LS-rankinglist: Houdini 3 tactical

[pohl4711]

The LS-rankinglist (LightSpeed-rankinglist), now with Houdini 3 tactical (Parameter Tactical Mode=true). -31 elo to Houdini 3. Excactly what R.Houdart expected...

Intel i7-2630QM (SSE42 support, Windows 7 64bit, 2 GHz Quadcore, FritzMark=20.2, Houdini (one core) 2.2 meganodes/s (middlegame)), 64 MB Hash, 1 core per engine (Hyperthreading off), no ponder, no endgame-bases, no resign. 500 selected opening-positions (all 8 moves deep, from Frank Q.-database)
Elos calculated with bayeselo (mm 0 1)(fixpoint Robbolito 0.085g3 3000 Elo). LittleBlitzerGUI (gauntlet-mode only, because this GUI chooses opening-positions per random in the round-robin-mode from the PGN-file...)
Time: 45''+500ms Fischerbonus (= 85-90 seconds per game/engine).

LS-rankinglist with best engine-versions only (no betas, no settings, no development-versions) (RoundRobin, 1000 games against each opponent):

Code: Select all

Rank Name                    Elo    +    - games score oppo. draws 
   1 Houdini 3 x64          3153    5    5 10000   67%  3028   42% 
   2 Critter 1.6a x64       3072    5    5 10000   55%  3036   53% 
   3 Strelka 5.5 x64        3069    5    5 10000   55%  3036   53% (singlecore)
   4 Komodo 5 x64           3059    5    5 10000   53%  3037   44% (singlecore)
   5 Robbolito 0.21Q x64s   3033    5    4 10000   49%  3040   57% 
   6 Ivanhoe 46h x64        3024    5    5 10000   48%  3040   56% (best open source)
   7 Bouquet 1.5 x64s       3019    5    5 10000   47%  3041   55% 
   8 Rybka 4.1 x64s         3012    5    5 10000   46%  3042   47% 
   9 Stockfish 2.3.1 x64s   3001    5    5 10000   44%  3043   46% 
  10 Robbolito 0.085g3 x64  3000    5    5 10000   44%  3043   54% (singlecore)(Ippolit 2009)
  11 Saros 3.0 x64          2987    5    5 10000   42%  3044   49% 

The results of the test-gauntlet:

Code: Select all

Houdini 3 tactical   - Critter 1.6a x64          588.0 - 412.0    +310/=556/-134    58.80%
Houdini 3 tactical   - Strelka 5.5 x64           555.0 - 445.0    +299/=512/-189    55.50%
Houdini 3 tactical   - Komodo 5 x64              568.5 - 431.5    +360/=417/-223    56.85%
Houdini 3 tactical   - Robbolito 0.21Q x64s      616.5 - 383.5    +393/=447/-160    61.65%
Houdini 3 tactical   - Ivanhoe 46h x64           649.0 - 351.0    +438/=422/-140    64.90%
Houdini 3 tactical   - Bouquet 1.5 x64s          647.5 - 352.5    +407/=481/-112    64.75%
Houdini 3 tactical   - Rybka 4.1 x64s            637.0 - 363.0    +431/=412/-157    63.70%
Houdini 3 tactical   - Stockfish 2.3.1 x64s      669.0 - 331.0    +474/=390/-136    66.90%
Houdini 3 tactical   - Robbolito 0.085g3 x64     665.5 - 334.5    +458/=415/-127    66.55%
Houdini 3 tactical   - Saros 3.0 x64             686.0 - 314.0    +495/=382/-123    68.60%

Results of Houdini 3 against the same opponents:

Code: Select all

Houdini 3 x64   - Critter 1.6a x64          619.0 - 381.0    +377/=484/-139    61.90%
Houdini 3 x64   - Strelka 5.5 x64           625.0 - 375.0    +375/=500/-125    62.50%
Houdini 3 x64   - Komodo 5 x64              612.5 - 387.5    +421/=383/-196    61.25%
Houdini 3 x64   - Robbolito 0.21Q x64s      674.5 - 325.5    +458/=433/-109    67.45%
Houdini 3 x64   - Ivanhoe 46h x64           695.0 - 305.0    +487/=416/-97    69.50%
Houdini 3 x64   - Bouquet 1.5 x64s          680.0 - 320.0    +460/=440/-100    68.00%
Houdini 3 x64   - Rybka 4.1 x64s            693.0 - 307.0    +508/=370/-122    69.30%
Houdini 3 x64   - Stockfish 2.3.1 x64s      679.0 - 321.0    +491/=376/-133    67.90%
Houdini 3 x64   - Robbolito 0.085g3 x64     715.5 - 284.5    +518/=395/-87    71.55%
Houdini 3 x64   - Saros 3.0 x64             736.0 - 264.0    +556/=360/-84    73.60%

The complete LS-rankinglist:

Code: Select all

   1 Houdini 3 x64          3151    5    5 13000   69%  3016   41% 
   2 Houdini 3 tactical     3120    5    5 10000   63%  3027   44% 
   3 Houdini 2.0c x64       3104    5    5 11000   62%  3015   42% 
   4 Houdini 1.5a x64       3084    5    5 10000   59%  3017   44% (best freeware (multicore))
   5 Strelka 5.5 x64        3072    4    4 17000   55%  3035   52% (singlecore)
   6 Critter 1.6a x64       3070    4    4 17000   55%  3035   53% 
   7 Komodo 5 x64           3061    4    4 17000   54%  3035   43% (singlecore)
   8 Robbolito 0.21Q x64s   3033    5    5 11000   48%  3046   56% 
   9 Ivanhoe 46h x64        3021    4    4 17000   48%  3038   54% (best open source)
  10 Robbolito 0.10 x64s    3019    4    4 15000   48%  3033   55% 
  11 Bouquet 1.5 x64s       3018    5    5 13000   47%  3041   56% 
  12 Rybka 4.1 x64s         3012    4    4 17000   46%  3038   46% 
  13 Robbolito 0.085g3 x64  3000    4    4 17000   44%  3039   53% (singlecore)(Ippolit 2009)
  14 Stockfish 2.3.1 x64s   3000    4    4 15000   45%  3039   45% 
  15 Stockfish 2.2.2 x64s   2995    5    5 11000   45%  3033   44% 
  16 Saros 3.0 x64          2988    4    4 17000   43%  3040   47% 
  17 Gull 2 beta2 x64       2985    5    5 11000   42%  3037   53% 
  18 Bouquet 1.4 x64s       2929    5    5 13000   34%  3044   42% 

(x64=64bit version, x64s=64bit SSE42-version)


deleted betas, development-versions, settings: none
aborted test-gauntlets (because of too bad results): Saros 3.1aDistro (-130 Elo to Saros 3.0 (2500 games))


If you want to get the games from the LS-list, send me an eMail-adress per PM. I will send the games (PGN-file) as soon as possible...

Greetings – Stefan

[Kyodai]

"aborted test-gauntlets (because of too bad results): Saros 3.1aDistro (-130 Elo to Saros 3.0 (2500 games))"

Oh, wonder what happened to Saros..

[carldaman]

There was a bad release of DS 3.1, then a fixed version. Which one was this?

[pohl4711]

There was a bad release of DS 3.1, then a fixed version. Which one was this?

It was the Version 3.1a Distro, which is still the newest on the Deep Saros Homepage...

Stefan