Fire 3.0 released

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.
pohl4711
Posts: 828
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Fire 3.0 released

Post by pohl4711 » Fri Dec 13, 2013 6:52 am

Good news: Fire 3.0 is there. Bad news: At the moment only modern CPUs are supported, because its a AVX-compile:

Intel Sandy Bridge
Intel Sandy Bridge E
Intel Ivy Bridge
Intel Ivy Bridge E
Intel Haswell
Intel Broadwell
AMD Bulldozer
AMD Piledriver
AMD Jaguar

But more compiles for older CPUs are announced...

Testrun for my LS-ratinglist will start, when the testruns of Stockfish 131210 and Komodo TCECr (bugfixed-version) are finished. If all works correctly the testrun of Fire 3.0 probably will start Tuesday and should end next Friday...

Here the link to the Fire 3.0 download: http://www.chesslogik.com/fire.htm


Stefan

User avatar
beachknight
Posts: 3533
Joined: Tue Jan 09, 2007 7:33 pm
Location: Antalya, Turkey
Contact:

Re: Fire 3.0 released

Post by beachknight » Fri Dec 13, 2013 12:52 pm

pohl4711 wrote:Good news: Fire 3.0 is there. Bad news: At the moment only modern CPUs are supported, because its a AVX-compile:

Intel Sandy Bridge
Intel Sandy Bridge E
Intel Ivy Bridge
Intel Ivy Bridge E
Intel Haswell
Intel Broadwell
AMD Bulldozer
AMD Piledriver
AMD Jaguar

But more compiles for older CPUs are announced...

Testrun for my LS-ratinglist will start, when the testruns of Stockfish 131210 and Komodo TCECr (bugfixed-version) are finished. If all works correctly the testrun of Fire 3.0 probably will start Tuesday and should end next Friday...

Here the link to the Fire 3.0 download: http://www.chesslogik.com/fire.htm


Stefan
SSE42 compile is also available.
hi, merhaba, hallo HT

carldaman
Posts: 1543
Joined: Sat Jun 02, 2012 12:13 am

Re: Fire 3.0 released

Post by carldaman » Fri Dec 13, 2013 4:55 pm

pohl4711 wrote:Good news: Fire 3.0 is there. Bad news: At the moment only modern CPUs are supported, because its a AVX-compile:

Intel Sandy Bridge
Intel Sandy Bridge E
Intel Ivy Bridge
Intel Ivy Bridge E
Intel Haswell
Intel Broadwell
AMD Bulldozer
AMD Piledriver
AMD Jaguar

But more compiles for older CPUs are announced...

Testrun for my LS-ratinglist will start, when the testruns of Stockfish 131210 and Komodo TCECr (bugfixed-version) are finished. If all works correctly the testrun of Fire 3.0 probably will start Tuesday and should end next Friday...

Here the link to the Fire 3.0 download: http://www.chesslogik.com/fire.htm


Stefan
Very interesting, thanks!

Regards,
CL

Werewolf
Posts: 929
Joined: Thu Sep 18, 2008 8:24 pm

Re: Fire 3.0 released

Post by Werewolf » Fri Dec 13, 2013 4:59 pm

is Fire derived from another engine?

carldaman
Posts: 1543
Joined: Sat Jun 02, 2012 12:13 am

Re: Fire 3.0 released

Post by carldaman » Fri Dec 13, 2013 5:04 pm

Werewolf wrote:is Fire derived from another engine?
I believe it's derived from both Ivanhoe and Robbolito.

User avatar
kranium
Posts: 1768
Joined: Thu May 29, 2008 8:43 am

Re: Fire 3.0 released

Post by kranium » Fri Dec 13, 2013 9:25 pm

carldaman wrote:
Werewolf wrote:is Fire derived from another engine?
I believe it's derived from both Ivanhoe and Robbolito.
Yes of course it's a direct Ippolit derivative...it started as Firebird (myself along with Milos S.)
complete history is here: http://www.chesslogik.com/fire.htm

Considering IvanHoe, I believe the problem was that the authors got much too caught up adding features
(monte carlo, robbobases, GUIs, etc., etc.)
That software is incredibly feature rich! They never concentrated on, or worked sufficiently at adding strength IMO
I also spent far too much time adding features to Fire 2.2, (170 configurable parameters, etc.)
because it seems most users judge the success of a release by ELO

Anyway, I've worked on Fire 3.0 privately most of this year since closing the source code, running tests on 2 6 core i7s 24/7 and concentrating on strength only...
I have made some progress with 3.0, but it remains to be seen how these improvements translate (scale) to longer TCs

PS - I strongly advise against anyone trying to develop or improve an existing chess engine unless they can endure endless and incredibly tedious work
recording tests in a spreadsheet, making a small change, compiling, setting up the test, waiting for results, and repeating this process over and over for many months, even years...yuck! (well i guess there's worse things)
Don D. said it best: people who develop chess engines don't have a life

IMO that's why people like Don Daily, Larry Kaufman, R. Houdart, Marco C. and the Stockfish team, deserve real kudos for their exceptional dedication and effort in this regard!

Ok my ramblings aside:

I just posted 4 more compiles
Fire 3.0 GEN
these are (generic) binaries should work on most systems that don't support AVX or SSE42
http://chesslogik.wix.com/fire

User avatar
velmarin
Posts: 1579
Joined: Mon Feb 21, 2011 8:48 am

Re: Fire 3.0 released

Post by velmarin » Fri Dec 13, 2013 9:56 pm

This last relase GEN, works fine on my Phenom II

Thanks Norman, and good luck and satisfaction.

greetings, Jose.

Max
Posts: 171
Joined: Tue Apr 13, 2010 8:41 am

Re: Fire 3.0 released

Post by Max » Sat Dec 14, 2013 9:52 am

kranium wrote:I just posted 4 more compiles
Fire 3.0 GEN
these are (generic) binaries should work on most systems that don't support AVX or SSE42
http://chesslogik.wix.com/fire
Hi Norman,
all 32-bit versions crash with this error while loading on my Mac via wine 1.7.7. Likely the same with wine on Linux.

Code: Select all

Unhandled exception: unimplemented function KERNEL32.dll.GetLargePageMinimum called in 32-bit code (0x7bc44528).
Register dump:
 CS:001b SS:0023 DS:0023 ES:0023 FS:1007 GS:000f
 EIP:7bc44528 ESP:0033fd5c EBP:0033fdc0 EFLAGS:00000202(   - --  I   - - - )
 EAX:00452a24 EBX:0033fd68 ECX:ffffffff EDX:00000000
 ESI:00000000 EDI:0043cad9
Stack dump:
0x0033fd5c:  00000000 00761308 004b9f8c 80000100
0x0033fd6c:  00000001 00000000 7bc44528 00000002
0x0033fd7c:  00452b56 00452a24 00000000 00000000
0x0033fd8c:  7b8665fc 0033fdb0 0000000f 0033fdc8
0x0033fd9c:  7b86549b 00000000 00000000 00000000
0x0033fdac:  7b86549b 00761308 00000000 00000000
0200: sel=1007 base=7ffc0000 limit=00000fff 32-bit rw-
Backtrace:
=>0 0x7bc44528 (0x0033fdc0)
  1 0x0034000f (0x0033fe68)
  2 0x0043ca83 in fire 3.0 w32 (+0x3ca82) (0x0033feb0)
  3 0x7b850e2c in kernel32 (+0x40e2b) (0x0033fec8)
  4 0x7b854687 in kernel32 (+0x44686) (0x0033ff08)
  5 0x7bc6d47c (0x0033ff28)
  6 0x7bc6e52a (0x0033ffa8)
  7 0x7bc6d442 (0x0033ffc8)
  8 0x7bc44bce (0x0033ffe8)
Would it be possible to create 32-bit versions of Fire 3.0 without (the not by wine supported) large pages?

Thanks,
Max

User avatar
kranium
Posts: 1768
Joined: Thu May 29, 2008 8:43 am

Re: Fire 3.0 released

Post by kranium » Sat Dec 14, 2013 11:58 am

Hi Max-
yes, NP
I could do that...

I was actually thinking last night of doing a native Linux compile
(got a couple email requests for such)

Give me some time and I'll do one or the other..

Best Regards-
Norm

CositasBuenas
Posts: 85
Joined: Tue Aug 03, 2010 5:36 pm

Re: Fire 3.0 released

Post by CositasBuenas » Sat Dec 14, 2013 12:31 pm

Pretty please a native linux compile!

Post Reply