IvanHoe V52 Tests

First topic message reminder :

[img][You must be registered and logged in to see this link.][/img]

KLO.s popcount vrs LeonardodoVincis Non popcount beta . same books + triple bases :cheers:

i fell asleep again :laf gal:

Depth 20 for opening move ?

the speeds absolutly wrong . have a look at task manager during game as this realy seems wrong , we are AMD quad but have 2 core as well and ivanhoes much much faster than the speeds the screen shot shows

Oh, you did it in Ponder ON? I think that's wrong if the engines are using all the cores at the same time, because they have to fight for the resources of your computer and these problems may appear. Test it at ponder OFF or use 1 core for each engine and activate Ponder.

Quote :

I have check it and it use 2 cpu

my system is XP64bits, may be i need to move to the last windows OS.
I will check it and i will do a new tournament when all will be fine.

Sorry for this mistake.
Nonetheless, is it normal than now Ivanhoe cannot compile on linux OS ?

Quote :

Oh, you did it in Ponder ON? I think that's wrong if the engines are using all the cores at the same time, they fight for the resources and these problems may appear. Test it at ponder OFF or use 1 core for each engine and activate Ponder .

ok i will verify it too.

Houdinis speed is far to fast

Peter Pan writes this to us


the speed of houdini is incorrect,especially when i look at his processors speed.

First he says houdini reaches 3000 kn's,which could be true... but that ivanhoe is 500kns,which is wrong.

Then... i see the screenshot where houdini has 4077 at starting position... hmmm

okay let me run the 2 core houdini verion on my pc using 2 cores from starting position...

and i will get back to you just now... so that we can compare with his speed and see if that is a possibility.Still does not explain slow speed of assmeds compile.

theropodes wrote:

my system is XP64bits, may be i need to move to the last windows OS.
I will check it and i will do a new tournament when all will be fine.

Sorry for this mistake.
Nonetheless, is it normal than now Ivanhoe cannot compile on linux OS ?

Quote :

Oh, you did it in Ponder ON? I think that's wrong if the engines are using all the cores at the same time, they fight for the resources and these problems may appear. Test it at ponder OFF or use 1 core for each engine and activate Ponder .

ok i will verify it too.

ponder on shoulnt cause this anomaly as it affects both engines

we test Ponder OFF so engines get full resources of cpu

Ivanhoe compile easily on linux

Peter Pan writes.............
i ran houdini-2core version x64 using only 2 cores @ starting position and get 3000kn's

He gets 4000kn's plus on his 2.2 ghz pc.Impossible!................

well,something seriously went wrong with that test

The speed of Houdini is too high,unless he has highly overclocked his pc ?! and it's running at close to 4 ghz,then that is possible,maybe,but still unlikely.

The speed of Ivanhoe is too low.Now this could be a glitch in my "friends" compile :)

Which is highly possible lol :)


scary stuff.

i dont' overclok my pc so i think i will format all and install last windows version.
i think your compile is good.

[You must be registered and logged in to see this link.] is the site where i can found the last version ?

theropodes wrote:

i dont' overclok my pc so i think i will format all and install last windows version.
i think your compile is good.

[You must be registered and logged in to see this link.] is the site where i can found the last version ?

the source codes from Decemberists can be found right under your nose here
( Ivanhoe engines AND codes ) or on Ippolite wikie

Hello great team,

good new, after formating windows xp 64 and use Fritz12 gui and taken the Houdini642Cpu and Ivanhoe T0.4beta2
i have done a tournament in 10 minutes by game.

Core2Duo 2.33 2CPU Blitz 10m 0

1 Houdini 1.03a x64 2_CPU 1½½½½½½110 6.0/10
2 IvanHoe T0.4 beta2 0½½½½½½001 4.0/10

I have attached the tournament.
At the beginning, Ivanhoe play with around 2000Kn/S
It take more time to analyse and at the middle game there is a difference of two minutes between Houdini and Ivanhoe.
At the beginning, the depth is around 20
At the middle game, the depth is 20
At the endgame, it is 24

For me, Ivanhoe is the powerful free chess engine.

Regards
TheRoPoDeS

were Bases on ivan ?

it's me, you're patient with me lol

it was without bases : only default configuration for the two engines.
I have like the 10 game when Ivanhoe found at the end , the good move to take the definitive advantage.

Regards
TheropodeS

theropodes wrote:

it's me, you're patient with me lol

it was without bases : only default configuration for the two engines.
I have like the 10 game when Ivanhoe found at the end , the good move to take the definitive advantage.

Regards
TheropodeS

If you can repeat test but with TripleBases on ivanhoe as houdini has built in base of some sort

ok i will downloaded it and start the test.
i need to found some time but i will do it

theropodes wrote:

ok i will downloaded it and start the test.
i need to found some time but i will do it

download it ?

to repeat test without robobases you need to download what ?

The test done was without robbobase
So now, if you want, i download Robbobase files and indicate to the engine parameter where found it.

may be i have missing something or done something wrong ?

theropodes wrote:

The test done was without robbobase
So now, if you want, i download Robbobase files and indicate to the engine parameter where found it.

may be i have missing something or done something wrong ?

NO , we assumed as you used IvanHoe you would allready have TripleBaseses , its real quite a lot stronger with them , our mistake to be presumptuose sorry

ok no problem
my two nuts, is that i prefer tested and use an engine without book and endgame base : only the knowledge of programmers with the engine code.

i'm waiting a new version, with Vlad0 correction, and i will start the same test.

Regards
TheropodeS

theropodes wrote:

ok no problem
my two nuts, is that i prefer tested and use an engine without book and endgame base : only the knowledge of programmers with the engine code.

i'm waiting a new version, with Vlad0 correction, and i will start the same test.

Regards
TheropodeS

[You must be registered and logged in to see this link.]

that is the sleep bug fixed version

Peter Pan posts

Blitz 1m+1s 2010


1 Ivanhoe B 52aHx64 +32 +23/=98/-11 54.55% 72.0/132
2 IvanHoe T0.4 beta4 x64 -32 +11/=98/-23 45.45% 60.0/132[/CODE]

No Robbobases was used.
Hash 1024 megs each.
LargePages OFF
Windows 7 x64
Q6600 @ 3.4 ghz.

B52aH speed was slighly faster than T 0.4 beta 4

This is a work in progress,and compile not ready for release yet...
I tried a new style in this test,i call it the "crazymanstyle"
Seems to play more aggressive,plus other changes made as well in source.

[You must be registered and logged in to see this link.]

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#

Skulltrail Intel QX Yorkfield (8X4090 MHz)
Time: 3'+0" . Ponder off. GUI Fritz 12. Hash 256. Tablebases 3-4-5 + Robbo Triple Bases. Book: HS-7moves plus.ctg.

Deep Rybka 4 - IvanHoe T0.4 beta5 x64 :
27.5 - 22.5 . ( +11 - 6 = 33) . One crash in game 43: in winning
position for IvanHoe with result Deep Rybka - IvanHoe 0-1 instead 1-0.
This beta 5 is weaker as the other beta compiles.

Sandi

Intel Core2Duo 2.33

Ivanhoe Vs Houdini 0


1 Houdini 1.03a x64 2_CPU +108 ½½½11½½½½1 6.5/10
2 IvanHoe T0.4 beta5 x64 -108 ½½½00½½½½0 3.5/10

Blitz five minutes 10 games : not so good

Like say Paul Keres, this is the road to the top

Tournament 10 games 7 minutes by game Core2Duo 2CPU 2.33

1 Houdini 1.03a x64 2_CPU +35 ½½½½½½½½1½ 5.5/10
2 Ivanhoe 52hUS-x64 -35 ½½½½½½½½0½ 4.5/10

Second step : the road to the perfection

Continue like this peter pan, you're rock

New Tournament result :

Ivanhoe Vs Rest Of The World 2010

Ivanhoe 52hUS-x64 - Houdini 1.03a x64 2_CPU 4.5 - 2.5 +3/=3/-1 64.29%
Ivanhoe 52hUS-x64 - Deep Rybka 4 x64 2.5 - 4.5 +2/=1/-4 35.71%

For me it is very strange because i was thinking than Houdini was the target, but it seems now it DeepRybka4

Configuration : Core2Duo 2.33 2 CPU XP64
Can we think than now, the second step begin : the road to the perfection

theropodes wrote:

New Tournament result :

Ivanhoe Vs Rest Of The World 2010

Ivanhoe 52hUS-x64 - Houdini 1.03a x64 2_CPU 4.5 - 2.5 +3/=3/-1 64.29%
Ivanhoe 52hUS-x64 - Deep Rybka 4 x64 2.5 - 4.5 +2/=1/-4 35.71%

For me it is very strange because i was thinking than Houdini was the target, but it seems now it DeepRybka4

Configuration : Core2Duo 2.33 2 CPU XP64
Can we think than now, the second step begin : the road to the perfection

Thank you
very strange result , though in all honesty seven games realy is not sufficent to tell much
, pleas try to run more ,30 would give the engines a better tesr test enen at lower tc if increment was added though the longer the games the more depth is reached
which gives a better idea of how good the compiles are

Skulltrail Intel QX 9775 Yorkfield (8X4090 MHz)
Benchmark 18.375 k/N/s
Time 3'+0" (equivalent to 7.5' on a standard quad and 10' on a dual core)
Hash 256. Ponder off. GUI Fritz 12. Tablebases 3-4-5 + Robbo Triple Bases.
Book: HS-7moves plus.ctg.

IvanHoe 52hUSo1-x64 - Deep Rybka 4 : 27.5 - 22.5 55% -45% .
( + 12 - 7 = 31 ) !!
After the analysis of the 50 games I have decided to play still a match with another 100 games on the same conditions to avoid random.
This compile has confirmed strength and a good scaling to 8 cores.
Congratulation to Vlad0 !!

Match 2 (100 games):
IvanHoe 52hUSo1-x64 - Deep Rybka 4 : 51 - 49 . 51% - 49% .
( +24 -22 = 54 )

Total after 150 games:
IvanHoe 52hUSo1-x64 - Deep Rybka 4 :
78.5/150 - 71.5/150 ( +36 - 29 = 85 )

Sandi
~~~~~~~~~~~~~~~~


Thank you Janee :gal4: