FruitFly 1.1

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Richard Allbert
Posts: 777
Joined: Wed Jul 19, 2006 7:58 am

Re: FruitFly 1.1

Post by Richard Allbert » Mon Jul 23, 2012 8:37 am

Well,

1. I'm using my PC to do your test for a few thousand games, to help you understand whether you have a strength increase over Fruit 2.1
2. Fruit isn't getting crushed - it is mid table for the opponents, which is why I chose the opponents.

You really should lighten up a little.

Richard Allbert
Posts: 777
Joined: Wed Jul 19, 2006 7:58 am

Re: FruitFly 1.1

Post by Richard Allbert » Mon Jul 23, 2012 9:26 am

ZirconiumX wrote: I don't program for elo, I program for style. If you want elo, use Toga.
Matthew:out
ZirconiumX wrote: FruitFly 1.1 is about 100 Elo stronger (My calculator shows 200 Elo, but I know the error margin is huge) than Fruit 2.1, scoring 75%.
ok....

User avatar
gleperlier
Posts: 1033
Joined: Sat Feb 04, 2012 9:03 pm

Re: FruitFly 1.1

Post by gleperlier » Mon Jul 23, 2012 11:08 am

ZirconiumX wrote:
Richard Allbert wrote:FF has played it's 6000 games, Fruit 2.1 has played 2000.

So far FF is -30 elo vs Fruit 2.1, test will be complete in 24 hours or so.
I don't program for elo, I program for style. If you want elo, use Toga.

Another thing. Most of these programs are post 2009. Fruit came out in 2004. Did you really expect a Fruit type program not to be crushed to pieces?

Matthew:out
Matthew,

Now that Richard and Sven are helping you, please be kind enough to thanks

:wink:

Cheers,

Gab

User avatar
lucasart
Posts: 3184
Joined: Mon May 31, 2010 11:29 am
Full name: lucasart
Contact:

Re: FruitFly 1.1

Post by lucasart » Mon Jul 23, 2012 1:19 pm

gleperlier wrote:
ZirconiumX wrote:
Richard Allbert wrote:FF has played it's 6000 games, Fruit 2.1 has played 2000.

So far FF is -30 elo vs Fruit 2.1, test will be complete in 24 hours or so.
I don't program for elo, I program for style. If you want elo, use Toga.

Another thing. Most of these programs are post 2009. Fruit came out in 2004. Did you really expect a Fruit type program not to be crushed to pieces?

Matthew:out
Matthew,

Now that Richard and Sven are helping you, please be kind enough to thanks

:wink:

Cheers,

Gab
I admire their patience...

ZirconiumX
Posts: 1327
Joined: Sun Jul 17, 2011 9:14 am

Re: FruitFly 1.1

Post by ZirconiumX » Mon Jul 23, 2012 6:12 pm

lucasart wrote:
gleperlier wrote:
ZirconiumX wrote:
Richard Allbert wrote:FF has played it's 6000 games, Fruit 2.1 has played 2000.

So far FF is -30 elo vs Fruit 2.1, test will be complete in 24 hours or so.
I don't program for elo, I program for style. If you want elo, use Toga.

Another thing. Most of these programs are post 2009. Fruit came out in 2004. Did you really expect a Fruit type program not to be crushed to pieces?

Matthew:out
Matthew,

Now that Richard and Sven are helping you, please be kind enough to thanks

:wink:

Cheers,

Gab
I admire their patience...
Yup Lucas. I am as thick as two planks. Well done for spotting that.

I am a 12 year old. What on all earth did you expect? Even you had to start from somewhere.

Matthew:out
Some believe in the almighty dollar.

I believe in the almighty printf statement.

ZirconiumX
Posts: 1327
Joined: Sun Jul 17, 2011 9:14 am

Re: FruitFly 1.1

Post by ZirconiumX » Mon Jul 23, 2012 6:40 pm

gleperlier wrote:
ZirconiumX wrote:
Richard Allbert wrote:FF has played it's 6000 games, Fruit 2.1 has played 2000.

So far FF is -30 elo vs Fruit 2.1, test will be complete in 24 hours or so.
I don't program for elo, I program for style. If you want elo, use Toga.

Another thing. Most of these programs are post 2009. Fruit came out in 2004. Did you really expect a Fruit type program not to be crushed to pieces?

Matthew:out
Matthew,

Now that Richard and Sven are helping you, please be kind enough to thanks

:wink:

Cheers,

Gab
Thank you, Gab, for suggesting I should thank Richard and Sven.

Thank you, Richard and Sven, for being kind and patient.

No thanks whatsoever to Lucas Braesch, for laughing at me, amongst other things.

Matthew:out
Some believe in the almighty dollar.

I believe in the almighty printf statement.

Richard Allbert
Posts: 777
Joined: Wed Jul 19, 2006 7:58 am

Re: FruitFly 1.1

Post by Richard Allbert » Tue Jul 24, 2012 7:20 am

Results, 5000 games not 6000 as I had to remove the Critter games - something was wrong and it lost all games.

Code: Select all

Name            Elo    +    -    games   score   oppo.   draws 
Fruit2.1        -59     9    9    5000      39%    28       23% 
Fruitfly1.1     -79     9    9    5000     37%     28       21% 

Which puts FF slightly worse than Fruit 2.1

User avatar
Jim Ablett
Posts: 1327
Joined: Fri Jul 14, 2006 5:56 am
Location: London, England
Contact:

Re: FruitFly 1.1

Post by Jim Ablett » Tue Jul 24, 2012 9:39 am

Hi Matthew,
I am a 12 year old. What on all earth did you expect? Even you had to start from somewhere.

First of all let me say that I think you are amazing for a 12 year old. I'm stunned really. Second it doesn't matter if you improve Fruit or not.
The coding experience alone which you are gaining makes it worthwhile and I'm sure you're having fun trying.

In my opinion trying to improve on Fruit which is so perfectly tuned and balanced already is an impossible task (unless you get lucky as was the case with the Toga author.)
I personally would have chosen an engine a such as Diablo to improve. The src code is written very well and it has lots of potential for improvement.

best of luck to you.

Jim.

User avatar
Zlaire
Posts: 62
Joined: Mon Oct 03, 2011 7:40 pm

Re: FruitFly 1.1

Post by Zlaire » Tue Jul 24, 2012 11:08 am

ZirconiumX wrote: Another thing. Most of these programs are post 2009. Fruit came out in 2004. Did you really expect a Fruit type program not to be crushed to pieces?
I think you've misunderstood how ratings work.

Theoretically you can be crushed to pieces and still get a valid rating, based on how crushed you got. If one version gets slightly less crushed than the other version, it means the first version is stronger.

But there's a but too of course, this would show that this version is stronger against that particular set of strong engines.

What Richard did was get a set with bother stronger and weaker engines to get a good mix.

Next point is 10 games of testing is like flipping a coin. A test of that size can show anything and is completely worthless in terms of determining strength. Even a 10-0 score should be taken lightly.

Richard Allbert
Posts: 777
Joined: Wed Jul 19, 2006 7:58 am

Re: FruitFly 1.1

Post by Richard Allbert » Tue Jul 24, 2012 7:53 pm

Hi Jim,

I just had a look at Diablo - it isn't Fruit in terms of strength, but it's pretty well balanced.

It has something which all strong engines have - and that I really struggle to find the secret to - its Evaluation is balanced such that it searches with a consistent low branching factor. It gets deep pretty quickly, and isn't doing much in the way of fancy pruning.

It might also prove to be a tough customer to improve!

Maybe something like TSCP - lots of features to add, speed ups to be gained by adding PieceLists, etc. ?

All the best

Richard

Post Reply