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.
-
Mdannib
- Posts: 75
- Joined: Mon Jul 12, 2010 10:06 pm
Post
by Mdannib » Sat Apr 20, 2013 4:53 pm
The new version is ready....and it is about
20 elo stronger thant the previous.
The change log :
- Large pages support with automatic detection
- "Unlimited" hash size
- Tuned evaluation
- New move ordering
This is the result of last test against the 1.2c
Code: Select all
Games Completed = 4000 of 4000 (Avg game length = 29.577 sec)
Settings = RR/128MB/10000ms+100ms/M 700000cp for 7 moves, D 150 moves/EPD:C:\Users\Matteo D'Annibale\Documents\LittleBlitzer-2.72\Opening.epd(4000)
Time = 17399 sec elapsed, 0 sec remaining
1. BlackMamba_1.4 x64 2119.5/4000 1166-927-1907 (L: m=927 t=0 i=0 a=0) (D: r=1426 i=345 f=90 s=6 a=40) (tpm=229.4 d=13.37 nps=1731666)
2. BlackMamba_1.2c x64 1880.5/4000 927-1166-1907 (L: m=1166 t=0 i=0 a=0) (D: r=1426 i=345 f=90 s=6 a=40) (tpm=231.1 d=13.31 nps=1703709)
This is the link
https://sites.google.com/site/mdannib/
Regards,
/Matteo & Maurizio
-
chetday
- Posts: 97
- Joined: Sun Nov 18, 2012 8:18 pm
Post
by chetday » Sat Apr 20, 2013 8:12 pm
Thanks for the new version. For what it's worth, 1.4 solved the Mate in 22 problem here on the forum significantly faster than the 1.2b version I was using before.
-
Graham Banks
- Posts: 30733
- Joined: Sun Feb 26, 2006 9:52 am
- Location: Auckland, NZ
Post
by Graham Banks » Sat Apr 20, 2013 8:13 pm
Mdannib wrote:The new version is ready....and it is about 20 elo stronger thant the previous.
Thanks.

My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz
-
pohl4711
- Posts: 828
- Joined: Sat Sep 03, 2011 5:25 am
- Location: Berlin, Germany
-
Contact:
Post
by pohl4711 » Sun Apr 21, 2013 7:18 am
Mdannib wrote:The new version is ready....and it is about
20 elo stronger thant the previous.
The change log :
- Large pages support with automatic detection
- "Unlimited" hash size
- Tuned evaluation
- New move ordering
This is the result of last test against the 1.2c
Code: Select all
Games Completed = 4000 of 4000 (Avg game length = 29.577 sec)
Settings = RR/128MB/10000ms+100ms/M 700000cp for 7 moves, D 150 moves/EPD:C:\Users\Matteo D'Annibale\Documents\LittleBlitzer-2.72\Opening.epd(4000)
Time = 17399 sec elapsed, 0 sec remaining
1. BlackMamba_1.4 x64 2119.5/4000 1166-927-1907 (L: m=927 t=0 i=0 a=0) (D: r=1426 i=345 f=90 s=6 a=40) (tpm=229.4 d=13.37 nps=1731666)
2. BlackMamba_1.2c x64 1880.5/4000 927-1166-1907 (L: m=1166 t=0 i=0 a=0) (D: r=1426 i=345 f=90 s=6 a=40) (tpm=231.1 d=13.31 nps=1703709)
This is the link
https://sites.google.com/site/mdannib/
Regards,
/Matteo & Maurizio
Never use the RoundRobin-mode of the LittleBlitzerGUI for your testwork. In the RoundRobin-mode the LBG chooses the opening positions per random out of the opening-positions-file and will not replay the position with reversed colors. Only in the Gauntlet-mode the opening positions are played sequentially from position 1 and replayed with reversed colors.
Best - Stefan
P.S: Testrun of Black Mamba 1.4 for the LS-ratinglist is running. Result in 5-6 days...
-
Mdannib
- Posts: 75
- Joined: Mon Jul 12, 2010 10:06 pm
Post
by Mdannib » Sun Apr 21, 2013 7:48 am
pohl4711 wrote:
Never use the RoundRobin-mode of the LittleBlitzerGUI for your testwork. In the RoundRobin-mode the LBG chooses the opening positions per random out of the opening-positions-file and will not replay the position with reversed colors. Only in the Gauntlet-mode the opening positions are played sequentially from position 1 and replayed with reversed colors.
Best - Stefan
P.S: Testrun of Black Mamba 1.4 for the LS-ratinglist is running. Result in 5-6 days...
Thanks for your suggestion.
/Matteo
-
jshriver
- Posts: 1103
- Joined: Wed Mar 08, 2006 8:41 pm
- Location: Morgantown, WV, USA
-
Contact:
Post
by jshriver » Mon Apr 22, 2013 5:07 am
Very cool, will update my copy for playing.
Thanks for the release!
-Josh
-
Mdannib
- Posts: 75
- Joined: Mon Jul 12, 2010 10:06 pm
Post
by Mdannib » Tue Apr 23, 2013 6:48 pm
Thank you, Jose.
The problem is the code used for the large pages; i will prepare a new compile for oldest pc (only for 32bit) asap but without any test (al my pc support the large pages).
There are some problems even with the Mac version running on Osx 10.5 and 10.6 but i've already got the fix
Regards
/Matteo
-
velmarin
- Posts: 1579
- Joined: Mon Feb 21, 2011 8:48 am
Post
by velmarin » Tue Apr 23, 2013 8:05 pm
You must completely disable large pages to 32 bits,
To my knowledge no 32-bit system supports them.
By the way, I probe BlackMamba with the Blockade pawn activated and had no breakage.

-
Mdannib
- Posts: 75
- Joined: Mon Jul 12, 2010 10:06 pm
Post
by Mdannib » Thu Apr 25, 2013 11:59 am
Some news.....
* Added the file BlackMamba_1.4_Win32_NLP for the systems that doesn't support the large pages and got the error related to KERNEL32.dll
* Recompiled the mac version that now runs on OSX 10.5 or above.
The NLP version is not tested
Regards
/Matteo