Testers for Maverick Leiden?

Steve Maughan · Post by **Steve Maughan** » Sat Jun 20, 2015 12:09 am

I'm preparing Maverick for Leiden in a week's time. If possible I'd like to squeeze some more strength in before the event (not that it's going to make too much difference LOL).

So I'm looking for a couple of people to help me test my latest version.

If have the time and are willing to put the latest version through it's paces drop me a private message and I'll send you a link to download the version I'm planning on using for Leiden. You'll need a modern processor (e.g. Intel i7 or i5). I'm particularly interested in any instability or bugs you may encounter (I think it's stable but who knows). I'm also interested in any insights into obvious weaknesses which can be easily corrected (i.e. low hanging fruit).

There is a new polyglot opening book which I'll be using. Any test or insights on the book would be appreciated.

Thanks - Steve

bob · Post by **bob** » Sat Jun 20, 2015 3:23 am

If you are interested, and if your program will compile/run under linux, I'd be happy to insert you into my gauntlet and play a bunch of games vs Crafty and send you the results...

Ferdy · Post by **Ferdy** » Sat Jun 20, 2015 9:16 am

When you will use it? I don't know the details of Leiden tour?

Henk · Post by **Henk** » Sat Jun 20, 2015 9:34 am

It may easily happen that you when you correct a last minute bug the correction may even perform worse. So always have a fallback version that plays a reasonable game.

Better not make any improvements in the last days.

lucasart · Post by **lucasart** » Sat Jun 20, 2015 10:21 am

Steve Maughan wrote:I'm preparing Maverick for Leiden in a week's time. If possible I'd like to squeeze some more strength in before the event (not that it's going to make too much difference LOL).

So I'm looking for a couple of people to help me test my latest version.

If have the time and are willing to put the latest version through it's paces drop me a private message and I'll send you a link to download the version I'm planning on using for Leiden. You'll need a modern processor (e.g. Intel i7 or i5). I'm particularly interested in any instability or bugs you may encounter (I think it's stable but who knows). I'm also interested in any insights into obvious weaknesses which can be easily corrected (i.e. low hanging fruit).

There is a new polyglot opening book which I'll be using. Any test or insights on the book would be appreciated.

Thanks - Steve

In my experiene, an engine of the level of Maverick can still gain lots of elo by fixing bugs. That's really the low hanging fruit at this level.

FIxing bugs is generally not hard, because they're mostly stupid and trivial. Finding them is difficult. Often, in DiscoCheck, I found bugs purely by chance. The most annoying thing is that it was typically just after I released a version, and the bugfix was a big elo gainer

I see your board code is well tested and has quite a few assert and board integrity checks, which is good. But what about your eval ? I don't see any assert there. Why don't you put lots of assert to pedantically verify things that are implied or assumed, and see if they fire by playing games with a debug compile ? Also, have you submitted your eval to a flip and flop symmetry test ?

Also there are assumptions you need to clarify and enforce (=assert) in the search, I think. Is it allowed to call qsearch_plus() in check ? Is it allowed to call eval when in check ?

Steve Maughan · Post by **Steve Maughan** » Sat Jun 20, 2015 11:52 am

bob wrote:If you are interested, and if your program will compile/run under linux, I'd be happy to insert you into my gauntlet and play a bunch of games vs Crafty and send you the results...

Thanks for the offer Bob but this version doesn't run on Linux and is only UCI.

- Steve

Steve Maughan · Post by **Steve Maughan** » Sat Jun 20, 2015 11:55 am

Ferdy wrote:When you will use it? I don't know the details of Leiden tour?

It's the WCCC:

http://www.grappa.univ-lille3.fr/icga/t ... php?id=287

http://icga.leidenuniv.nl/

Steve

Steve Maughan · Post by **Steve Maughan** » Sat Jun 20, 2015 11:56 am

Henk wrote:It may easily happen that you when you correct a last minute bug the correction may even perform worse. So always have a fallback version that plays a reasonable game.

Better not make any improvements in the last days.

Version 1.0 is my fallback!

- Steve

Steve Maughan · Post by **Steve Maughan** » Sat Jun 20, 2015 11:57 am

Hi Lucas,

You're right. I added a load of asserts in the board manipulation routines but need more in the evaluation. Bug hunting will be my focus for the next couple of days.

Thanks - Steve

Ferdy · Post by **Ferdy** » Sat Jun 20, 2015 12:29 pm

Steve Maughan wrote:
Ferdy wrote:When you will use it? I don't know the details of Leiden tour?
It's the WCCC:

http://www.grappa.univ-lille3.fr/icga/t ... php?id=287

http://icga.leidenuniv.nl/

Steve

Code: Select all

From 2015-06-30 to 2015-07-06, Leiden University, Leiden, The Netherlands.

I assume the TC is 45' + 15s inc.
You can send the link thru PM, I will test it at this TC with ponder on.

My system is i7-2600k, windows 7.

Testers for Maverick Leiden?

Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?

Re: Testers for Maverick Leiden?