In support of the IvanHoe authors

[Izak Pretorius]

Jose i remember in a previous TCEC tournament it was the same.Some criticizing everything.Why didn't you use this compile,why didn't you use that compile,why didn't you use the official ivanhoe compile,as if there was an official compile etc etc... the list goes on.

I think it is purely based on jealousy and the people had some personal interest,perhaps their own compiles or modifications and was/is sour grapes for them not being used in the tournament.

Quite frankly i don't care about all this silly politics,i wish only to see good and strong and stable chess and enjoy chess.

I know a lot of hard work went into Bouquet and i appreciate the efforts you made with your engine and it is hurtful to always have this negativity from other jealous people.

I wish this will stop.

Happy Chess
Izak

[velmarin]

Right here, in this thread,

forget that we are people working for chess, like them.

We are called "crooks". As a "almost always" with impunity.

As always pointing to the party
  "LucasArts" and "Kranium" in their eagerness to be the father of all that is wagging.

These two insufferable I campaigned on "motor origins" against Bouquet , but failed to stop the project.
but not shut his mouth,
always to the party.

[kranium]

There seems to be some misunderstandings here regarding IvanHoe and which is the strongest etc... as an expert on the field let me explain to those who may be ignorant or just simply don't know better.

Some person here claimed that Bouquet and PanChess is just as strong as the strongest IvanHoe.Okay lets look at that for a moment shall we..

On the Stefan Pohl rating list we have as follows :

15 Ivanhoe 50kQ x64s 3037 4 4 19000 52% 3021 57% (cp)
16 PanChess 00.537 x64 3037 5 5 11000 47% 3059 57%
17 Bouquet 1.7 beta x64 3037 5 5 13000 47% 3057 54%
18 Ivanhoe 46eQi x64s 3035 4 4 31000 50% 3034 57%
19 PanChess 00.400 x64 3030 4 4 17000 48% 3046 57%
20 Firenzina 2.3.1 x64s 3030 4 4 19000 48% 3047 55%
21 Fire trap120925 x64s 3029 5 5 11000 48% 3041 59%
22 Robbolito 0.21Q x64s 3028 3 3 44000 50% 3027 58%
23 Firenzina 2.2.2 x64s 3022 4 4 16000 49% 3030 54%
24 Bouquet 1.6 x64s 3022 3 3 35000 50% 3022 57%
25 Stockfish 3 x64s 3020 5 5 12000 46% 3053 45%
26 Ivanhoe 46h x64 3018 4 4 18000 48% 3033 54%
27 Gull 2.1 x64 3018 5 5 10000 45% 3053 55%
28 Bouquet 1.5 x64s 3016 5 5 15000 47% 3036 57%
29 Robbolito 0.10 x64s 3016 4 4 15000 48% 3030 55%

IvanHoe 50kQ x64 is not just a compile,but a modification by a user who claims this is just a compile.
Secondly IvanHoe 50kQ has some stability issues.

are you confusing (cp) with (sp)?

50kQ has 19000 Lightspeed games under it's belt
if there was a stability issue, I'm sure Stefan would have reported it

(cp) refers to compatibility...i.e. does not run in Fritz, etc.

[Izak Pretorius]

are you confusing (cp) with (sp)?

50kQ has 19000 Lightspeed games under it's belt
if there was a stability issue, I'm sure Stefan would have reported it

(cp) refers to compatibility...i.e. does not run in Fritz, etc.

No not really.
I was referring to my own experiences with testing this compile IvanHoe 50kQ.
I see on Stefan's site it says that it does not work on AMD processors.
compatibility problems:
"
Ivanhoe 50kQ not running on AMD-CPUs and not running with the ShredderGUI & FritzGUI (Arena works!)"

And to top it off,i have 2 intel pc's,and it also have technical issues on both of them,on the one it will not execute at all even.

If you want to call it compatibility,or stability,i do not care,it is not a fit for use modification of Ivanhoe,because it certainly isn't a compile of Ivanhoe version 50k.It even beats your RobboLito which is also weaker than "IvanHoe" 50kQ by the way.

But stay tuned,end of next week my next release of PanChess will be out and then "IvanHoe 50kQ" will be old news again anyway.

And there are 2 engines just as strong without those issues,PanChess,which is my modification of IvanHoe 46h,
and the very strong engine Bouquet by Jose mº velasco,which is also in that rating class.He assured me it is not a modification of Ivanhoe,i assumed this because of your accusations etc... and a misunderstanding from his post earlier.

I think Norman,you should stop flaming and rather just relax.

Best Regards
Izak

[kranium]

Izak-
i have made no accusations, not have i flamed anyone...please reread my posts and i think you find my manner respectful and polite

the point of this topic was to express my support for the Ippolit authors
...they are the original authors of IvanHoe and all of it's many derivatives
(including robbolito, panchess, bouquet, mars, firenzina, etc.)

the fact is:

in the Lightspeed rating list,
the Ippo derivatives (as well as the recompiles w/ new and faster Intel optimizations) are bunched extremely close together...
with no derivative clearly distinguishing itself

and there at least 2 compiles of IvanHoe equal to anything else

i believe if a match/tournament organizer decides to chose one Ippo engine,
(not counting Houdini which has evolved enough IMO to be considered a different engine)...

and no derivative is significantly stronger...
the proper thing to do would be to chose the 'original' engine,
and by doing so, give the engine it's proper place, and the authors some credit

i realize you and Jose don't like this, but it's my opinion, as is apparently shared by others
Norm

[Izak Pretorius]

Izak-
i have made no accusations, not have i flamed anyone...please reread my posts and i think you find my manner respectful and polite

the point of this topic was to express my support for the Ippolit authors
...they are the original authors of IvanHoe and all other derivatives
(including robbolito, panchess, bouquet, mars, firenzina, etc.)

the fact is:
in the Lightspeed rating list,
the Ippo derivatives are bunched extremely close together...with no derivative distinguishing itself
and there at least 2 compiles of IvanHoe equal to anything else

i believe if a match/tournament organizer decides to chose 1 Ippo engine
(not counting Houdini which has evolved enough IMO to be considered a different engine)...

and no derivative is significantly stronger...
the proper thing to do would be to chose the 'original' engine,
and by doing so, give the authors proper credit

i realize you and Jose don't like this, but it's my opinion, as is apparently shared by others

Norman

Okay i accept your explanation of non-flaming:)
So lets move on then to your desire,it is your opinion and you are welcome to it.

The problem is,there is no "official" compile since Roberto Pescatore didn't compile any of their sources.
You know well as i do that it was me,and KLO and few others that since start of Ivanhoe were working hard and compiling their changes at regular times.
Just so that there is NO misunderstanding,the Ivanhoe 46h is a direct compilation from the ippolit wiki sources.In this source,i did NOT change any values,or anything else that could change the engines thinking process in any shape or form.
Unfortunately Martin has brought to my attention that this compile does not even work on his system.
He tried the KLO compile from,the same source,if KLO didn't alter anything else,but this compile gives him bogus mate scores etc etc.

So you would rather let Martin use a buggy more than 1 year old compile of Ivanhoe sources so that it can be pure ?!

You can see that in PanChess i clearly state the original programmers as the authors,and me as the modifiaction maker,and Peter Pankovsky as the compiler.

I do not see a problem here.

But obviously you are not happy,i do not know why though.

Best Regards
Izak

[kranium]

as you well know, Intel publishes new versions of it's compilers on a regular basis
normally these releases include new faster optimizations, updated intrinsic functions, etc..

there's no doubt in my mind that a new IvanHoe compile would greatly benefit from this new 'technology'...
(faster popcnt routines, prefetch improvements, etc)

but there's really no need to do it: Q has already done that for us with 46e:

15 Ivanhoe 50kQ x64s 3037 4 4 19000 52% 3021 57% (cp)
16 PanChess 00.537 x64 3037 5 5 11000 47% 3059 57%
17 Bouquet 1.7 beta x64 3037 5 5 13000 47% 3057 54%
18 Ivanhoe 46eQi x64s 3035 4 4 31000 50% 3034 57%
19 PanChess 00.400 x64 3030 4 4 17000 48% 3046 57%
20 Firenzina 2.3.1 x64s 3030 4 4 19000 48% 3047 55%

Qi = Quocvuong Intel

[Izak Pretorius]

as you well know, Intel publishes new versions of it's compilers on a regular basis
normally these releases include new faster optimizations, updated intrinsic functions, etc..

there's no doubt in my mind that a new IvanHoe compile would greatly benefit from this new 'technology'...
(faster popcnt routines, prefetch improvements, etc)

but there's really no need to do it: Q has already done that for us with 46e:

15 Ivanhoe 50kQ x64s 3037 4 4 19000 52% 3021 57% (cp)
16 PanChess 00.537 x64 3037 5 5 11000 47% 3059 57%
17 Bouquet 1.7 beta x64 3037 5 5 13000 47% 3057 54%
18 Ivanhoe 46eQi x64s 3035 4 4 31000 50% 3034 57%
19 PanChess 00.400 x64 3030 4 4 17000 48% 3046 57%
20 Firenzina 2.3.1 x64s 3030 4 4 19000 48% 3047 55%

Qi = Quocvuong Intel

It's 3 elo weaker than PanChess and the strong Bouquet engine.
Next week i publish the next PanChess.
I see no point really.

Which brings as back to the Q compiles,it is not pure.
It is not just compiler optimizations.
And this can be the case for any compile.
anyone can say it is pure,but Q compiles are never accompanied with sources,and it does not give the same corresponding output than the original 46e,could you care to explain that perhaps?

KLO compiles have sources,but it is the non altered exact ippolit wiki sources,so they are not pure either.In other words,he did not publish his alteratiosn.

On occasions i have released my sources,and if i am not mistaken i have released the source for Ivanhoe 46h also at some stage,it is not pure either,but it plays and thinks exactly like the original sources cause the only changes made was those to make it work properly

There are no "pure" compiles of IvanHoe.

I can right now with the latest Intel compiler recompile Ivanhoe 46h and it will NOT be 19 elo stronger.
Perhaps 5 or maybe 10 at the most at these fast time controls.

PanChess is a modification,not a re-compile,with code additions to the search and evaluation and more...

I have to get back to work on PanChess.
I would love to chat some more,but i have elo's to add.

best regards
Izak

[kranium]

It's 3 elo weaker than PanChess and the strong Bouquet engine.

there's actually 2 ELO difference, but regardless if 2 or 3, considering the error margins it essentially means they are equal

anyone can say it is pure,but Q compiles are never accompanied with sources,and it does not give the same corresponding output than the original 46e,could you care to explain that perhaps?

because it's much faster


anyway, best of luck with your project!
Norm

[velmarin]

Within what Ippolit family,
leaving out Houdini, Critter, Equinox, Blackmamba.

Gull, but this is a special case with a tremendous merit.

In the rest in my opinion, what remains is the following:

IvanHoe45, born and died, no work ...
Firenzina equal ....

Robbolite, abandoned, only to growl ...
Fire, equal ...

Mars has not yet executable for 64 no_popcnt or 32-bit,
Fire quick compilation, I have doubts about his future ...

PanChess, Izak hard work, with much enthusiasm,
live project, ...

Bouquet, equal to Panchess, daily work, daily tests, few means to make progress, hopefully, but tired of insults and other things.
Soon I will leave and go into other things, worth might not keep fighting.

After I post similarity test...