New version of NoonianChess Rev 3.0

[Charles Roberson]

Geschrieben von: / Posted by: Charles Roberson at 01 July 2001 11:20:47:

Currently Pentium Pro Binaries only (Linux & MS).
Pentium tomorrow after some sleep.
As usual, I'd like to see time to 7 ply benchmarks on various
machines. Just mail me results of nc3_0 -bench -fd 7
Time to 9 ply is interesting as well.
New version of NoonianChess released: version 3.0.
http://members.home.net/noonian/GameThe ... nload.html
Lots of modifications and bug fixes.
o Fractional extensions
o King Attack and defences
o Algorithm restructuring to allow later mods
o More endgame knowledge
o Timing algorithm changed
o Extensions added and maxextensions increased
o Quiescence algorithm rewritten
o Pawn Structure knowledge improved and bug fixed
o Opening penalties added/increased for poor moves
o Parameter tuning yielded 20% for plys less than 6
and 2.5x speedup for 9 ply searches.
Performance comparison to rev 2.12a
Pentium III 800 MHz tc=5 mps=200
Rev 2.12a NC Rev 3.0 NC
W L D %Pts W L D %Pts
Betsy 5.1 12 33 5 29 3 4 1 44
Averno031 5 27 18 28 9 19 72 45

[Pierre Muzac]

Geschrieben von: / Posted by: Pierre Muzac at 01 July 2001 15:28:31:
Als Antwort auf: / As an answer to: New version of NoonianChess Rev 3.0 geschrieben von: / posted by: Charles Roberson at 01 July 2001 11:20:47:

Currently Pentium Pro Binaries only (Linux & MS).
Pentium tomorrow after some sleep.
As usual, I'd like to see time to 7 ply benchmarks on various
machines. Just mail me results of nc3_0 -bench -fd 7
Time to 9 ply is interesting as well.
New version of NoonianChess released: version 3.0.
http://members.home.net/noonian/GameThe ... nload.html
Lots of modifications and bug fixes.
o Fractional extensions
o King Attack and defences
o Algorithm restructuring to allow later mods
o More endgame knowledge
o Timing algorithm changed
o Extensions added and maxextensions increased
o Quiescence algorithm rewritten
o Pawn Structure knowledge improved and bug fixed
o Opening penalties added/increased for poor moves
o Parameter tuning yielded 20% for plys less than 6
and 2.5x speedup for 9 ply searches.
Performance comparison to rev 2.12a
Pentium III 800 MHz tc=5 mps=200
Rev 2.12a NC Rev 3.0 NC
W L D %Pts W L D %Pts
Betsy 5.1 12 33 5 29 3 4 1 44
Averno031 5 27 18 28 9 19 72 45

Why does noonian accept en passant but won't capture en passant.What if the advantage of a position lies in the "en passant" capture.Isn't it a weakness in strategic play?

[Charles Roberson]

Geschrieben von: / Posted by: Charles Roberson at 01 July 2001 19:44:58:
Als Antwort auf: / As an answer to: Re: New version of NoonianChess Rev 3.0 geschrieben von: / posted by: Pierre Muzac at 01 July 2001 15:28:31:

Currently Pentium Pro Binaries only (Linux & MS).
Pentium tomorrow after some sleep.
As usual, I'd like to see time to 7 ply benchmarks on various
machines. Just mail me results of nc3_0 -bench -fd 7
Time to 9 ply is interesting as well.
New version of NoonianChess released: version 3.0.
http://members.home.net/noonian/GameThe ... nload.html
Lots of modifications and bug fixes.
o Fractional extensions
o King Attack and defences
o Algorithm restructuring to allow later mods
o More endgame knowledge
o Timing algorithm changed
o Extensions added and maxextensions increased
o Quiescence algorithm rewritten
o Pawn Structure knowledge improved and bug fixed
o Opening penalties added/increased for poor moves
o Parameter tuning yielded 20% for plys less than 6
and 2.5x speedup for 9 ply searches.
Performance comparison to rev 2.12a
Pentium III 800 MHz tc=5 mps=200
Rev 2.12a NC Rev 3.0 NC
W L D %Pts W L D %Pts
Betsy 5.1 12 33 5 29 3 4 1 44
Averno031 5 27 18 28 9 19 72 45

Why does noonian accept en passant but won't capture en passant.What if the advantage of a position lies in the "en passant" capture.Isn't it a weakness in strategic play?

It is a weakness. Currently, that piece of code doesn't work. It is
bubbling up to the top of my list of things to fix/improve. The problem
is that I rarely get more than 4 hours per week to work on Noonian and
those 4 hours are usely spread across 3 days. I hope to get it fixed in
version 3.1 as I think it is now the number 1 or 2 issue holding Noonian's
rating down.