Hi all,
A new version of Hannibal is available at:
https://sites.google.com/site/edapostol/hannibal
Primary changes:
-rewrote most of the support framework
-improved hash table handling
-improved SMP, now using C++11 threads
-improved endgame knowledge
-improved search including move selection, prunings, reductions, etc.
-Multi-PV support
-pawn hash and eval cache are per threads now
Hannibal 1.5 release
Moderators: hgm, Rebel, chrisw
-
- Posts: 550
- Joined: Thu Apr 24, 2008 9:31 am
- Location: Belgium
Re: Hannibal 1.5 release
Wow..thanks Sam !
So many new engines these last days..just great.
Kind regards,
JP.
So many new engines these last days..just great.
Kind regards,
JP.
-
- Posts: 3018
- Joined: Thu Mar 09, 2006 11:58 am
- Location: Antalya/Turkey
Re: Hannibal 1.5 release
+1jpqy wrote:Wow..thanks Sam !
So many new engines these last days..just great.
Kind regards,
JP.
-
- Posts: 6808
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: Invasion of TOP Engines!
Hi there,
I have enough to do for the next 6 weeks.
That is fantastic !!
Thanks to the Hannibal team. Waiting of this version.
How big is the improvement the team await?
Best
Frank
I have enough to do for the next 6 weeks.
That is fantastic !!
Thanks to the Hannibal team. Waiting of this version.
How big is the improvement the team await?
Best
Frank
-
- Posts: 1154
- Joined: Fri Jun 23, 2006 5:18 am
Re: Invasion of TOP Engines!
I am hesitant to give a number, though I will say the new version is "noticeably better". I will say in my own testing the longer the time control and the more cores available, the better Hannibal1.5 does compared to Hannibal 1.4.
-Sam
-Sam
Frank Quisinsky wrote:Hi there,
I have enough to do for the next 6 weeks.
That is fantastic !!
Thanks to the Hannibal team. Waiting of this version.
How big is the improvement the team await?
Best
Frank
-
- Posts: 1154
- Joined: Fri Jun 23, 2006 5:18 am
Re: Hannibal 1.5 release
Yes, all my fault. Edsel did a lot of good work improving SMP and reworking our code to give us a much stronger base to develop from in the future, while I spent a good year just focussed on work/family and only helping a little here and there with testing and minor debugging and such. If I was matching his level of effort I am sure we would have released a long time ago. Only after stopping by TCEC in the most recent cycle and getting a little frustrated about Hannibal's play did I really pitch in and help get the current version ready for release.SzG wrote:Thanks, Sam and Edsel. That was long awaited.
-Sam
-
- Posts: 2871
- Joined: Wed Mar 08, 2006 10:09 pm
- Location: Germany
- Full name: Werner Schüle
Re: Hannibal 1.5 release
Many thanks to Sam and Edsel for the new version!BubbaTough wrote:Hi all,
A new version of Hannibal is available at:
https://sites.google.com/site/edapostol/hannibal
Werner
-
- Posts: 41432
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: Hannibal 1.5 release
Thanks Sam and Edsel.BubbaTough wrote:Hi all,
A new version of Hannibal is available at:
https://sites.google.com/site/edapostol/hannibal
Primary changes:
-rewrote most of the support framework
-improved hash table handling
-improved SMP, now using C++11 threads
-improved endgame knowledge
-improved search including move selection, prunings, reductions, etc.
-Multi-PV support
-pawn hash and eval cache are per threads now
gbanksnz at gmail.com
-
- Posts: 15
- Joined: Sun Dec 07, 2014 3:14 am
- Location: Columbus, Ohio
Re: Hannibal 1.5 release
Thank you very much Sam and Edsel for your hard work and for your generous gift to the chess community.
Hannibal was already one of my favorite engines because of its attractive playing style. I keep a list of interesting games that occur in the tournaments I run and Hannibal is very well represented there with many examples. So this new version is a very welcome addition.
I ran a short tournament this morning with Hannibal 1.4b and the result was +4 =17 -1 in favor of Hannibal 1.5. The games were 3' + 2" per move. Computer is i7 running 4 cores. I used a small book and each engine got to play white and black once with each line.
Not much of a sample but for what it's worth, probably indicates about +40 to +50 ELO gain.
Your efforts are much appreciated here! Thank you again!
Hannibal was already one of my favorite engines because of its attractive playing style. I keep a list of interesting games that occur in the tournaments I run and Hannibal is very well represented there with many examples. So this new version is a very welcome addition.
I ran a short tournament this morning with Hannibal 1.4b and the result was +4 =17 -1 in favor of Hannibal 1.5. The games were 3' + 2" per move. Computer is i7 running 4 cores. I used a small book and each engine got to play white and black once with each line.
Not much of a sample but for what it's worth, probably indicates about +40 to +50 ELO gain.
Your efforts are much appreciated here! Thank you again!
-
- Posts: 1154
- Joined: Fri Jun 23, 2006 5:18 am
Re: Hannibal 1.5 release
Glad to hear it David. Hopefully your small sample is representational of our elo progress. Your comments on Hannibal's style are particularly appreciated. Given where our engine falls in the elo pecking order, and the likelihood we will climb to the top of the pack given the amount of time Edsel and I have to dedicate to this hobby, it would be highly demotivating to think that elo is the only goal.
-Sam
-Sam
deefree49 wrote:Thank you very much Sam and Edsel for your hard work and for your generous gift to the chess community.
Hannibal was already one of my favorite engines because of its attractive playing style. I keep a list of interesting games that occur in the tournaments I run and Hannibal is very well represented there with many examples. So this new version is a very welcome addition.
I ran a short tournament this morning with Hannibal 1.4b and the result was +4 =17 -1 in favor of Hannibal 1.5. The games were 3' + 2" per move. Computer is i7 running 4 cores. I used a small book and each engine got to play white and black once with each line.
Not much of a sample but for what it's worth, probably indicates about +40 to +50 ELO gain.
Your efforts are much appreciated here! Thank you again!