Test II: Vitruvius vs. Komodo- Never Had A Chance

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

kranium
Posts: 2129
Joined: Thu May 29, 2008 10:43 am

Re: Test II: Vitruvius vs. Komodo- Never Had A Chance

Post by kranium »

and if 'yes'....

is this perhaps because you are now beta testing, testing parameter changes, as a part of the 'Vitruvius' team?

if so:
:roll: :shock: :x :oops:
User avatar
Graham Banks
Posts: 44626
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Test II: Vitruvius vs. Komodo- Never Had A Chance

Post by Graham Banks »

kranium wrote:.....major problem: as a CCRL tester (former...not sure here?...apparently this is a 'private' CCRL internal matter, public has no right to be informed!)
(but i do see you are no longer using the CCRL logo as your Avatar, can you explain?)
George no longer tests for CCRL.
gbanksnz at gmail.com
Robert Flesher
Posts: 1287
Joined: Tue Aug 18, 2009 3:06 am

Re: Test II: Vitruvius vs. Komodo- Never Had A Chance

Post by Robert Flesher »

geots wrote:Vitruvius_1.0C_x32 vs. Komodo32 Version 4-c2


It is a bit complicated to explain what we are doing right now by playing an engine and then repeating the match with some changes. Although I am making the setting decisions, it is Roberto's place to explain things- not mine.

So far:

Match 1- vs. Rybka 4.1 w32, Vitruvius scores +14

Match 2- vs. Rybka 4.1 w32 (after changes), Vitruvius scores +42!


Match 1- vs. Komodo 4 32bit, Vitruvius scores +7

Match 2- vs. Komodo 4 32bit (after changes-below), Vitruvius scores +85!




1CPU/32bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games


Vitruvius_1.0c_x32..........+85......+20/-8/=22......62.00%......31.0/50


Komodo32 Version 4-c2....-85......+8/-20/=22......38.00%......19.0/50





Not sure who said this here yesterday, and do not care to waste the time looking. But it was said that Vitruvius was just a weak Ivanhoe. In his dreams.

I looked back at every Ivanhoe version (10 or 12) that I had run ag. Komodo 4 32bit. They all beat Komodo, as did Deep Saros and the new RobboLito 0.10. BUT- not one of them scored even 50% as high as Vitruvius did in this match. And the only engine to score higher against Komodo than Vitruvius did here, was Critter 1.4. And it was only 7 points higher than Vitruvius! Which amounts to only 1 game.

Some people will just have to face it- Vitruvius is here to stay and it is extremely, extremely strong.

Hell, if I did not already have the complete package to beta test, "Normal" and "HEM" files- I would buy it in a second.



Bye for now,

g
e
o
r
g
e


Now you are seeing the results I have been getting. Vitruvius plays amazing chess!
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

Re: Test II: Vitruvius vs. Komodo- Never Had A Chance

Post by geots »

tomgdrums wrote:
geots wrote:
WuShock wrote:Can you say what your settings are ??
Is this the only major change :
Bases=NONE
Ponder_Learning=OFF
I don't see anything about " Learning "
And , while I'm at it , what are the : "HEM" files ??

Since I did pay for it , I wouldn't mind knowing the secrets......

Thanks....... Tom


Look, I will help anyone I can. But you have to realize you have it all wrong. I don't get a penny for helping Roberto. It is his program, not mine. But having said that, I don't understand most of what you are asking. In fact I have no idea. I can tell you that HEM stands for "high end machines" , being the new generation that support popcount and the SSEs. All I know about them is I think they give a speed increase. How much, I do not know. Buyers were allowed to choose between the "Normal executables" and the "HEM executables". But they could not get both. I am not trying to be cute or anything, but I don't understand how you made the decision as to which to buy without knowing what they were. Because if you bought the HEMs, and they were not compatible with your computer, they would crash and not run.


I have no idea where you are going with bases and learning. Learning has been shut off in the gui for ALL engines by me. It has nothing to do with any one engine in particular.

I hope I have been of some help. And yes, I am the one who came up with new settings. They are no secret. But the whole point of Roberto calling for me- he has a monster of an engine that was handicapped by some settings. I am merely straightening things out. First, the worst thing I could do is give you new settings before I am 99.9% sure I have his problem fixed. Secondly, I don't have the authority to do thst anyway. Because I don't own the engine. I have carte blanche to do anything I want with his settings, run ag.any engine of my choice with any time limit I choose- but my freedom stops there. When it is settled very,very soon- all buyers will know.


It is a bit confusing because when I veer off on my own settings- I call it "default". Technically it is not, rather it just shows it is different from his suggested changes. I make them, and default is just a way for you to see they are different. It is possible some may remain default if I decide so.

Be patient. We will get this optimum for everone who bought it. People who bought it should thank their lucky stars they have an author/seller who cares more about his customers than he does money. I am constantly having to fight with him over his being impatient. He is so damn anxious you guys are all satisfied, and I tell him it is no good unless we are sure. He says he knows, but he wants so badly for you all to be happy.

I will tell you to go ahead and just use the actual default settings right now- just load and run. Those are strong enough to beat almost anything out there.



Best,

george
Hey George!



Vitruvius is an excellent analysis partner! (which is my whole reason for purchasing it) Especially the Human version. I really am finding it to be a great partner with Houdini, and Komodo!

That is exactly what the "Human " version is for- analysis. That does not mean it can't beat other engines on that setting- just that mainly it is for analysis. That is another thing I am going to verify also.

If you happened to download the pdf. with the instructions PLEASE IGNORE any settings he mentioned were best for blitz and long games- and MAKE NO CHANGES AT ALL until I can get all this straightened out. And I will. And thank you for your patience.


Best regards,

george
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

Re: Test II: Vitruvius vs. Komodo- Never Had A Chance

Post by geots »

WuShock wrote:I don't understand how you made the decision as to which to buy without knowing what they were. Because if you bought the HEMs, and they were not compatible with your computer, they would crash and not run.

That's why i didn't buy the HEM edition , because I never found an explanation of exactly what files would be included in HEM. I have an old i7 / 940 , and it does do the SEE 4.2 , but doesn't do some of the newer ones.......

I have no idea where you are going with bases and learning

Just something I copied and pasted out of your first post :

(after changes-below)

1CPU/32bit
128MB hash
Bases=NONE
Ponder_Learning=OFF -------** Thought maybe these were changes **
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games

Sorry to bother you


You are not bothering me at all. I don't mind helping you. I realized after reading my thread earlier, that I should not have said "changes below". Because that is what caused you to look at bases, learning and so on, as they were below. When I said below- I was referring to the match results that changed from a 1 game win in the 1st to a 12 game win in the 2nd.

Ponder and Learning = OFF are not engine changes. They do control how the engine plays, but they are made in the gui, not any specific engine. Ponder off means neither engine can think on its opponents time. So that is something both engines must abide by. Same with Learning. It is set in the gui or the engines' parameters so that neither can benefit from learning. Meaning if it plays a certain line, and it is bad, if learning was ON the next time that line showed up it would remember and go a different way.

Bases refer to end game tablebases- that is all. Just remember that the list you copied of mine were the match settings that BOTH ENGINES had to abide by. And remember it is not a setting change for an engine when it is merely a match setting that BOTH ENGINES must adhere to.

And any time you have a question- please ask. That is the only way to learn. I assure you that you are not bothering me.


Best,

george
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

Re: Test II: Vitruvius vs. Komodo- Never Had A Chance

Post by geots »

kranium wrote:
WuShock wrote:I don't understand how you made the decision as to which to buy without knowing what they were. Because if you bought the HEMs, and they were not compatible with your computer, they would crash and not run.

That's why i didn't buy the HEM edition , because I never found an explanation of exactly what files would be included in HEM. I have an old i7 / 940 , and it does do the SEE 4.2 , but doesn't do some of the newer ones.......

I have no idea where you are going with bases and learning

Just something I copied and pasted out of your first post :

(after changes-below)

1CPU/32bit
128MB hash
Bases=NONE
Ponder_Learning=OFF -------** Thought maybe these were changes **
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games

Sorry to bother you

Andres (Chess2u tester) wrote:

4min+2sec
Code:
1 Critter 1.4 64-bit +53 +23/=46/-11 57.50% 46.0/80
2 Vitruvius_1.0C_x64 -53 +11/=46/-23 42.50% 34.0/80

4min+2sec
Code:
1 RobboLito 0.10 SMP 64bit +47 +15/=21/-9 56.67% 25.5/45
2 Vitruvius 1.0 x64 -47 +9/=21/-15 43.33% 19.5/45

4min+2sec
Code:
1 Critter 1.4 64-bit +60 +11/=19/-5 58.57% 20.5/35
2 Vitruvius 1.0 x64 -60 +5/=19/-11 41.43% 14.5/35

http://www.chess2u.com/t5420p45-vitruvius


Immortal223:
1. RobboLito 0.10 SMP x64 129.0/200 85-27-88 (L: m=27 t=0 i=0 a=0) (D: r=64 i=12 f=4 s=0 a=8) (tpm=152.9 d=15.0 nps=2851074)
2. Vitruvius 1.0 Cons. x64 71.0/200 27-85-88 (L: m=85 t=0 i=0 a=0) (D: r=64 i=12 f=4 s=0 a=8) (tpm=130.9 d=14.1 nps=2681924)
Thank you very much for the new and weakest engine from IvanHoe family. Especially for its price!

http://immortalchess.net/forum/showthread.php?t=14200

no thanks!
...even if George has? created magically secret Vitruvius super parameters!
(as implied above)
:lol:


Norman, enough with the silly shit- ok. I already have a bad headache.