Chessmaster 11 speed tests

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

Moderators: hgm, Rebel, chrisw

User avatar
Werner
Posts: 2872
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Chessmaster 11 speed tests

Post by Werner »

Hi all,
I managed to install it on my dual and on my quad. Here some interesting
things:
- single CM11 engine is running in Arena, Chessbase,... with OPK
- not managed to use more CPUs in other GUIs
- 3 engines are delivered: TheKing333, TheKing350, TheKing350_64

The CM GUI is loading (I have not tested in 32bit GUI yet, only 64bit Win):
low ressouce engine: 333 (32bit)
medium ressource engine: 350 (32bit)
high ressource engine: 350_64 (64bit)

Some speed Tests:
on dual engine, running start position
350.exe using only 1CPU
0:34 - 2/13 - 19.564.370 kn = 575kns
3:00 - 3/15 - 104.639.314 kn = 581 kns

350_64.exe using 2CPUs
0:19 - 2/13 - 20.558.800 kn = 1082 kns
1:39 - 3/15 - 110. 675.234 kn = 1118 kns

now on my quad, which is a little faster:
350.exe using only 1 CPU
2:42 - 3/15 - 104.639.314 kn = 646kns

350_64.exe using 3CPUs (75%), one is used by game.exe (25%)
1:16 - 3/15 - 140.369.223 kn = 1847 kns

so I do not know how to use 2 CPUs on my quad :-)))
and I do not know why TheKing is not using all 4 CPUs - never saw an engine
using 3CPUs :-))
Werner
User avatar
mclane
Posts: 18755
Joined: Thu Mar 09, 2006 6:40 pm
Location: US of Europe, germany
Full name: Thorsten Czub

Re: Chessmaster 11 speed tests

Post by mclane »

johan said i think in leiden that he never tested it on more than 2 cpu's.
did you found a way to test it on the dual core with PB off so that you can let it run 2 cpu versus another MP program using 2 cores and PB off?

or is this also not possible ?!
User avatar
Werner
Posts: 2872
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: Chessmaster 11 speed tests

Post by Werner »

mclane wrote:johan said i think in leiden that he never tested it on more than 2 cpu's.
did you found a way to test it on the dual core with PB off so that you can let it run 2 cpu versus another MP program using 2 cores and PB off?

or is this also not possible ?!
Yes, on dual core I played some test games against our Spike MP version installed as wb engine without an adapter - that works with ponder off. But I do not have a neutral short book for my tests yet.

On my quad I played against scorpio 2.0 4CPU eng-eng matches with ponder off: Scorpio used all 4 CPUs - the Chessmaster used 3 CPUs on its turn.
Werner
Kaj Soderberg

Re: Chessmaster 11 speed tests

Post by Kaj Soderberg »

Oh well, the good thing about using 3 cores is that you keep 1 core to do other stuff on your PC :wink:

But seriously, Johan de K. should have a look at a few things:
- use of more cores, to some maximum of course
- selection by the user of how many cores to use
- access and processing of EGDB's
ernest
Posts: 2041
Joined: Wed Mar 08, 2006 8:30 pm

Re: Chessmaster 11 speed tests

Post by ernest »

Werner wrote:Some speed Tests:
on dual engine, running start position
350.exe using only 1CPU
0:34 - 2/13 - 19.564.370 kn = 575kns
3:00 - 3/15 - 104.639.314 kn = 581 kns

350_64.exe using 2CPUs
0:19 - 2/13 - 20.558.800 kn = 1082 kns
1:39 - 3/15 - 110. 675.234 kn = 1118 kns
Why didn't you also test 350_64.exe (64-bit engine, I suppose) using only 1CPU, in order to check if 32-bit => 64-bit brings something?
User avatar
Werner
Posts: 2872
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: Chessmaster 11 speed tests

Post by Werner »

Hi,
this does not work inside Chessmaster GUI as the engine set is automatic.
So I tried to do it inside Chessbase GUI with wb2uci adapter and opk number. The result is not ok, as I got the same numbers with either engine loaded (checked with Taskmanager):
2:39 T15 101MNs = 635 kns.
Perhaps a restart of the PC would help to see the real difference. Yesterday I had a 5% advantage of the 64bit engine.
The same results under Arena. Here on one line 32 bit engine was faster with 649 kns. Maybe some hash problems.
Werner