TY - CONF AU - Kaczmarek, O. AU - Schmidt, C. AU - Steinbrecher, P. AU - Wagner, M. TI - Conjugate gradient solvers on Intel Xeon Phi and NVIDIA GPUs IS - arXiv:1411.4439 CY - Hamburg PB - Deutsches Elektronen-Synchrotron, DESY M1 - PUBDB-2015-05358 M1 - arXiv:1411.4439 M1 - DESY-PROC-2014-05/28 SP - 157-162 PY - 2015 AB - Lattice Quantum Chromodynamics simulations typically spend most of the runtime in inversions of the Fermion Matrix. This part is therefore frequently optimized for various HPC architectures. Here we compare the performance of the Intel R © Xeon Phi TM to current Kepler-based NVIDIA R © Tesla TM GPUs running a conjugate gradient solver. By exposing more parallelism to the accelerator through inverting multiple vectors at the same time, we obtain a performance greater than 300 GFlop / s on both architectures. This more than doubles the performance of the inversions. We also give a short overview of the Knights Corner architecture, discuss some details of the implementation and the effort required to obtain the achieved performance T2 - GPU Computing in High-Energy Physics CY - 10 Sep 2014 - 12 Sep 2014, Pisa (Italy) Y2 - 10 Sep 2014 - 12 Sep 2014 M2 - Pisa, Italy KW - lattice field theory (INSPIRE) KW - quantum chromodynamics (INSPIRE) KW - numerical calculations (INSPIRE) KW - multiprocessor: graphics (INSPIRE) KW - programming (INSPIRE) KW - performance (autogen) KW - accelerator (autogen) LB - PUB:(DE-HGF)8 ; PUB:(DE-HGF)15 DO - DOI:10.3204/DESY-PROC-2014-05/28 UR - https://bib-pubdb1.desy.de/record/291386 ER -