We present results of the implementation of one MILC lattice QCD application-simulation with dynamical clover fermions using the hybrid-molecular dynamics R algorithm-on the Cell Broadband Engine processor. Fifty-four individual computational kernels responsible for 98.8% of the overall execution time were ported to the Cell's Synergistic Processing Elements (SPEs). The remaining application framework, including MPI-based distributed code execution, was left to the Cell's PowerPC processor. We observe that we only infrequently achieve more than 10 GFLOPS with any of the kernels, which is just over 4% of the Cell's peak performance. At the same time, many of the kernels are sustaining a bandwidth close to 20 GB/s, which is 78% of the Cell's peak. This indicates that the application performance is limited by the bandwidth between the main memory and the SPEs. In spite of this limitation, speedups of 8.7x (for 8x8x16x16 lattice) and 9.6x (for 16x16x16x16 lattice) were achieved when comparing a 3.2 GHz Cell processor to a single core of a 2.33 GHz Intel Xeon processor. When comparing the code scaled up to execute on a dual-Cell blade and a quad-core dual-chip Intel Xeon blade, the speedups are 1.5x (8x8x16x16 lattice) and 4.1x (16x16x16x16 lattice).
With recent developments in parallel supercomputing architecture, many core, multi-core, and GPU processors are now commonplace, resulting in more levels of parallelism, memory hierarchy, and programming complexity. It has been necessary to adapt the MILC code to these new processors starting with NVIDIA GPUs, and more recently, the Intel Xeon Phi processors. We report on our efforts to port and optimize our code for the Intel Knights Landing architecture. We consider performance of the MILC code with MPI and OpenMP, and optimizations with QOPQDP and QPhiX. For the latter approach, we concentr…
We investigate the coupling constant $g_{σππ}$ for the hadronic decay $σ\toππ$ only using the relevant three-point function, which is evaluated by the moving-wall source technique with a pretty good noise-to-signal ratio. This simulation is carried out on a $40^3\times96$ MILC gauge configuration with $N_f=2+1$ flavor of the "Asqtad" improved staggered dynamical sea quarks at the lattice spacing $a \approx 0.09$ fm. Our estimated value for this given MILC fine lattice gauge ensemble $g_{σππ}=2.71(42)$ GeV.
We review our work done to optimize the staggered conjugate gradient (CG) algorithm in the MILC code for use with the Intel Knights Landing (KNL) architecture. KNL is the second gener- ation Intel Xeon Phi processor. It is capable of massive thread parallelism, data parallelism, and high on-board memory bandwidth and is being adopted in supercomputing centers for scientific research. The CG solver consumes the majority of time in production running, so we have spent most of our effort on it. We compare performance of an MPI+OpenMP baseline version of the MILC code with a version incorporating …
We deliver lattice results for the $I=0$ $ππ$ elastic $s$-wave scattering length calculated with the MILC $N_f=3$ flavors of the Asqtad-improved staggered fermions. The scattering phase shifts are determined by Lüscher's formula from the energy-eigenvalues of $ππ$ systems at one center of mass frame and four moving frames using the moving wall source technique. Our measurements are good enough to resolve the scattering length $a$ and effective range $r$, moreover, it allows us to roughly estimate the shape parameter $P$. Using our lattice results, the scattering length $a$ and effective range …
Electromagnetic effects play an important role in many phenomena such as isospin-symmetry breaking in the hadron spectrum and the hadronic contributions to g-2. We have generalized the MILC QCD code to include the electromagnetic field. In this work, we focus on simulations including charged sea quarks using the RHMC algorithm. We show details of the dynamical QCD+QED simulation algorithm with compact QED. We analyze the code performance and results for hadron-spectrum observables.
Untangling the heavy-flavor mess: status of the Fermilab-MILC calculation of the $B_{(s)}\to D^{(\ast)}_{(s)}\ellν$ form factors
Universidad de Zaragoza（萨拉戈萨大学）；University of Utah（犹他大学）；University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Illinois Center for Advanced Studies of the Universe, University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校宇宙高级研究中心）；Indiana University（印第安纳大学）；Colorado State University（科罗拉多州立大学）；Fermi National Accelerator Laboratory（费米国家加速器实验室）
We present the first lattice QCD calculation of the $B_s$ and $B_d$ mixing parameters with physical light quark masses. We use MILC gluon field configurations that include $u$, $d$, $s$ and $c$ sea quarks at 3 values of the lattice spacing and with 3 values of the $u/d$ quark mass going down to the physical value. We use improved NRQCD for the valence $b$ quarks. Preliminary results show significant improvements over earlier values.
Colorado State University（科罗拉多州立大学）；University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；University of Illinois（伊利诺伊大学）；University of Utah（犹他大学）；Indiana University（印第安纳大学）；Fermi National Accelerator Laboratory（费米国家加速器实验室）；Syracuse University（雪城大学）；Universidad de Zaragoza（萨拉戈萨大学）
We present the status of on-going calculations of the $K\toπlν$ and $D\to K(π) lν$ semileptonic form factors at $q^2=0$. These form factors are important for the determination of the CKM matrix elements $\lvert{V_{us}}\rvert$ and $\lvert{V_{cs(d)}}\rvert$ respectively. This work uses the HISQ action for both valence quarks and sea quarks on MILC $N_f=2+1+1$ configurations. We employ twisted boundary conditions to calculate the form factors at zero momentum transfer directly. The $K\toπ$ results are an update to previously published results with new data at the physical point. The $D\to K(π)$ r…
We present an update on the ongoing calculations by the Fermilab Lattice, HPQCD, and MILC Collaboration of the leading-order (in electromagnetism) hadronic vacuum polarization contribution to the anomalous magnetic moment of the muon. Our project employs ensembles with four flavors of highly improved staggered fermions, physical light-quark masses, and four lattice spacings ranging from $a \approx 0.06$ to 0.15 fm for most of the results thus far.
Using conventional constituent-quark model, $I=1/2$ scalar $κ$, vector $K^\ast(892)$, and axial vector $K_1$ mesons are studied in the asqtad-improved staggered fermion with the wall-source and point-sink interpolators. The mass ratio of $m_κ/m_{K^\ast(892)}$ is numerically confirmed to vary apparently with quark mass, and the experimental ordering $m_{K^\ast(892)} > m_κ$ is elegantly hold when the light $u/d$ quark masses are sufficiently small, while the valence strange quarks are fixed to its physical values. We also get reasonable signals for $K_1$ meson suggested by SCALAR Collaboration f…
Neutral-meson mixing is loop suppressed in the Standard Model, leading to the possibility of enhanced sensitivity to new physics. The uncertainty in Standard Model predictions for $B$-meson oscillation frequencies is dominated by theoretical uncertainties within the short-distance $B$-meson hadronic matrix elements, motivating the need for improved precision. In $D$-meson mixing, the Standard Model short-distance contributions are further suppressed by the GIM mechanism allowing for the possibility of large new physics enhancements. A first-principle determination of the $D$-meson short-distan…
The Cabibbo-Kobayashi-Maskawa (CKM) matrix element $|V_{cb}|$ is extracted from exclusive semileptonic $B \to D^{(*)}$ decays adopting a novel unitarity-based approach which allows to determine in a full non-perturbative way the relevant hadronic form factors (FFs) in the whole kinematical range. By using existing lattice computations of the $B \to D^{(*)}$ FFs at small recoil from FNAL/MILC and JLQCD Collaborations, we show that it is possible to extrapolate their behavior also at large recoil without assuming any specific momentum dependence and without constraining their shape using experim…
