CoreWar "Koenigstuhl" Page

This hill is named after the location of the place where I worked (Max-Planck-Institute for Astronomy) until 1997 ('Koenigstuhl' translates to 'King's Chair'). Similar to Planar's Mt.Olympus (last update: 1998-08-31) this is an infinite hill. The difference is that I separated the hill into several categories and I let every program fight every other program:

Standard Hills (CORESIZE==8000)
  • 88 ('88 instruction set) - KotH: Lithobolia C yvfp by Steve Gunnell
    TOP-50: Top 50 of the 88-Koenigstuhl - KotH: Into the Unknown by John Metcalf
  • 94nop ('94 instruction set, no Pspace, no '88 programs) - KotH: Hullab3loo by Roy van Rijn
    TOP-50: Top 50 of the 94nop-Koenigstuhl - KotH: Hullab3loo by Roy van Rijn
  • Pspace (Pspace-only) - KotH: Sunset by David Moore
    TOP-50: Top 50 of the Pspace-Koenigstuhl - KotH: Sunset by David Moore
    • Classical Hill ('88 instruction set)
  • ICWS/64 (CORESIZE==8192 && MAXPROCESSES==64)
    • KotH: Twimp by Monika Keindl
  • ICWS (CORESIZE==8192 && MAXPROCESSES==8000)
    • KotH: Eye of the Storm T by John Metcalf
    Other Hills (94draft instruction set)
  • Limited Process (MAXPROCESSES==8 && MAXLEN==200)
    • KotH: R.O.U.S. by S.Fernandes
  • X (CORESIZE==55440 && MAXLEN==200 && MAXCYC==500000)
    • KotH: Fatamorgana X by Zul Nadzri
  • Tiny (CORESIZE==800 && MAXLEN==20 && MAXCYC==8000)
    • KotH: dfd94f0-408f53a4-b1f60e08.rc by bvowk
  • Nano (CORESIZE==80 && MAXLEN==5 && MAXCYC==800)
    • KotH: [RS] Nextratulated Sturvinator 2 by inversed

  • Odd Hill (non-competitive, special tournament rule sets)

  • '88 hill(s): "-8 -r 1000" (648 entries: 88.tar.gz)

    Recursive 88-Koenigstuhl -- Battle Royale 88-Koenigstuhl

    Re-run with "-P" (all permutations) for the top-200 and "-r 1000" for all weaker warriors.
    The score on the "Royale" hill is sampled at the iteration when the warrior was in the last position.
    I also switched to pmars-0.9.4 and a 32-bit (little endian) score table (was 16-bit and pmars-0.9.2).

    Recent entries: (rank on recursive 88-Hill, program, author, date-of-entry)
    8: Torment , John Metcalf, (Jun/08)
    15: Prometheus (decoy) , John Metcalf, (Sep/06)
    163: ]enigma[ , Michal Janeczek, (Jun/10 -- moved from ICWS)
    171: Flash Paper3.8 , Matt Hastings, (Jun/08)

  • 88-TOP-50 Hill, "-8 -P", '88 rules (15602 rounds)

    Any program that reaches rank-50 or higher on the '88-Koenigstuhl will also enter the TOP-50 Koenigstuhl. Programs that fall below rank-50 on both ('88 and TOP-50) hills will be removed from the TOP-50 Koenigstuhl.

    TOP-50 of the '88-Koenigstuhl (88top50.tar.gz)

    Recent entries: (rank on TOP-50 hill, program, author, date-of-entry)
    3: Prometheus (decoy) , John Metcalf, (Sep/06)
    21: Torment , John Metcalf, (Jun/08)

    Recent leaves: (program, author, date)
    Agony 4.1, Stefan Strack, (Sep/09)

  • '94nop-hill(s), "-r 1000", no Pspace (1018 entries: 94.tar.gz)

    Recursive 94nop-Koenigstuhl -- Battle Royale 94nop-Koenigstuhl

    Re-run with "-P" (all permutations) for the top-300 and "-r 1000" for all weaker warriors.
    The score on the "Royale" hill is sampled at the iteration when the warrior was in the last position.
    I also switched to pmars-0.9.4 and a 32-bit (little endian) score table (was 16-bit and pmars-0.9.2).

    Recent entries: (rank on recursive 94nop hill, program, author, date-of-entry)
    31: Discord (decoy) , John Metcalf, (Jun/08)
    63: Mercenary , John Metcalf, (Sep/06)
    68: Elixir , John Metcalf, (Jun/08)
    97: Trouble , John Metcalf, (Nov/08)
    104: Pendulum-4009 , John Metcalf, (Nov/08)
    113: He Scans Alone qhoz , Anton Marsden, (Nov/01)

  • 94nop-TOP-50 Hill, "-P", 94nop rules (15602 rounds)

    Any program that reaches rank-50 or higher on the 94nop Koenigstuhl will also enter the TOP-50 Koenigstuhl. Programs that fall below rank-50 on both (94nop and TOP-50) hills will be removed from the TOP-50 Koenigstuhl.

    TOP-50 of the 94nop-Koenigstuhl (94top50.tar.gz)

    Recent entries: (rank on TOP-50 hill, program, author, date-of-entry)
    12: Discord (decoy) , John Metcalf, (Jun/08)

    Recent leaves: (program, author, date)
    Herbal Avenger , Michal Janeczek , (Jun/08)
    HazyLazy C 11 , CS SG , (Jun/08)
    HazyLazy A 70 , Steve Gunnell , (Jun/08)
    Dawn 2 , Roy van Rijn , (Jun/08)
    Hairy Puffy Twins , Roy van Rijn , (Jun/08)
    paper(paper(paper(clear))) , Sascha Zapf , (Jun/08)

  • Pspace-hill(s), "-r 1000", Pspace ok (205 entries: pspace.tar.gz)

    Recursive Pspace-Koenigstuhl -- Battle Royale Pspace-Koenigstuhl

    Re-run with pmars-0.9.4 and a 32-bit (little endian) score table.

    Note: "-P" (randomizing all possible starting positions) will not make the battles reproducible, because the order of the fights makes a difference with P-space.

    Recent entries: (rank on recursive Pspace-Hill, program, author, date-of-entry)
    37: Electric Head chhm , Anton Marsden, (Nov/01)

  • Pspace-TOP-50 Hill, "-P", Pspace ok (15602 rounds)

    Any program that reaches rank-50 or higher on the "full" hill will also enter the TOP-50 hill. Programs that fall below rank-50 on both hills will be removed from the TOP-50 Koenigstuhl.

    TOP-50 of the Pspace-Koenigstuhl (psptop50.tar.gz)

    Recent entries: (rank on recursive Pspace-Hill, program, author, date-of-entry)
    46: Electric Head chhm , Anton Marsden, (Nov/01)

    Recent leaves: (program, author, date)

  • ICWS/64 hill(s): "-r 1000 -8 -s 8192 -p 64"

    Recursive ICWS/64-Koenigstuhl (475 entries: icws64.tar.gz)

    Flat (un-weighted) scores against all of the ICWS/64-hill. Scores against top 50%, 25%, 10% of the 'flat' ICWS/64-hill. (Bold typeface ranks indicate programs that are within the top N-% of the 'flat' hill)

  • ICWS hill(s): "-r 1000 -8 -s 8192 -p 8000 -c 100000"

    Recursive ICWS-Koenigstuhl (479 entries: icws.tar.gz)

    Flat (un-weighted) scores against all of the ICWS-hill. Scores against top 50%, 25%, 10% of the 'flat' ICWS-hill. (Bold typeface ranks indicate programs that are within the top N-% of the 'flat' hill)

    Recent entries: (rank on recursive ICWS-Hill, program, author, date-of-entry)
    9: Prometheus (decoy) T , John Metcalf, (Sep/06)

  • LP hill(s): "-r 1000 -p 8 -l 200 -d 200" (88, 94, Pspace)

    Recursive LP-Koenigstuhl (149 entries: lp.tar.gz)

    Flat (un-weighted) scores against all of the LP-hill. Scores against top 50%, 25%, 10% of the 'flat' LP-hill. (Bold typeface ranks indicate programs that are within the top N-% of the 'flat' hill)

    Recent entries: (rank on recursive LP-Hill, program, author, date-of-entry)
    1: R.O.U.S. , S.Fernandes, (Mar/22)

  • X hill(s): "-r 1000 -s 55440 -c 500000 -p 10000 -l 200 -d 200" (203 entries: x.tar.gz)

    Recursive X-Koenigstuhl Recursive X-Koenigstuhl

    I switched to pmars-0.9.4 and a 32-bit (little endian) score table (was 16-bit and pmars-0.9.2).

    Recent entries: (rank on recursive X-Hill, program, author, date-of-entry)
    14: Pendulum X , John Metcalf, (Nov/08)
    20: Despair X , John Metcalf, (Nov/08)

  • Tiny hill(s): "-P -s 800 -c 8000 -l 20 -d 20 -p 800" (1522 rounds) (436 entries: tiny.tar.gz)

    Recursive Tiny-Koenigstuhl -- Battle Royale Tiny-Koenigstuhl

    The score on the "Royale" hill is sampled at the iteration when the warrior was in the last position.
    I also switched to pmars-0.9.4 and a 32-bit (little endian) score table (was 16-bit and pmars-0.9.2).

    Recent entries: (rank on recursive Tiny-Hill, program, author, date-of-entry)
    1: dfd94f0-408f53a4-b1f60e08.rc , bvowk, (Sep/06)
    25: BiShot Redux , Schmidt/Metcalf, (Jun/08)
    52: The Black Knight Returns , Christian Schmidt, (Jun/08)
    143: Infernal Machine , John Metcalf, (Sep/06)
    356: Curiosity , John Metcalf, (Sep/06)

  • Nano hill(s): "-P -s 80 -c 800 -l 5 -d 5 -p 80" (142 rounds) (449 entries: nano.tar.gz)

    Recursive Nano-Koenigstuhl -- Battle Royale Nano-Koenigstuhl

    The score on the "Royale" hill is sampled at the iteration when the warrior was in the last position.
    I also switched to pmars-0.9.4 and a 32-bit (little endian) score table (was 16-bit and pmars-0.9.2).

    Recent entries: (rank on recursive Nano-Hill, program, author, date-of-entry)
    1: [RS] Nextratulated Sturvinator 2 , inversed, (Sep/20)
    2: Combat Arithmetic , inversed, (Sep/20)
    5: Another MEVO Thing , Terry Newton , (Jul/05)
    6: Nonlocality 2 , inversed, (Sep/20)
    31: Nabla , inversed, (Sep/20)
    50: Bifurcation , inversed, (Sep/20)
    56: Diffraction Limit S5 , inversed, (Sep/20)
    112: Diffraction Limit S42 , inversed, (Sep/20)

  • Odd hill: "-r 50 -l 800 -d 800" (88, 94 & Pspace) (1094 entries: odd.tar.gz)

    Hill for all programs that do not fit any other hill (eg. special tournament rounds or non-competitive programs).

    Flat (un-weighted) scores against all of the Odd-hill.

    Easter 2016 Tournament
    Tiny Stone Challange (May 2015)
    Valentine 2015 Tournament
    Halloween 2014 Tournament
    Spring 2014 Tournament

    "Flat" Scores:
    On each hill a new challanger fights a N-round battle against every programs (different from Mt.Olympus) already on the hill. The result of this battle changes the score for both opponents:

    For the programs already on the hill:
    new_score = {old_score * (#hill-1) + score_against_challenger} / #hill
    For the 'challenger':
    score = (score1 + score2 + ...) / (#hill)
    After doing all battles for a new challenger I re-order the hill and write the new hill-page(s).

    "Recursive" Scores:
    Current: The recursive algorigthm weights scores against better programs more and calculates a new hill from the data provided by the "flat" hill. This procedure is repeated (with decreasing absolute weights) until a stable hill is reached (or a cut-off of 25 programs is reached).

    Maybe in the future? The recursive algorighm starts with the round-robin (flat) scores of the entire hill. The bottom half is fixed in place, then the top half is re-evaluated only within itself. This process is repeated until the top part reaches a cut-off size of 50 programs.

    If you cannot find your program(s) here, please send your program to birk AT carnegiescience DOT edu with a subject line like 'Koenigstuhl entry'. Please note that the Koenigstuhl is an "archive" hill, ie. published programs only. If you would like to see how your program would do on the Koenigstuhl before you publish it, please download the appropriate hill and run your program against it.

    back to Christoph's CoreWar page
    1997-06-18, Christoph C. Birk (last updated: 2022-11-29)