Ernest wrote:

-----------------

Hi Vas,

Since I have both Rybka 2.2n2 mp and Rybka 2.2n2 (single), I did some position analysis on my single-core AMD.
PVs were the same, except that the nodes of the 2.2n2 mp numbered exactly 1.7 times the nodes of the 2.2n2 (single)...

Isn't that an unwanted side-effect of your latest knodes, knodes/sec display? Laughing

------------------

Yes, this is a bug in the multi-processor versions of Rybka 2.2n2. The nodes per second calculations have a huge bug.

More specifically, if you are using X cores, the reported nodes are calculated as if you were using (X+1) cores.

This is fixed in Rybka 2.3, and does not affect the single-processor version of Rybka 2.2n2.

Vas