Lattice QCD with Domain Decomposition on Intel Xeon Phi Co-Processors

Abstract

The gap between the cost of moving data and the cost of computing continues to grow, making it ever harder to design iterative solvers on extreme-scale architectures. This problem can be alleviated by alternative algorithms that reduce the amount of data movement. We investigate this in the context of Lattice Quantum Chromodynamics and implement such an alternative solver algorithm, based on domain decomposition, on Intel Xeon Phi co-processor (KNC) clusters. We demonstrate close-to-linear on-chip scaling to all 60 cores of the KNC. With a mix of single- and half-precision the domain-decomposition method sustains 400-500 Gflop/s per chip. Compared to an optimized KNC implementation of a standard solver [1], our full multi-node domain-decomposition solver strong-scales to more nodes and reduces the time-to-solution by a factor of 5.

12 pages, 7 figures, presented at Supercomputing 2014, November 16-21, 2014, New Orleans, Louisiana, USA, speaker Simon Heybrock; SC '14 Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, pages 69-80, IEEE Press Piscataway, NJ, USA (c)2014

References in corpus (12)

• The Chroma Software System for Lattice QCD
• Schwarz-preconditioned HMC algorithm for two-flavour lattice QCD
• Adaptive multigrid algorithm for the lattice Wilson-Dirac operator
• Scaling Lattice QCD beyond 100 GPUs
• Local coherence and deflation of the low quark modes in lattice QCD
• Nucleon mass and sigma term from lattice QCD with two light fermion flavors
• Solution of the Dirac equation in lattice QCD using a domain decomposition method
• Scaling study of dynamical smeared-link clover fermions
• An adaptive aggregation based domain decomposition multilevel method for the lattice wilson dirac operator: multilevel results
• Adaptive Aggregation Based Domain Decomposition Multigrid for the Lattice Wilson Dirac Operator
• Domain Decomposition method on GPU cluster
• Deflation and Flexible SAP-Preconditioning of GMRES in Lattice QCD Simulation

Cited by in corpus (18)

• Novel quark smearing for hadrons with high momenta in lattice QCD
• Charmonium-like resonances with $J^{PC}=0^{++},2^{++}$ in coupled $D\bar{D}$, $D_s{\bar{D}}_s$ scattering on the lattice
• Preparation of the SU(3) Lattice Yang-Mills Vacuum with Variational Quantum Methods
• Hadro-quarkonium from Lattice QCD
• Masses and decay constants of the $\eta$ and ${\eta }^{\prime }$ mesons from lattice QCD
• Status and Future Perspectives for Lattice Gauge Theory Calculations to the Exascale and Beyond
• Towards the quark mass dependence of $T_{cc}^{+}$ from lattice QCD
• Charmonium resonances with $J^{PC}=1^{--}$ and $3^{--}$ from $\bar{D}D$ scattering on the lattice
• ECM modeling and performance tuning of SpMV and Lattice QCD on A64FX
• Identifying spin and parity of charmonia in flight with lattice QCD
• Lattice study of $cc\bar{u}\bar{s}$ tetraquark channel in $D^{(\ast )}{D}_s^{(\ast )}$ scattering
• Towards Lattice Quantum Chromodynamics on FPGA devices
• A performance evaluation of CCS QCD Benchmark on the COMA (Intel(R) Xeon Phi${}^{TM}$, KNC) system
• QPACE 2 and Domain Decomposition on the Intel Xeon Phi
• Adaptive algebraic multigrid on SIMD architectures
• Exact diagonalization of quantum lattice models on coprocessors
• Charmonium resonances on the lattice
• Practical Implementation of Lattice QCD Simulation on Intel Xeon Phi Knights Landing