A First Glance At the New Vitruvius Engine!!

[geots]

This version will be released to all previous buyers very, very soon- at no cost of course. As well to any others interested in a very good buy!

It is also named 1.01, so I would advise anyone running the previous 1.01 either unload it or add a notation to the old zip so you don't get them mixed up. Below I have for you its first match on my end.

Note the default settings are now for blitz and up to 15 min. and maybe even 20 min. games. So only one change to make for the long games.



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

Code: Select all

1   Strelka 5.1             +42      +12/-6/=32    56.00%    28.0/50
2   Vitruvius_1.01C_x32     -42      +6/-12/=32    44.00%    22.0/50

So in the last 4 Strelka matches ag. the strongest Ivanhoe versions, only showing the wins and losses in the 50-game matches:

Strelka 5.1..........+13.......+56
Ivanhoe B46e......+5.........-56

Strelka 5.1..........+25.......+123
Ivanhoe B47d......+8.........-123

Strelka 5.1..........+14.......+70
Ivanhoe 46h........+4.........-70

Strelka 5.1.............+19......+85
Ivanhoe B47f0.2....+7........-85

Strelka 5.1.............+12.......+42
Vitruvius 1.01C.......+6.........-42



Sometimes a loss is really a win. Only Ivanhoe B46e came close to turning in a score anywhere near as good as the new Vitruvius. It bested all 4 of the top Ivanhoes!

Statistical noise- ok. But I had much rather be starting on the plus end of the noise.

I am extremely proud of Roberto and his efforts.


Off to rest-

g
e
o
r
g
e

[geots]

Too late to edit my post, so I will add here that in the specs where I have each engine using 128MB hash- it was changed to 64MB at the last minute and I am sorry I failed to change it in my post.

Best,

george

[TimoK]

Hi George,

why post results from Strelka against some Ivanhoes and Vitruvius? Sorry, but if you want to prove that Vitruvius is stronger than Ivanhoe then that is the wrong way I think. Please post results of Vitruvius against several opponents, that is the only way to estimate its strength.

Best regards
Timo

[MM]

Hi George,

Please post results of Vitruvius against several opponents, that is the only way to estimate its strength.

Best regards
Timo

+1

[Carotino]

Has just finished the 'Silver Test', played between Vitruvius and Houdini 2.0-Pro.
Of course Vitruvius he lost ( it's the end of all the engines at this time! ), but it proved a good competitor. Vitruvius also took some brilliant victories. Despite everything, I'm pretty satisfied with the result.

Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz 1596 MHz W=21.0 plies; 2. 998kN/s B=23.4 plies; 2.729kN/s

Andrea-HP, Blitz 5m
Rome, 2012.03.10 - 2012.03.11
Score 12345678901234567890123456789012345678901234567890
----------------------------------------------------------------------------------------------------------------------------------------------
1: Houdini 2.0c Pro x64 Aq 60.5 / 100 ===1011010111==1=111=0=11=100==0===1101===11=1===0==1=1===0======0=01110===1111===11111010======1=10
2: Vitruvius_1.01C_HEM_x64 39.5 / 100 ===0100101000==0=000=1=00=011==1===0010===00=0===1==0=0===1======1=10001===0000===00000101======0=01
----------------------------------------------------------------------------------------------------------------------------------------------
100 games: ELO -74

PGN: http://www.mediafire.com/?af4m577qm9q9mzo

best,
Roberto.

[MM]

Has just finished the 'Silver Test', played between Vitruvius and Houdini 2.0-Pro.
Of course Vitruvius he lost ( it's the end of all the engines at this time! ), but it proved a good competitor. Vitruvius also took some brilliant victories. Despite everything, I'm pretty satisfied with the result.

Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz 1596 MHz W=21.0 plies; 2. 998kN/s B=23.4 plies; 2.729kN/s

Andrea-HP, Blitz 5m
Rome, 2012.03.10 - 2012.03.11
Score 12345678901234567890123456789012345678901234567890
----------------------------------------------------------------------------------------------------------------------------------------------
1: Houdini 2.0c Pro x64 Aq 60.5 / 100 ===1011010111==1=111=0=11=100==0===1101===11=1===0==1=1===0======0=01110===1111===11111010======1=10
2: Vitruvius_1.01C_HEM_x64 39.5 / 100 ===0100101000==0=000=1=00=011==1===0010===00=0===1==0=0===1======1=10001===0000===00000101======0=01
----------------------------------------------------------------------------------------------------------------------------------------------
100 games: ELO -74

PGN: http://www.mediafire.com/?af4m577qm9q9mzo

best,
Roberto.

Thanks for sharing and informations

Best Regards

[TimoK]

Hi Roberto,

thanks for sharing the information and games. If you have more results (e.g. against other engines) please share them also with us. I think there are some people who are really interested in Vitruvius (including me) and are thinking about buying. But more information is needed, e.g. when do you release version 1.01?

Best regards
Timo

[Carotino]

Hi Timo,

The release of version 1.01 is now a matter of days. Before issuing the update, I wanted to make a series of tests to eliminate any bugs. I have to thank George for the enthusiasm and the great experience that he made ​​available. Thanks unto his counsels I was able to improve engine performance. Vitruvius is still a 'young engine' and many things can be improved. There is still much work to do, but I think the way we started, is the right one...

Here is a link of other test-games (by my friend Jean Paul!):

http://www.mediafire.com/?d7azsdh5j3x2jdb

best,
Roberto.

[gleperlier]

You can also see my "Vitrivius 1.01C" playing sometimes on Playchess under my account "Numerobis".

Best wishes,

Gabriel

[geots]

Hi George,

why post results from Strelka against some Ivanhoes and Vitruvius? Sorry, but if you want to prove that Vitruvius is stronger than Ivanhoe then that is the wrong way I think. Please post results of Vitruvius against several opponents, that is the only way to estimate its strength.

Best regards
Timo

Timo my friend, I appreciate your input. I will try to explain this, which I normally do not do when someone tries to explain to me the proper way to test an engine. I had just finished and posted the 4 Ivanhoe matches against Strelka. That had been done, and they are posted here. I only decided to include the new Vitruvius as an after thought after Roberto sent it to me. I was looking for something other than Critter or Houdini to stop Strelka. It made perfect sense to try the new engine.

This result was 1 50-game match. I said nothing about Vitruvius being stronger than these four Ivanhoes, and I was not attempting to prove anything one way or the other. It performed better here in this case than the Ivanhoes. Can it beat them itself. Haven't even thought about it- and that is not important right now to me. As I said, this was not a huge plan hatched by me. The Ivanhoe matches were already in the book. If you think I was trying to prove it was stronger than Ivanhoe THIS WAY, you know very little about me.

I welcome constructive criticism, but your comments do border on an insult to my abilities. If I had set out to prove Vitruvius was stronger than Ivanhoe this way, I would have to be a complete idiot- or have many more games to toss into the ring.

As for results against other engines, I have one from a match earlier today against Chiron 1.1a. It will come a little later today. Then there will be Stockfish, Rybka, Komodo. Then I will switch to 64bit. I think you will get all the matches you could ever want to see.

I take no offense, because I know you meant no harm and was trying to be helpful. But please, give me at least a little credit for having a modicum of common sense.

And I do hope you enjoy the engine.


Best regards,

george