Page 2 of 3

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 7:19 am
by Jimmy Huggins
geots wrote:
Jimmy Huggins wrote:Larry I have around 90kns on my 32 bit machine. But my is around 2.14 Ghz so I don't have the fastest PC.

Mine is around 2995 and my node count is no better than yours.


george

Yeah then that maybe a problem.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 7:39 am
by Werner
Here on my Intel (no SSE) I have:

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq -

Engine: Komodo32 3 32bit (256 MB)
von Don Dailey, Larry Kaufman

18.20 0:34 +0.17 1.d4 Sf6 2.Sc3 d5 3.Lf4 Lf5 4.e3 e6
5.Ld3 Sc6 6.Sf3 Lb4 7.O-O a6 8.h3 O-O
9.a3 Lxc3 10.bxc3 Se4 (16.127.663) 469

Bester Zug: d2-d4 Zeit: 0:47.362 min K/s: 469.066 Knoten: 22.164.220

and

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq -

Engine: Komodo32 Version 4 (256 MB)
von Don Dailey, Larry Kaufman

16.20 0:29 +0.16 1.d4 d5 2.Sf3 Sf6 3.Lf4 Lf5 4.e3 Sc6
5.Ld3 e6 6.O-O Ld6 7.Sc3 O-O 8.Lxf5 exf5
9.Dd3 g6 (4.550.989) 153

17.01 0:37 +0.20 1.d4 d5 2.Sf3 Sf6 3.c4 e6 4.Sc3 Sc6
5.e3 Le7 6.cxd5 exd5 7.a3 Se4 8.Ld3 Sxc3
9.bxc3 Dd6 10.O-O Lg4 (5.798.938) 153

17.20 0:41 +0.20 1.d4 d5 2.Sf3 Sf6 3.c4 e6 4.Sc3 Sc6
5.e3 Le7 6.cxd5 exd5 7.a3 Se4 8.Ld3 Sxc3
9.bxc3 Dd6 10.O-O Lg4 (6.345.822) 153

Bester Zug: d2-d4 Zeit: 0:47.393 min K/s: 153.502 Knoten: 7.228.242


Komodo 3 is much faster

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:01 am
by Graham Banks
Graham Banks wrote:
Graham Banks wrote:
lkaufman wrote:Thanks, but could you tell us what Komodo 3 does on the same hardware? It should be in the 140-150 range unless something is wrong with Komodo 4 32 bit.
Will take a look shortly and let you know.
I can add that in an endgame position later in the same game, Komodo 4 32-bit is consistently up around 220 kns.
This is on a Q6600 quad using 1CPU.
From my i5 running on 1CPU.

Image

Image

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:32 am
by Graham Banks
From Ray:

Image

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:42 am
by rodolfoleoni
Graham Banks wrote:From Ray:

Image
This screenshot seems to show Komodo 4 x64 is about 4 times faster than 32 bit version.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 10:30 am
by Sarciness
Hmm, I will purchase when 32-bit gets up to speed. I know it's not a high priority for Larry and Don, but I don't have a 64-bit OS.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 12:20 pm
by geots
Sarciness wrote:Hmm, I will purchase when 32-bit gets up to speed. I know it's not a high priority for Larry and Don, but I don't have a 64-bit OS.

Ishan, you can go ahead and purchase it according to what I was told. It is as fast as Komodo 3- Don told me that a little while ago. Now don't if you are concerned, but I am going to go ahead and run it for CCRL in 40/4 blitz ag. Houdini. My 64 bit system I have not yet had time to set up yet. I am not promising anything- just what Don told me in an email. It satisfied me- but you be the judge.


Best regards,

george

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 12:25 pm
by geots
Werner wrote:Here on my Intel (no SSE) I have:

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq -

Engine: Komodo32 3 32bit (256 MB)
von Don Dailey, Larry Kaufman

18.20 0:34 +0.17 1.d4 Sf6 2.Sc3 d5 3.Lf4 Lf5 4.e3 e6
5.Ld3 Sc6 6.Sf3 Lb4 7.O-O a6 8.h3 O-O
9.a3 Lxc3 10.bxc3 Se4 (16.127.663) 469

Bester Zug: d2-d4 Zeit: 0:47.362 min K/s: 469.066 Knoten: 22.164.220

and

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq -

Engine: Komodo32 Version 4 (256 MB)
von Don Dailey, Larry Kaufman

16.20 0:29 +0.16 1.d4 d5 2.Sf3 Sf6 3.Lf4 Lf5 4.e3 Sc6
5.Ld3 e6 6.O-O Ld6 7.Sc3 O-O 8.Lxf5 exf5
9.Dd3 g6 (4.550.989) 153

17.01 0:37 +0.20 1.d4 d5 2.Sf3 Sf6 3.c4 e6 4.Sc3 Sc6
5.e3 Le7 6.cxd5 exd5 7.a3 Se4 8.Ld3 Sxc3
9.bxc3 Dd6 10.O-O Lg4 (5.798.938) 153

17.20 0:41 +0.20 1.d4 d5 2.Sf3 Sf6 3.c4 e6 4.Sc3 Sc6
5.e3 Le7 6.cxd5 exd5 7.a3 Se4 8.Ld3 Sxc3
9.bxc3 Dd6 10.O-O Lg4 (6.345.822) 153

Bester Zug: d2-d4 Zeit: 0:47.393 min K/s: 153.502 Knoten: 7.228.242


Komodo 3 is much faster


Werner, I cannot dispute what you show, but in an email not long ago, Don told me not to worry, it was just as fast as Komodo 3- just measuring more accurately. This is not from me- but I trust what Don tells me. (And what the hell do I know anyway)


Best regards,

george/CCRL

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 2:09 pm
by Sarciness
geots wrote:
Sarciness wrote:Hmm, I will purchase when 32-bit gets up to speed. I know it's not a high priority for Larry and Don, but I don't have a 64-bit OS.

Ishan, you can go ahead and purchase it according to what I was told. It is as fast as Komodo 3- Don told me that a little while ago. Now don't if you are concerned, but I am going to go ahead and run it for CCRL in 40/4 blitz ag. Houdini. My 64 bit system I have not yet had time to set up yet. I am not promising anything- just what Don told me in an email. It satisfied me- but you be the judge.


Best regards,

george
Good enough for me. I purchased it!

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 2:27 pm
by Don
Jimmy Huggins wrote:
geots wrote:
Jimmy Huggins wrote:Larry I have around 90kns on my 32 bit machine. But my is around 2.14 Ghz so I don't have the fastest PC.

Mine is around 2995 and my node count is no better than yours.


george

Yeah then that maybe a problem.
As Larry mentioned we are counting nodes differently. More specifically, both versions have always counted a node when a move is made, but Komodo 3 would count nodes that were forward pruned (if the move was actually executed) and Komodo 4 does not. In Komodo 3 I figured that if any work was done (such as attack testing after making the move) it should count as a node. But in Komodo 4 we changed this because the move is technically being pruned (even if we had to check it out a bit first.) Of course if a move is searched or reduced it is counted in both versions.

It's mostly semantics, but we get more stable node counts and this works better for fixed node testing which we sometimes do.

I have several things in mind for getting serious speedups in future versions of Komodo which will bring the speed of Komodo back up to that of other programs, or at least not as far away. I believe Komodo is one of the slower programs in terms of raw search speed as we have put a really heavy focus on having a smart program and little on heavy optimization. The reasoning is the same for delaying the MP stuff too, we want to start with a really strong program before we play the optimization games.