Greetings,
Myrddin 0.84 is now officially released. The engine now plays around 2050 ELO, which is an improvement of about 50 points over the previous version released seven months ago. This is also the first version to support the Gaviota tablebases and be compiled by Jim Ablett.
Many thanks to all who assisted in its development and testing, and especially to Ron Murawski, who generously hosts Myrddin's site at:
http://computer-chess.org/doku.php?id=c ... ddin:index
Enjoy!
jm
Myrddin 0.84 release
Moderators: hgm, Harvey Williamson, bob
Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
-
Graham Banks
- Posts: 30733
- Joined: Sun Feb 26, 2006 9:52 am
- Location: Auckland, NZ
Re: Myrddin 0.84 release
Thanks John. 
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz
Re: Myrddin 0.84 release
Thanks, nice to know that the engine is making a good progress and recent versions certainly seem to be doing better in STS.
Best regards,
Swami
Best regards,
Swami
Re: Myrddin 0.84 release
Hi Swami,
Because I use GradualTest to run my test suites, I can't use the STS Stat program on your website. So I wrote a little program, called GTScan, to scan the EPD output from GradualTest and produce a score. I can send it to you if you are interested.
There may even be some corner cases in which it does not agree with STS Stat, so if you could find the time to do a little bit of comparative testing it would be very much appreciated.
Many thanks,
jm
Because I use GradualTest to run my test suites, I can't use the STS Stat program on your website. So I wrote a little program, called GTScan, to scan the EPD output from GradualTest and produce a score. I can send it to you if you are interested.
There may even be some corner cases in which it does not agree with STS Stat, so if you could find the time to do a little bit of comparative testing it would be very much appreciated.
Many thanks,
jm
Re: Myrddin 0.84 release
Thanks a lot, John! I'd be interested in this program. I'd certainly do test sometimes in the weekend and compare results with output from STS Stat.JVMerlino wrote:Hi Swami,
Because I use GradualTest to run my test suites, I can't use the STS Stat program on your website. So I wrote a little program, called GTScan, to scan the EPD output from GradualTest and produce a score. I can send it to you if you are interested.
There may even be some corner cases in which it does not agree with STS Stat, so if you could find the time to do a little bit of comparative testing it would be very much appreciated.
Many thanks,
jm
If in some case the result don't match, It's likely that there's difference between testing the engine in Arena, and testing it in Gradtest.
You can send it to me at nswami15 at yahoo dot com.
Best Regards,
Swami
Re: Myrddin 0.84 release
A blast from the past. John I wish you good luck on your endeavors.
Myrddin 0.84b release
Myrddin version 0.84b has now been released. Currently it is only available on Jim Ablett's site (http://homepages.tesco.net/henry.ablett/jims.html), but I will also make it available on my official site (http://computer-chess.org/doku.php?id=c ... ddin:index) within a few days.
It fixes an initialization bug that can only occur when trying to play local matches (i.e. not on ICS) using Winboard and the Gaviota tablebases. The bug would cause Winboard to think that Myrddin claimed a draw before the game began. The bug does not occur under any other circumstance or with any other chess GUI.
Otherwise, the code is identical to 0.84. So you don't have to download it if you hadn't been getting the bug.
Many thanks to Olivier Deville for reporting the problem!
jm
It fixes an initialization bug that can only occur when trying to play local matches (i.e. not on ICS) using Winboard and the Gaviota tablebases. The bug would cause Winboard to think that Myrddin claimed a draw before the game began. The bug does not occur under any other circumstance or with any other chess GUI.
Otherwise, the code is identical to 0.84. So you don't have to download it if you hadn't been getting the bug.
Many thanks to Olivier Deville for reporting the problem!
jm