -35, { -2
djn.x $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Aborning Chute
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 144.100006

mov.i > 29, $ 13
spl.i # -26, > 22
mov.i } -11, { 0
mov.i > -25, { -2
djn.x $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Laundry OSHA
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 144.678574
mov.i > -4, $ 30
spl.i # -26, > 22
mov.i } -6, { 0
mov.i > -21, { -2
djn.i $ -2, $ -3
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Snapback Sprite
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
; scored 144.857147
; born on 88
spl.i # -19, > 34
mov.i > 23, { -3
mov.i > -17, { -2
mov.i < -38, { -2
dat.b > 24, { -4
end 0

;redcode-nano
;assert CORESIZE == 80
;name rdrc: Breakaway Carte
;author Dave Hillis
;strategy - Created using RedRace.c.
;strategy - An evolving population playing KOTH.
spl.f # -34, > -18
mov.i } -6, { -1
mov.i } -23, { -1
mov.i > 25, { -1
mov.i # 3, $ 1
end 0
 >> Stay informed about: nano hill 

I don't have anything on the hill right now. I don't know if these
will help you because they have been off the hill for a long time.
But, I was planning to publish them anyway, so here they are.

;redcode-nano
;assert CORESIZE == 80
;name Sabertooth
;author Ken Hubbard
;strategy Two concurrent clears, created by RedRace: SPL# MOV MOV JMP-2

start spl.f # 9, } start
attack1 mov.i > target, } start
attack2 mov.i > -22, > target
jmp.i attack1, } attack2
target spl.x } 0, } 24

end start


Sabertooth lived to be 206 on the Nano Hill, and was pushed off around
6 months ago, I think. I'm not sure why it lasted so long, because
it's a very simple program. It is 100% created by Dave Hillis's
Redrace evolver. I didn't change anything, except to add the labels.
(I think it's easier to read that way, but I imagine I'm in the
minority on that.)


Here is my best nano warrior:

;redcode-nano
;assert CORESIZE == 80
;name Military Grade Nano
;author Ken Hubbard
;strategy RedRace assisted in making this d-clear variant w/
incendiary bomb.

spl.i # 9, } 0
mov.i < 52, } -1 ;evolved as >-32, }-1
mov.i $ 2, } -2
djn.i $ -1, } -3
mov.i * 5, < 0
end 0


I had a version with labels somewhere, but can't find it right now.
Military Grade Nano was also created by Redrace, but I manually changed
the constant in line 2 after some testing against the then-current
hill.

If you look at the code of the great "On-Speed" by brx, you can see a
lot of similarity. If I remember correctly, I had seeded Redrace with
several of the top warriors at the time. This one is structurally
identical to On-Speed, but the two significant changes that Redrace
found are:

(1) shortening the djn loop in the 4th line to a -1 instead of -2. I'm
not sure the significance of this, except that it means most of the
functioning processes are in a 2 cell range, rather than a 3 cell
range. Probably makes it more resistant to a SPL bomb hitting line 2.

(2) Where OnSpeed used a DAT bomb, MGN uses a MOV bomb. The way the
constants work out causes the process that executes the bomb to usually
pick up another copy of the MOV instruction and write it 1 cell behind
the one executing. So, the opponent over-writes himself backwards.
Obviously that increases your chances of taking out the opponent's
all-important SPL instruction, and also seems to cause a wounded
opponent to die a few cycles faster than he otherwise might.

Anyway, this is my favorite nano warrior. He lived to be 299 and was
pushed off last November, I think. Big thanks to Dave (for Redrace)
and brx (for publishing On-Speed), because otherwise I never could have
made this one.


I had two other programs on the hill for a while. I think they both
lived over a 100 cycles, but I seem to have lost the code for both. I
recall they were both hand-coded efforts to fix the weakness of
Military Grade Nano (which is that when his djn stream hits a 1,1 imp
trail and drops through, he eats his own incendiary bomb). If I
remember correctly, Moving Mars placed the bomb at the top of the
warrior instead of the bottom, while Heavy Ash use a hard jump instead
of a djn in line 4. Both warriors were inferior to the original.

By the way, can anyone tell me how to submit Sabertooth and Military
Grade Nano to the infinite hill?
 >> Stay informed about: nano hill 

I agree with Dave that the strength of nano is almost about the same.
The KOTH can easily be pulled down by sending several 'anti-KOTH'. How
much good variations can we get from 5 lines? Not much
 >> Stay informed about: nano hill 

> I agree with Dave that the strength of nano is almost about the same.
> The KOTH can easily be pulled down by sending several 'anti-KOTH'.

Not at all. That depends on your strategy. Evolvers probably find it as
difficult as it was some time ago, but we handcoders notice some
balance variations.
As instance, scanners once met a great success, and took many points
from those evolved clears/mad bombers. Actually they have lost their
advantage and it is very hard for them to enter in the top half of the
hill.

> How much good variations can we get from 5 lines? Not much

More than it seems
 >> Stay informed about: nano hill 

Neogryzor wrote:
> > I agree with Dave that the strength of nano is almost about the same.
> > The KOTH can easily be pulled down by sending several 'anti-KOTH'.
>
> Not at all. That depends on your strategy. Evolvers probably find it as
> difficult as it was some time ago, but we handcoders notice some
> balance variations.
> As instance, scanners once met a great success, and took many points
> from those evolved clears/mad bombers. Actually they have lost their
> advantage and it is very hard for them to enter in the top half of the
> hill.
>
> > How much good variations can we get from 5 lines? Not much
>
> More than it seems

What I said was intended as conjecture. There's no consensus on
this list as to what it means when you say "the hill is getting
stronger," or on how to measure it, or even whether it's desirable.
When I was evolving for the nano hill, and using a koth style
fitness function, for a hill size of 50, non-transitive effects
dominated. That was bad for what I was trying to do, so I used a hill
size of 200 which was slightly better.
As idle speculation: if we had a very large but finite nano hill
(say size = 2000) and the evolvers abandoned all restraint and good
manners, then I think we would see what most of us would consider a
stronger hill.
I'd like to hear other people's opinions.
For the record, I think the nano hill is fine the way it is. The
art of figuring out what's on a hill and designing to it may be lost on
us evolvers (me anyway) but it gives corewars some of its charm.
Dave Hillis
 >> Stay informed about: nano hill 

>so they should appear on the infinite hills soon.

They did -- Military Grade Nano showed up in 9th place. Amazing....


Here are two more recent additions to the nano hill:

;redcode-nano
;assert CORESIZE == 80
;name Bear
;author Ken Hubbard
;strategy Redrace-evolved clear. I adjusted the pointers in the
;strategy first line to reduce gaps in the coverage

start spl.b # 32, > 56 ;evolved as spl.b #28, >61
mov.i } -29, { -4
mov.i > start, { -1
mov.i > 33, { start
dat.ab > 35, { start

end 0


Bear is not doing too well, and is dipping in and out of the bottom 10.



;redcode-nano
;assert CORESIZE == 80
;name Different Day
;author Ken Hubbard
;strategy Another clear/bomber thing via RedRace

start spl.f # -14, } -31
mov.i { start, { -4
mov.i < start, { start
mov.i > 24, < start
div.i $ 12, } 0

end 0

Tested a few modifications, but I couldn't find anything that scored
better against the test set (which was all published warriors currently
on the SAL hill).

Different Day has a structure somewhat similar to Bear, but currently
scores much better. It has been holding in or near the top 10
(although it's only age 6, so who knows what will happen).

I have to confess, however, that I'm having a hard time figuring out
what the DIV instruction is doing. I stepped through several times in
Corewin, and it initially appears that it does nothing useful. It
doesn't modify the core in any way that I can see.

Here is the puzzling thing:

--WITH the div line, the warrior self terminates after 268 cycles.

--WITHOUT it, he doesn't self-terminate.

You would think NOT killing yourself would help your score, but it
doesn't. In fact, it scores about 20 points higher WITH the Div line.


I eventually figured out that after the clear wraps around it uses that
line as a bomb one time, and then uses the a-field value of that
location as a pointer for the second pass. The increased score is
apparently due to the fact that there are two simultaneous bombing runs
that are spread out (and therefore more effective). Without the div
12, }0, there are still two bombing runs, but they are so close
together that there might as well be only one.

So, problem solved? No... if that's all there was to it, then a Dat
12, }0 would function just as well as the Div, but it doesn't. So, I
can only assume that the Div itself is changing something else
somewhere that is relevant.

I got tired of studying it. I imagine somebody else can probably make
sense of it immediately, but I apparently can't. I sort of feel like
I'm cheating when I enter a warrior that I don't understand. That's
one of the reasons that I don't play on the regular hill -- I'm not
good enough to write effective warriors, and I don't want to just hack
and copy, and use code that is gibberish to me (like those crazy Q ^ 4
thingies).

Anyway, I thought I would publish a couple of warriors since I have
been making use of warriors that other folks have published.

--Ken
 >> Stay informed about: nano hill 

Neo: > How much good variations can we get from 5 lines? Not much
More than it seems

Just look at the warriors posted here and also the top warriors on the
infinite hill. Virtually all are using spl/mov with majority using
spl/mov/djn for looping. I focused on 'strong' warriors. I think some
of the warriors are already clones for some of other warriors.

Lukasz: "Let me know if you want to see more of my calculations
results, "

Yes, show me you results
 >> Stay informed about: nano hill