A question about scalability on Nehalem architecture

A question about scalability on Nehalem architecture


I was reading a paper about FC (flat combining) , and i came accross
some graphs that shows the scalability on the Nehalem architecture:

Here is the paper and the graphs in it:


As you will notice the Nehalem Intel architecture is giving a negative
scalability for all the Queues, but on the SPARC architecture there is
positive scalability.

So my question is:

Why all of the queues are giving a negative scalability on the Intel
Nehalem architecture ?

They are not saying on the paper why all the queues have a negative
scalability .on the Intel Nehalem architecture..

thank you.

Amine Moulay Ramdane.

1 post / 0 new
For more complete information about compiler optimizations, see our Optimization Notice.