Difference in Node Speeds With Same Hardware - Fritz 12

Hi,

I have Fritz 12 installed on my system. Recently, I noticed something curious - when Fritz 12 and Houdini 1.5a were given one core each to work with, the node speed of Fritz 12 was approximately 2200kN/s, while that of Houdini 1.5a was nearly 1300-1400kN/s. Here are the images:

Is this OK? Aren't they both supposed to have nearly the same node speed given the same hardware? Oh, and get this: even with a significantly smaller node speed, Houdini 1.5a still manages to beat Fritz 12 in an engine match, like it should. On using two cores, I get about 2500kN/s with Houdini, which is fine. Am I doing something wrong, or is this the way it is? If it is of any help, I have an Intel Pentium E5200 processor clocked at 2.5GHz.

And where exactly am I supposed to post questions on chess software? I couldn't find any section that explicitly mentioned chess software...so here it is...

Thanking you and best regards,

Aashish Satyajith

You cannot compare node speed or search depth of different engines. They are different, after all.

That's what I thought too. But all the other UCI engines like Critter 1.6a and Stockfish 3 nearly get the same node speeds as Houdini 1.5a, and I simply couldn't believe that Fritz 12 was nearly 50% faster on the same hardware. But thanks anyway...

How much RAM is in your computer?

I have a 1GB RAM and I've selected 128MB as Hashtable Size.

There was this one statement that I found intriguing (it was given in the help file): "The native Chessbase engines are linked as DLL’s to the user interface, which means that unlike the UCI engines they are not external processes."

But I think Estragon and duck_and_cover are probably right...it might simply be a matter of design after all.

These numbers also vary with the game, the position and the depth/time settings.  You have apparently compared the same position in your screenshots so the variables left would be a question of the "instantaneous" time at which the screenshot was taken.  Note that the working depths are different and the working time in the lower right corner are different.  The node speed numbers are variable (random sampling at determined intervals) and also dependent on the hash tables used and RAM available after the engine is loaded.

I think you have to be very careful of trying to compare two engines in this way.  In my Fritz 13, there have been numerous updates to the program, some of them apparently involve improvement of the engine efficiency as the self-rating numbers do change.