Page 1 of 1

Scorpio 2.5 egbb problems

PostPosted: 28 Mar 2010, 18:23
by Denis P. Mendoza
I hope Daniel is still visiting the forum as we have problems with the new Scorpio in utilizing the bitbases.
I've tried 3-4 men and 3-4-5 men on 32/64-bit versions and results show that Scorpio is better without the egbb! Odd!
attached are the Arena endgame epd test results.

http://www.mediafire.com/download.php?3ji1dz2njjt

The engines were verified loading the egbbprobe prior to tests.

Others report that this new version ignores the egbb path in the ini file with :

#path to egbb
egbb_path C:\egbbnew(64-bit)\

feature done=0
ht 2097152 X 64 = 128MB
eht 2097152 X 16 = 32MB
pht 699050 X 24 = 15MB
processors [1]
EgbbProbe not Loaded!
loading_time = 0s


Has anyone noticed it too?

Thanks.

Denis

Re: Scorpio 2.5 egbb problems

PostPosted: 29 Mar 2010, 15:37
by Volker Pittlik
Denis P. Mendoza wrote:...
#path to egbb
egbb_path C:\egbbnew(64-bit)\

...
EgbbProbe not Loaded!
loading_time = 0s

Has anyone noticed it too?
....


Same here on Linux 64-bit since several versions.


vp

Re: Scorpio 2.5 egbb problems

PostPosted: 29 Mar 2010, 21:17
by Werner Schüle
Hi,
just tested here in console mode:

#path to egbb
egbb_path J:/negbb64/

egbb_load_type 1

feature done=0
ht 4194304 X 64 = 256MB
eht 524288 X 16 = 8MB
pht 349525 X 24 = 7MB
processors [1]
EgbbProbe 3.3 by Daniel Shawul
Egbbs loaded !
loading_time = 2s

it would work with 5men too, but I normally do not use as loading time may hurt the engine in matches!

feature done=0
ht 4194304 X 64 = 256MB
eht 524288 X 16 = 8MB
pht 349525 X 24 = 7MB
processors [1]
EgbbProbe 3.3 by Daniel Shawul
Egbbs loaded !
loading_time = 120s

kann ich diese 75 Endspiele haben (wo runterladen)?
Werner

edit: just used Eigenmann endgametest too with 10 sec/postion: 31 solved with 5men and 36 solved without egbbs :?:

Re: Scorpio 2.5 egbb problems

PostPosted: 30 Mar 2010, 13:36
by Werner Schüle
Correction:
used Scorpio 2.5 x64 1CPU and new egbbs

just used Eigenmann endgametest too with 10 sec/postion: 31 solved with 5men and 36 solved without egbbs :?:

but: now the same with 60 seconds as in original test:
Solved 47 without egbbs and solved 56 with egbbs

So they work fine here!!
from Eigenmann test: 028. Scorpio 1.91 43/100 01:07:35
I used a AMD 64x2 2600+ running on 2.6 GHz
Werner

Re: Scorpio 2.5 egbb problems

PostPosted: 02 Apr 2010, 00:40
by Denis P. Mendoza
Thanks Volker and Werner. I also avoided 3-4-5 men on my Scorpio gauntlets as loading time really hurts. I only asked this because of Shaun Brewer's experience with the new Scorpio. I didn't notice this strange behavior, but I'll recheck again and see what happens. I'll probably compile the source again and compare results.

Denis