Performance evaluation and tuning of lattice QCD on the next generation Blue Gene

Blue Gene/P is a next generation supercomputer following the technology of Blue Gene/L that supports ultra-scalability and efficient performance for many applications, including lattice QCD. We had experience optimizing lattice QCD problems on Blue Gene/L and we reported on the good performance and scalability. In this paper we describe how we are porting and tuning the lattice QCD code on Blue Gene/P and report on the initial code tuning efforts that take advantage of the new architectural features.