Learn how to deliver more performance and better energy efficiency with the right processor. In this paper, we investigate the porting and optimization of a test problem for the Intel Xeon Phi coprocessor. The test problem is a basic N-body simulation using common code in C for the host processor and for the coprocessor.
Download this exclusive paper here and learn more about the results how you can benefit.
I actually do not fully understand the mathematics but looks very very impressive.
AVX also has a limited-precision reiprocal square root instruction. Using -Ofast -xAVX on the Sandy Bridge, I can recover the 1:3 performance ratio between SNB and Phi for the optimized code version.
@GeorgHager Thank you, Georg! Indeed, we had overlooked the argument -xAVX. I have published an addendum to the paper at http://research.colfaxinternational.com/post/2013/01/07/Nbody-Xeon-Phi.aspx with corrected results.