Multicore task on Tower 7910

Hi Friends,

My knowledge of computers is pretty limited, so my question might be silly one. Hope someone can help me with this.

I am trying to run some acadamic computation work using package called Quantum Espresso on Precision tower 7910 with 40 CPU and 252 GB RAM, having Linux OS. 

My problem is that the speed of the computation is not satisfactory as compared to what suggested by the computational package. Speed is highest when I use 80 processors, but decreases if I increase the number of processors. Whereas, I see in the manual of package that people use even high numbers of processors at it helps in faster calculations. 

As suggested by package manual, I have installed open-mpi, intel MKL libraries etc. 

Experts on the software package have already given up, would be thankful if anyone can suggest me solution to this problem.

thanks

SG