OpenChess

AMD Phenom(tm) II X6 1090T Processor6x @ 3.2 GHz 4,096 MB Memory
Windows 7 Home Premium Edition (Build 7600)
Fritz Benchmark:
Speed: 23.33
KNS: 11196
GUI: CB Rybka 3
Book: Perfect 2009- 10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

Nice experiment and improvement!

i7 940 @ 3.75 / 2 cores ea / ponder off / F11 / Vistax64 / 512 mb / klo openings / egtb + old Robbobases - On

Time Control : 4 mins + 2 sec

Rybka 4................53..........+19,-13,=68
IH B50tAx64p.........47


Rybka 4..............53............+22,-16,=62
IH T0.5.4.2 .........47

Tom Glenn

Linux/Windows builds for beta versions 999950r and 999950t are now available at http://chess.cygnitec.com/engine/ivanhoe/

The attached 999950t file to this post is just a build with an alternative compiler.

Intel(R) Core(TM) i7 Q- 960 3.2 GHz8x @ 4.005 GHz 6.00 MB Memory
Windows 7 Professional Professional (Build 7600)
Fritz Benchmarks:
Speed: 24.89
KNS: 11946
GUI: CB Rybka 3
Hash: 256
Book: HS-7 moves
RB and TB: ON
Ponder: OFF
Blitz:10' 0

Slight bug fix for [url=hhttp://chess.cygnitec.com/engine/ivanhoe/IvanHoe999950t.7z]IvanHoe 999950t[/url] 'Analysis Mode' only. No point in correcting 999950r and previous builds. The "bug" is not found in the original source.

What happened is when auto thread detect and uci parameter was added, the code depended on string "Ponder" being present to insert 'Threads' setting in the appropriate position when the for loop iterates through the struct array. But when building 'Analysis Mode,' 'Ponder' which is made available through preprocessor in 'Game Mode' is no longer present. So one cannot change number of cores when using 'Analysis Mode' -- fixed and uploaded!

Intel(R) Core(TM)2 Quad Q9550 2.83GHz4x @3.5 GHz 4,096 MB Memory
Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)
Fritz Benchmark:
Speed: 20.90
KNS: 10032
GUI: CB Rybka 3
Book: HS-10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

kingliveson wrote:Slight bug fix for [url=hhttp://chess.cygnitec.com/engine/ivanhoe/IvanHoe999950t.7z]IvanHoe 999950t[/url] 'Analysis Mode' only.

How does this KLO compile compare with the PeterPan one (IvanhoeB50tAX64)?

Intel(R) Core(TM)2 Quad Q9550 2.83GHz4x @3.5 GHz 4,096 MB Memory
Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)
Fritz Benchmark:
Speed: 20.90
KNS: 10032
GUI: CB Rybka 3
Book: HS-10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

ernest wrote:

kingliveson wrote:Slight bug fix for [url=hhttp://chess.cygnitec.com/engine/ivanhoe/IvanHoe999950t.7z]IvanHoe 999950t[/url] 'Analysis Mode' only.

How does this KLO compile compare with the PeterPan one (IvanhoeB50tAX64)?

The approach/philosophy is a little different when looking at the different compiles. In the begining, things were really disorganized, so the goal was always to have a place where you could find all the builds and released sources. Also, to have a build that reflected the developers' intent -- a generic build with no bells and whistles. Time management settings were tweaked in a couple builds, which probably hurt more than helped, so it is now back to default.

There is good chance that Peterpan's and ahmed's builds might be a few ELO stronger because they tend to tweak some values to get the best performance. Here is a post by Ahmed:

Ahmed wrote: First, Thanks to all testers in the Project, especially to Roparz
after a lot of tests i think this Material values is better than the Default values about 7 ELO
Pawn = 100
Knights = 325
Bishop = 350
Rook = 500
Queen = 950
I will use this values in all my compiles

and this test for T0.5.4.2 too by Roparz
Intel Core i7 X 980 @ 3.33GHz
Threads = 12
Book:.SilverSuite (50 positions)
Time: 1m + 0s

Code: Select all

IvanHoe T0.5.4.2 x64            +49 / =111 / -40 52.25%
Deep Rybka 4 SSE42 x64          +40 / =111 / -49 47.75%

Other strong compiles come from Vlad0, though I haven't seen any builds of the latest IvanHoe sources released. The best part of it all is choice.

Franklin