Hakkapeliitta 2.0 release

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
User avatar
Bloodbane
Posts: 154
Joined: Thu Oct 03, 2013 2:17 pm

Hakkapeliitta 2.0 release

Post by Bloodbane » Thu Dec 25, 2014 12:43 pm

Merry Christmas to everyone. I have a present for you which I got ready just a day before, a(nother) total rewrite of Hakkapeliitta. I expect it to be rated around 2775, give or take 25 elo. It still contains the same highly tactical style that 1.0 had and still has Syzygy tablebase support. The current version should detect whether to use hardware POPCNT automatically but I do not have any machines without it so that feature is untested. I have also added support for Linux though there are bound to be problems with that as that is not my main OS.

I plan on releasing Hakkapeliitta 3.0 around next fall or winter and it should contain some combination of the things on the following list:

Pondering: this got left out of 2.0 due to me having too little time.
Large pages: almost done already, only have to update the TT code.
Multithreading: depends on how masochistic I am feeling.
Major evaluation tuning and improvements: the current eval is so ridiculously bad that this is my number one priority.
Small search improvements: there are some things which I haven't tried yet like singular extension and SEE pruning at low depths.

The link to the repo containing Hakkapeliitta, binaries are in the "bin" folder: https://github.com/mAarnos/Hakkapeliitta

If you have any problems feel free to send me a message or post in this thread.
Functional programming combines the flexibility and power of abstract mathematics with the intuitive clarity of abstract mathematics.
https://github.com/mAarnos

User avatar
vittyvirus
Posts: 641
Joined: Wed Jun 18, 2014 12:30 pm

Re: Hakkapeliitta 2.0 release

Post by vittyvirus » Thu Dec 25, 2014 1:29 pm

Thanks for your interest in Chess Programming.

BTW, instead of
Winglet, ???
write
Winglet, Stef Luijten
in the acknowledgment (on git).
'I would only believe in a god who could dance.' - Friedrich Nietzsche

SzG
Posts: 2447
Joined: Fri Mar 10, 2006 6:20 am
Location: Szentendre, Hungary

Re: Hakkapeliitta 2.0 release

Post by SzG » Thu Dec 25, 2014 2:09 pm

Thanks Mikko. I will test is as soon as I can.

EDIT:

I have encountered a problem. Hakkapeliitta detects popcnt on my computer, which is an Intel Q6600 without popcnt. And, unfortunately, the engine crashes.
Gabor Szots

CCRL testing group

Jamal Bubker
Posts: 260
Joined: Mon May 24, 2010 2:32 pm

Re: Hakkapeliitta 2.0 release

Post by Jamal Bubker » Thu Dec 25, 2014 2:30 pm

Kiitos paljon Mikko !

User avatar
Bloodbane
Posts: 154
Joined: Thu Oct 03, 2013 2:17 pm

Re: Hakkapeliitta 2.0 release

Post by Bloodbane » Thu Dec 25, 2014 3:04 pm

I have fixed the problem (I think) and updated the binaries. I was accidentally checking for MMX and not POPCNT thanks to a off-by-one bug. Sorry for the trouble.
Functional programming combines the flexibility and power of abstract mathematics with the intuitive clarity of abstract mathematics.
https://github.com/mAarnos

SzG
Posts: 2447
Joined: Fri Mar 10, 2006 6:20 am
Location: Szentendre, Hungary

Re: Hakkapeliitta 2.0 release

Post by SzG » Thu Dec 25, 2014 3:19 pm

Bloodbane wrote:I have fixed the problem (I think) and updated the binaries. I was accidentally checking for MMX and not POPCNT thanks to a off-by-one bug. Sorry for the trouble.
It's OK now, thank you.
Gabor Szots

CCRL testing group

Wolfgang
Posts: 313
Joined: Fri May 12, 2006 11:08 pm

Re: Hakkapeliitta 2.0 release

Post by Wolfgang » Thu Dec 25, 2014 6:37 pm

Bloodbane wrote:... I expect it to be rated around 2775, give or take 25 elo...
First of all: Thanks! :-)

Do you really expect +400 to version 1.0 which is rated 2373 at our (CEGT) 40/4 rating list?

http://www.husvankempen.de/nunn/40_4_Ra ... /1038.html
Best
Wolfgang
CEGT-Team

User avatar
Bloodbane
Posts: 154
Joined: Thu Oct 03, 2013 2:17 pm

Re: Hakkapeliitta 2.0 release

Post by Bloodbane » Thu Dec 25, 2014 6:47 pm

Oops, forgot to specify 2775 CCRL. I think on CEGT 40/4 Hakkapeliitta 2.0 is closer to 2620. I really hope you didn't put it against engines with CEGT ratings of 2775 or so because that is going to be an slaughter.
Functional programming combines the flexibility and power of abstract mathematics with the intuitive clarity of abstract mathematics.
https://github.com/mAarnos

Wolfgang
Posts: 313
Joined: Fri May 12, 2006 11:08 pm

Re: Hakkapeliitta 2.0 release

Post by Wolfgang » Thu Dec 25, 2014 6:56 pm

ok, thanks for the info! I'll test it vs. various opponents from 2400 to 2700 (CEGT).

Btw: 2620 would be an impressive improvement too!! :D
Best
Wolfgang
CEGT-Team

User avatar
Graham Banks
Posts: 30733
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Hakkapeliitta 2.0 release

Post by Graham Banks » Thu Dec 25, 2014 7:16 pm

Bloodbane wrote:Merry Christmas to everyone. I have a present for you which I got ready just a day before, a(nother) total rewrite of Hakkapeliitta. I expect it to be rated around 2775, give or take 25 elo. ....
Thanks Mikko. Merry Christmas to you. 8-)
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

Post Reply