IvanHoeV57 /V55 Tests

First topic message reminder :

[img][You must be registered and logged in to see this link.][/img]
same books , ponder off
Whilst test is an unfair one as DR4 had no Namilov nor large pages we let it run on to test integrity of Ivan 57 and it was exellent, could be faster but , it was stable , no crashes and read bases
Anyoe who can enable Large page and or Namilov please test same engines

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


i stopped test at +5 =16-0 as i felt embaressed for this engines maker

Q9550-XP64, Blitz:10' 0

1 IvanHoe 9.57b x64 3185 +10/-6/=34 54.00% 27.0/50
2 Deep Rybka 4 x64 3190 +6/-10/=34 46.00% 23.0/50

Four times now i have tried to get this compile to load the bases , written in the correct path each time ...
But there is NO loadbases box....an oversight of JRs perhaps

So all four engine test have begun without its use of bases
however it is a strong engine and with bases may be a contender for top spot or second place. i think this short and unfair test (as 9.57b had bases ) confirms its potential

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

EDIT.....
This engine is strong , score now 4 wins each 22 draws

To me the results seem to be all over the place.

I ran another marathon tournemnt with T57A engine and did 188 games against DR4SSE4.2 X64 and T57A was only loosing by 2 games against rybka4. in Time control of 1' +1" More than likely as more results come in looks like .57 engines are about same strength as rybka4 basically. Back to the drawing board for the Deciembrists

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

To me the results seem to be all over the place.

I ran another marathon tournemnt with T57A engine and did 188 games against DR4SSE4.2 X64 and T57A was only loosing by 2 games against rybka4. in Time control of 1' +1" More than likely as more results come in looks like .57 engines are about same strength as rybka4 basically. Back to the drawing board for the Deciembrists

test better versions then

Maybe true. But newer versions are not proving to be any better. I retested this one since i had orginally gotten very good results. I see now that maybe 57A was lucky. Anyway newer versions are not really showing an improvment. Atleast not yet.

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

Maybe true. But newer versions are not proving to be any better. I retested this one since i had orginally gotten very good results. I see now that maybe 57A was lucky. Anyway newer versions are not really showing an improvment. Atleast not yet.

test KLO's 57a and use 9.57b engine which is in folder ..then test JRs engine..basic 64 bit no bases and extremely strong

by slankmen

Hi JR,
his was a great score.So far the best on this quad against Rybka 4 and on 10 min. per engine.
I just wish that the RBs were working also but looks like they were not that important in this match.

Q9550-XP64, Blitz:10' 0

1 IvanHoe 57a2.x64 JR 3185 +11/-4/=15 61.67% 18.5/30
2 Deep Rybka 4 x64 3190 +4/-11/=15 38.33% 11.5/30


it strong allright

Bettina (Sobel ?) wrote:

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

Maybe true. But newer versions are not proving to be any better. I retested this one since i had orginally gotten very good results. I see now that maybe 57A was lucky. Anyway newer versions are not really showing an improvment. Atleast not yet.

test KLO's 57a and use 9.57b engine which is in folder ..then test JRs engine..basic 64 bit no bases and extremely strong

Unfortunately i cannot test any of the 9.57x series since they are not built properly to load up on a I7 980x even though they load fine for everyone else. I am limited because of this engine loading problem. I may try the jr engine.

[quote="wajit@cfl.rr.com"]

Bettina (Sobel ?) wrote:

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

Maybe true. But newer versions are not proving to be any better. I retested this one since i had orginally gotten very good results. I see now that maybe 57A was lucky. Anyway newer versions are not really showing an improvment. Atleast not yet.

test KLO's 57a and use 9.57b engine which is in folder ..then test JRs engine..basic 64 bit no bases and extremely strong

Unfortunately i cannot test any of the 9.57x series since they are not built properly to load up on a I7 980x even though they load fine for everyone else. I am limited because of this engine loading problem. I may try the jr engine.[/quo sorry thought you had 2 pcs ...yes try JRS as even without bases its very strong ...on AMD anyway ,hope its ok for your beast

"B" wrote:

Four times now i have tried to get this compile to load the bases , written in the correct path each time ...
But there is NO loadbases box....an oversight of JRs perhaps

So all four engine test have begun without its use of bases
however it is a strong engine and with bases may be a contender for top spot or second place. i think this short and unfair test (as 9.57b had bases ) confirms its potential

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

EDIT.....
This engine is strong , score now 4 wins each 22 draws

Ivanhoe back

long life to IvanHoe

Despite the "claims " by immortals tester "slackman" we see no improvement in JRs compile now it reads RobboBases...in fact we see it crashing now, something previose versions did not do, twice in less than 14 games its crashed , i took screen shot of the test after first crash

/image_preview.php?i=55&u=14751437][/url][/img]

as it shows it is equal ( as its lose was due to crashing ) in this short test to the KLO compile which to date is the strongest V57 weve found for AMD pc's, we will try more games at longer TC and hope others will also test this as we had very high hopes it would meet expectations

EDIT...

16 more games without another crash , honours still even . more tests later

Engine to "Buggy " after crashes ,,many vrs rykba4 it stops reading its bases allthough pathway is still in paramiters , this compile needs :gal4: it is not well as now without bases is much weaker than before it was lost to DR$ very badly when i saw this i watch games and see no "tb" hits in last 10 games it looses, i reset play again , it crashes just before checkmateing DR$ first game then no more reads bases ,it is friend to DR4(JRa4 ) i think,

Disable Total bases , we begin to think the may problem lay with them at least as far as crashes maybe concerned as we have had crash with 9.57b since installing these ******* Total (waste of our time) Bases, it no longer showed table base hits after continuation following crash...its worth a try

Yes , i disabled Totalbases and just used TripleBases ,JRs engine did not crash but this is NOT the same strength as his other compiles ...look at this very short test

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

DR4 had No large pages NOR bases... i shall let test run on to ensure no crashes

As promised i ran on test and no crashes nor "no moves" but it became to painfull to watch

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

perhaps some Intell users will test please

10 mins per game NO robbobases ponder off... no bases or large pages..same books
DR4 Loses 1-7 , 7 draws

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

We had to cut short test (persnal reasons ) but belive this compile is (to date) the strongest of all at least on AMD quads , perhaps some else could try 15 , or 20 min games test overnight ?

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

KLO says this version on a par with or stronger than DR4

We say.. KLOs version on all 4 cores PLUS TripleBases .... DR4 NO Bases NO large pages...

You decide ...

Realy goes to show how good beta999955 code was when engines NEED Bases to beat DR4 without namilov or large pages helping it


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

Hi girls!

How is Angel today?


Thanks for the useful info! May be it will help
to reduce the number of "no move" bugs in my 55mU compilation.
This night I ran 200 games with it against "official" 9.55mb. The result:

1 IvanHoe 55m6U-x64 +22 +35/=135/-23 53.11% 102.5/193
2 IvanHoe 9.55b x64 -22 +23/=135/-35 46.89% 90.5/193

In fact, there was 201 games played. On the game 34 one occurs "sleeping bug" in 55mU.
I.e. the engine has made no move after book moves and loosed due to timeout.
But after this Fritz has stopped the match instead of starting the next game.
I've started the new match of 167 games (due to my arithmetic error the number of games in 2 matches
became 201 instead of planned 200). There were 8 "no move" bugs: 7 of them in my engine and 1 in official.
So my engine is more unstable than official one. But at the same time seems to be more strong.
If one adds 7 points to 9.55mb and 1 point to 55mU my engine will be ahead of official one by 6 point yet
Vlad0

Slankman writes...


Including these ,now I have 90 games so far against Rybka 4 and the final score is 22-9 and 61 draws.
So far I think this is the best score at 10 min.per engine.
It looks that we have very strong engine that can beat Rybka 4.Ivanhoe programmers have fulfilled their task and they should be thanked for their great work.
More testing will help clarify this fact but also there is a need to test it against Houdini.This is what I am going to do next.
I know it takes time to do slightly longer games but that will give us more information.
Happy testing.

Q9550-XP64, Blitz:10' 0



1 IvanHoe 55mU-x64 3185 +7/-3/=20 56.67% 17.0/30
2 Deep Rybka 4 x64 3190 +3/-7/=20 43.33% 13.0/30



~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Congratulations Vlamir (Vlad0 )
Danke Angels sister and "B" for the many hours of testing this version so well before its release and for testing the V57 of Vlads also

Thank you Vlad0 and all s

Janee and i for s77 for full recovery soon ,we miss her and thank her sister and "as sisters " for being there for her and for us

Ivan B 55mPP + bases , no Large Page

DR4 No Bases No LargePages

same books , 512 engine has,h windows64 vista

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

Speaks for itself realy