Questa è una vecchia versione del documento!
The text for the project is available on this page
In the lessons we stated performance of Xeon PHI is 2 to 4 times smaller than the performance observed on the host machine Xeon processors.
Actually, performance scale factor varies depending on the kind of code executed. Code with quite a number of branches may experience 10x and more slowdown w.r.t. host Xeon processor.
As far as the project report is concerned, please report scalability on PHI and times achieved when using parallelism degree equal to one on both Xeon PHI and host Xeon processor.
This page hosts all the informations relative to the access to lab machines.