Time management: Rybka vs IvanHoe

As in chess tournaments and matches...
benstoker
Posts: 110
Joined: Thu Jun 10, 2010 7:32 pm
Real Name: Ben Stoker

Re: Time management: Rybka vs IvanHoe

Post by benstoker »

kingliveson wrote:

Code: Select all

Rybka vs IvanHoe , Blitz:4'+2  2010 

1   IvanHoe 9.55b x64       +14/=29/-7 57.00%   28.5/50
2   Deep Rybka 4 SSE42 x64  +7/=29/-14 43.00%   21.5/50
† 25 Positions (test suite)
† 3-4-5 Nalimov/Robbobase
† 256 MB Hash
† Ponder Off
† Phenom II 940
† 4 Cores
† 4+2 Time Control

Rybka
TC Buffer 3
TC Normal Move Time 100
TC Max Move Time 150

IvanHoe
TimeEasyFactor 18
TimeOrdinaryFactor 69
TimeAbsolutePercent 23
BufferTime 2000
Pawn Hash 16 MB
AllowInstantMoveFromHash Off


Game data available: http://chess.cygnitec.com/pgn/

100 game 4+2 match currently running on the slower Athlon 64 x2 system with above time management settings -- after 58 games, +19/=30/-9 in favor of IvanHoe.
Can you theorize why turning off AllowInstantMoveFromHash seems to increase ELO?
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

benstoker wrote:
Can you theorize why turning off AllowInstantMoveFromHash seems to increase ELO?
Hey Ben, what AllowInstantMoveFromHash does is try to gain time advantage by moving quickly on hash it -- that is if the opponent played the expected move. For simplicity, you could compare it to move on ponder hit. As you see, it says to turn it off when pondering is enabled. Can you see why it would have some side effect?!

here is the code:

Code: Select all

 /* "PREVIOUS_FAST" is inverse named plus don't too this with UCI_PONDER */
  if (EXACT_DEPTH >= PREVIOUS_DEPTH - 6 && EXACT_MOVE == HASH_MOVE
      && !UCI_PONDER && EXACT_MOVE && PREVIOUS_FAST && PREVIOUS_DEPTH >= 18
      && ALLOW_INSTANT_MOVE && MyOK (POSITION, EXACT_MOVE)
      && Value < 25000 && Value > -25000)
    {
      ROOT_SCORE = Value;
      ROOT_BEST_MOVE = EXACT_MOVE;
      ROOT_DEPTH = EXACT_DEPTH;
      PREVIOUS_FAST = FALSE;
      if (!IsCheck)
	v = MyExclude (POSITION, Value - 50, PREVIOUS_DEPTH - 6, EXACT_MOVE);
      else
	v = MyExcludeCheck (POSITION, Value - 50, PREVIOUS_DEPTH - 6, EXACT_MOVE);
      if (v < Value - 50)
	return;
    }
PAWN : Knight >> Bishop >> Rook >>Queen
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

Code: Select all

Ivanhoe v Rybka R008, Blitz:4'+2  2010                     

1   IvanHoe 9.55b x64  +29/=54/-17 56.00%   56.0/100
2   Deep Rybka 4 x64   +17/=54/-29 44.00%   44.0/100
† 50 Positions (test suite)
† 3-4-5 Nalimov/Robbobase
† 128 MB Hash
† Ponder Off
† Athlon 64 X2 6000+ @ 3100 MHz
† 2 Cores
† 4+2 Time Control

Rybka
TC Buffer 3
TC Normal Move Time 100
TC Max Move Time 150

IvanHoe
TimeEasyFactor 18
TimeOrdinaryFactor 69
TimeAbsolutePercent 23
BufferTime 2000
Pawn Hash 8 MB
AllowInstantMoveFromHash Off


Game data available: http://chess.cygnitec.com/pgn/

More games needed. Also, others can use these settings to see if similar results can be reproduced.
PAWN : Knight >> Bishop >> Rook >>Queen
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

Code: Select all

MS, 10'/40+10'/40+10'/40  2010             

1   IvanHoe 9.55b w32  +15/=28/-7 58.00%   29.0/50
2   Rybka 4 w32        +7/=28/-15 42.00%   21.0/50
† 25 Positions (test suite)
† No Tablebases
† 128 MB Hash
† Ponder Off
† AMD Turion(tm) 64 Mobile Technology ML-37
† 1 CPU
† 40/10 Repeating Time Control

Rybka
TC Buffer 3
TC Normal Move Time 100
TC Max Move Time 150

IvanHoe
TimeEasyFactor 16
TimeOrdinaryFactor 64
TimeAbsolutePercent 24
BufferTime 2000
Pawn Hash 8 MB
AllowInstantMoveFromHash Off


Game data available: http://chess.cygnitec.com/pgn/
PAWN : Knight >> Bishop >> Rook >>Queen
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

Code: Select all

Rybka vs IvanHoe , Blitz:4'+2  2010                      

1   IvanHoe 9.55b x64       +18/=70/-12 53.00%   53.0/100
2   Deep Rybka 4 SSE42 x64  +12/=70/-18 47.00%   47.0/100
† 50 Positions (test suite)
† 3-4-5 Nalimov/Robbobase
† 256 MB Hash
† Ponder Off
† Phenom II 940
† 4 Cores
† 4+2 Time Control

Rybka
TC Buffer 3
TC Normal Move Time 100
TC Max Move Time 150

IvanHoe
TimeEasyFactor 18
TimeOrdinaryFactor 69
TimeAbsolutePercent 23
BufferTime 2000
Pawn Hash 16 MB
AllowInstantMoveFromHash Off


Game data available: http://chess.cygnitec.com/pgn/
PAWN : Knight >> Bishop >> Rook >>Queen
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

Sample long (40/40 repeating) games being tested on 2 systems. Exact positions and engine settings. The difference is, No EGTB for the much much slower hardware, and of course, Hash is scaled to fit each system. This is a preview on how Rybka/IvanHoe manage time. The sample games can be plugged into your favorite GUI -- and you can see how they spend time through 40 moves, including difficult and not so difficult positions. Rybka is using recommended Time management configuration for repeating games as is IvanHoe. Use drop menu to select game, E1 to move forward, D1 to move backward, and D8 to get full PGN source.

4 CPUs:


1 CPU:


Edit: IvanHoe version used in all these tests is unmodified from source, and is available here: http://chess.cygnitec.com/engine/
PAWN : Knight >> Bishop >> Rook >>Queen
User avatar
DaveD
Posts: 38
Joined: Thu Jun 10, 2010 3:04 am

Re: Time management: Rybka vs IvanHoe

Post by DaveD »

kingliveson wrote: Edit: IvanHoe version used in all these tests is unmodified from source, and is available here: http://chess.cygnitec.com/engine/
None of the 9.55b w32 versions work for me on XP! I get this error message -

Image
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

DaveD wrote:
kingliveson wrote: Edit: IvanHoe version used in all these tests is unmodified from source, and is available here: http://chess.cygnitec.com/engine/
None of the 9.55b w32 versions work for me on XP! I get this error message -

Image

You can easily resolve the problem by getting Windows 7 Professional Upgrade version for only $199.99.
PAWN : Knight >> Bishop >> Rook >>Queen
User avatar
DaveD
Posts: 38
Joined: Thu Jun 10, 2010 3:04 am

Re: Time management: Rybka vs IvanHoe

Post by DaveD »

kingliveson wrote:
DaveD wrote:
kingliveson wrote: Edit: IvanHoe version used in all these tests is unmodified from source, and is available here: http://chess.cygnitec.com/engine/
None of the 9.55b w32 versions work for me on XP! I get this error message -

Image

You can easily resolve the problem by getting Windows 7 Professional Upgrade version for only $199.99.
You must be kidding! :roll:
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Time management: Rybka vs IvanHoe

Post by kingliveson »

DaveD wrote:
kingliveson wrote:
DaveD wrote:
kingliveson wrote: Edit: IvanHoe version used in all these tests is unmodified from source, and is available here: http://chess.cygnitec.com/engine/
None of the 9.55b w32 versions work for me on XP! I get this error message -

Image

You can easily resolve the problem by getting Windows 7 Professional Upgrade version for only $199.99.
You must be kidding! :roll:
Please see: http://www.open-chess.org/viewtopic.php ... 3693#p3693
PAWN : Knight >> Bishop >> Rook >>Queen
Post Reply