- Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften
Boltzmannstr. 1
D-85748 Garching bei Muenchen
Volker Weinberg
Leibniz Supercomputing Centre, High Performance Systems Division, Department Member
Research Interests:
As code optimisation techniques are getting more and more important in HPC, LRZ @ GCS as one of the six European PRACE Advanced Training Centres (PATC) has extended its curriculum by a new PATC course " HPC code optimisation workshop " ,... more
As code optimisation techniques are getting more and more important in HPC, LRZ @ GCS as one of the six European PRACE Advanced Training Centres (PATC) has extended its curriculum by a new PATC course " HPC code optimisation workshop " , which took place at LRZ on May 4, 2017 for the first time.
Research Interests:
For the fourth time the Czech-Bavarian Competence Centre for Supercomputing Applications, CzeBaCCA, has organised a technical Intel Many Integrated Core (MIC) programming workshop combined with a scientific workshop about HPC simulations... more
For the fourth time the Czech-Bavarian Competence Centre for Supercomputing Applications, CzeBaCCA, has organised a technical Intel Many Integrated Core (MIC) programming workshop combined with a scientific workshop about HPC simulations in the field of environmental sciences. The Czech-Bavarian Competence Centre was established in 2016 by the Leibniz Supercomputing Centre (LRZ), the Department of Informatics of the Technical University of Munich (TUM) and the IT4Innovations National Supercomputing Centre of the Czech Republic to foster the Czech-German collaboration in high performance computing. One of the main objectives of the Competence Centre is to organise a series of Intel Xeon Phi specific technical workshops combined with scientific symposia on topics like optimisation of simulation codes in environmental science.
Research Interests:
The Leibniz Supercomputing Centre publishes in this booklet the complete material of the Intel MIC programming workshop that took place at LRZ on June 26 – 28, 2017. The workshop discussed Intel’s Many Integrated Core (MIC) architecture... more
The Leibniz Supercomputing Centre publishes in this booklet the complete material of the Intel MIC programming workshop that took place at LRZ on June 26 – 28, 2017. The workshop discussed Intel’s Many Integrated Core (MIC) architecture and various programming models for Intel Xeon Phi co-/processors. The workshop covered a wide range of topics from the description of the hardware of the Intel Xeon Phi co-/processors through information about the basic programming models as well as information about vectorisation and MCDRAM usage up to tools and strategies how to analyse and improve the performance of applications. The workshop mainly concentrated on techniques relevant for Knights Landing (KNL) based systems. During a plenary session on the last day 8 invited speakers from IPCC@LRZ, IPCC@TUM, IPCC@IT4Innovations, Intel, RRZE, the University of Regensburg, IPP and MPCDF talked about Intel Xeon Phi experience and best practice recommendations. Hands-on sessions were done on the Knights Corner (KNC) based system SuperMIC and two KNL test systems at LRZ.
Research Interests:
For the third time the Czech-Bavarian Competence Centre for Supercomputing Applications, CzeBaCCA, has organised a technical Intel Many Integrated Core (MIC) programming workshop combined with a scientific workshop about HPC simulations... more
For the third time the Czech-Bavarian Competence Centre for Supercomputing Applications, CzeBaCCA, has organised a technical Intel Many Integrated Core (MIC) programming workshop combined with a scientific workshop about HPC simulations in the field of geo- and environmental sciences. The Czech-Bavarian Competence Centre was established in 2016 by the Leibniz Supercomputing Centre (LRZ), the Department of Informatics of the Technical University of Munich (TUM) and the IT4Innovations National Supercomputing Centre of the Czech Republic to foster the Czech-German collaboration in high performance computing. One of the main objectives of the Competence Centre is to trigger new collaborations between Germans and Czech via scientific workshops and many-core architecture specific technical trainings.
Research Interests:
To foster the Czech-German collaboration in high performance computing, the Leibniz Supercomputing Centre (LRZ), the Department of Informatics of TUM and the National Supercomputing Centre of the Czech Republic, IT4Innovations, recently... more
To foster the Czech-German collaboration in high performance computing, the Leibniz Supercomputing Centre (LRZ), the Department of Informatics of TUM and the National Supercomputing Centre of the Czech Republic, IT4Innovations, recently joined forces and established the Czech-Bavarian Competence Centre for Supercomputing Applications (CzeBaCCA). Besides their joint research program around simulation software and tools for Salomon, one of the main objectives of the new Competence Centre is to organise a series of scientific workshops and Intel MIC (Many Integrated Core) architecture specific trainings. The article reports on the second series of workshops that took place at LRZ in Garching in June 2016 and combined a three-day “Intel MIC Programming Workshop” (June 27 – 29, 2016) with a three-day scientific workshop on “High Performance Computing for Water Related Hazards” (June 29 – July 1, 2016).
Research Interests:
On 18-22 April 2016 the Leibniz Supercomputing Centre hosted the 21st VI-HPS Tuning Workshop in a very fruitful cooperation with the Jülich Supercomputing Centre (JSC) and the VI-HPS consortium. This series of tuning workshops gives an... more
On 18-22 April 2016 the Leibniz Supercomputing Centre hosted the 21st VI-HPS Tuning Workshop in a very fruitful cooperation with the Jülich Supercomputing Centre (JSC) and the VI-HPS consortium. This series of tuning workshops gives an overview of the VI-HPS performance analysis and tuning tools suite, explains the functionality of individual tools and how to use them effectively, and offers hands-on experience and expert assistance using these tools on participants' own applications.
Research Interests:
Most HPC systems are clusters of shared memory nodes. These SMP nodes can be small multi-core CPUs up to large many-core CPUs. Parallel programming may combine the distributed memory parallelisation on the node interconnect (e.g., using... more
Most HPC systems are clusters of shared memory nodes. These SMP nodes can be small multi-core CPUs up to large many-core CPUs. Parallel programming may combine the distributed memory parallelisation on the node interconnect (e.g., using MPI) with the shared memory parallelisation inside of each node (e.g., using OpenMP or MPI-3.0 shared memory). As such hybrid programming techniques are getting more and more important in HPC, GCS as one of the 6 European PRACE Advanced Training Centres (PATC) has extended its curriculum by a new PATC course on hybrid programming techniques, which took place at LRZ on January 14, 2016 for the first time.
Research Interests:
To foster the Czech-German collaboration in high performance computing, the Leibniz Supercomputing Centre (LRZ), the Department of Informatics of TUM and the National Supercomputing Centre of the Czech Republic, IT4Innovations, recently... more
To foster the Czech-German collaboration in high performance computing, the Leibniz Supercomputing Centre (LRZ), the Department of Informatics of TUM and the National Supercomputing Centre of the Czech Republic, IT4Innovations, recently joined forces and established the Czech-Bavarian Competence Centre for Supercomputing Applications (CzeBaCCA).
Research Interests:
With the ever-increasing energy demands and prices and the need for uninterrupted services, data centre operators must find solutions to increase energy efficiency and reduce costs. Running from October 2011 to June 2015, the aim of the... more
With the ever-increasing energy demands and prices and the need for uninterrupted services, data centre operators must find solutions to increase energy efficiency and reduce costs. Running from October 2011 to June 2015, the aim of the European project Mont-Blanc has been to address this issue by developing an approach to Exascale computing based on embedded power-efficient technology. The main goals of the project were to i) build an HPC prototype using currently available energy-efficient embedded technology, ii) design a Next Generation system to overcome the limitations of the built prototype and iii) port a set of representative Exascale applications to the system.
Research Interests:
Research Interests:
Since the standards of parallel programming languages are becoming more and more complex and extensive, it can be hard to stay up to date with recent developments. LRZ has thus invited leading HPC experts to give updates of recent... more
Since the standards of parallel programming languages are becoming more and more complex and extensive, it can be hard to stay up to date with recent developments. LRZ has thus invited leading HPC experts to give updates of recent advances in parallel programming languages during a 1-day workshop on June 8 2015 at LRZ.
Research Interests:
We discuss a detailed strong and weak scaling analysis of PICCANTE, an open source, massively parallel, fully-relativistic Particle-In-Cell (PIC) code. PIC codes are widely used in plasma physics and astrophysics to study the cases where... more
We discuss a detailed strong and weak scaling analysis of PICCANTE, an open source, massively parallel, fully-relativistic Particle-In-Cell (PIC) code. PIC codes are widely used in plasma physics and astrophysics to study the cases where kinetic effects are relevant. PICCANTE is primarily developed to study laser-plasma interaction. Within a PRACE Preparatory Access Project, various revisions of different routines of the code have been analysed on the HPC systems JUQUEEN at J\"ulich Supercomputing Centre (JSC), Germany, and FERMI at CINECA, Italy, to improve the parallel scalability and the I/O performance of the application. The diagnostic tool Scalasca is used to filter out suboptimal routines. Different output strategies are discussed. The detailed strong and weak scaling behaviour of the improved code is presented in comparison with the original version of the code.
We discuss a detailed weak scaling analysis of GEM, a 3D MPI-parallelised gyrofluid code used in theoretical plasma physics at the Max Planck Institute of Plasma Physics, IPP at Garching b. M\"unchen, Germany. Within a PRACE Preparatory... more
We discuss a detailed weak scaling analysis of GEM, a 3D MPI-parallelised gyrofluid code used in theoretical plasma physics at the Max Planck Institute of Plasma Physics, IPP at Garching b. M\"unchen, Germany. Within a PRACE Preparatory Access Project various versions of the code have been analysed on the HPC systems SuperMUC at LRZ and JUQUEEN at J\"ulich Supercomputing Centre (JSC) to improve the parallel scalability of the application. The diagnostic tool Scalasca has been used to filter out suboptimal routines. The code uses the electromagnetic gyrofluid model which is a superset of magnetohydrodynamic and drift-Alfv\'en microturbulance and also includes several relevant kinetic processes. GEM can be used with different geometries depending on the targeted use case, and has been proven to show good scalability when the computational domain is distributed amongst two dimensions. Such a distribution allows grids with sufficient size to describe small scale tokamak devices. In order to enable simulation of very large tokamaks (such as the next generation nuclear fusion device ITER in Cadarache, France) the third dimension has been parallelised and weak scaling has been achieved for significantly larger grids.
As the complexity and size of challenges in science and engineering are continually increasing, it is highly important that applications are able to scale strongly to very large numbers of cores (>100,000 cores) to enable HPC systems... more
As the complexity and size of challenges in science and engineering are continually increasing, it is highly important that applications are able to scale strongly to very large numbers of cores (>100,000 cores) to enable HPC systems to be utilised efficiently. This paper presents results of strong scaling tests performed with an MPI only and a hybrid MPI + OpenMP version of the Lattice QCD application BQCD on the European Tier-0 system SuperMUC at LRZ.
With the rapidly growing demand for computing power new accelerator based architectures have entered the world of high performance computing since around 5 years. In particular GPGPUs have recently become very popular, however programming... more
With the rapidly growing demand for computing power new accelerator based architectures have entered the world of high performance computing since around 5 years. In particular GPGPUs have recently become very popular, however programming GPGPUs using programming languages like CUDA or OpenCL is cumbersome and error-prone. Trying to overcome these difficulties, Intel developed their own Many Integrated Core (MIC) architecture which can be programmed using standard parallel programming techniques like OpenMP and MPI. In the beginning of 2013, the first production-level cards named Intel Xeon Phi came on the market. LRZ has been considered by Intel as a leading research centre for evaluating coprocessors based on the MIC architecture since 2010 under strict NDA. Since the Intel Xeon Phi is now generally available, we can share our experience on programming Intel's new MIC architecture.
Intel Array Building Blocks is a high-level data-parallel programming environment designed to produce scalable and portable results on existing and upcoming multi- and many-core platforms. We have chosen several mathematical kernels - a... more
Intel Array Building Blocks is a high-level data-parallel programming environment designed to produce scalable and portable results on existing and upcoming multi- and many-core platforms. We have chosen several mathematical kernels - a dense matrix-matrix multiplication, a sparse matrix-vector multiplication, a 1-D complex FFT and a conjugate gradients solver - as synthetic benchmarks and representatives of scientific codes and ported them to ArBB. This whitepaper describes the ArBB ports and presents performance and scaling measurements on the Westmere-EX based system SuperMIG at LRZ in comparison with OpenMP and MKL.
Research Interests:
Knowledge of the derivative of the topological susceptibility at zero momentum is important for assessing the validity of the Witten-Veneziano formula for the eta' mass, and likewise for the resolution of the EMC proton spin problem.... more
Knowledge of the derivative of the topological susceptibility at zero momentum is important for assessing the validity of the Witten-Veneziano formula for the eta' mass, and likewise for the resolution of the EMC proton spin problem. We investigate the momentum dependence of the topological susceptibility and its derivative at zero momentum using overlap fermions in quenched lattice QCD simulations. We expose the role of the low-lying Dirac eigenmodes for the topological charge density, and find a negative value for the derivative. While the sign of the derivative is consistent with the QCD sum rule for pure Yang-Mills theory, the absolute value is overestimated if the contribution from higher eigenmodes is ignored.
Research Interests:
Research Interests:
A detailed comparison is made between the topological structure of quenched QCD as revealed by the recently proposed over-improved stout-link smearing in conjunction with an improved gluonic definition of the topological density on one... more
A detailed comparison is made between the topological structure of quenched QCD as revealed by the recently proposed over-improved stout-link smearing in conjunction with an improved gluonic definition of the topological density on one hand and a similar analysis made possible by the overlap-fermionic topological charge density both with and without variable ultraviolet cutoff lambdacut. The matching is twofold, provided
Research Interests:
Research Interests:
We present selected recent results of the QCDSF collaboration on the localization and dimensionality of low overlap eigenmodes and of the topological density in the quenched SU(3) vacuum. We discuss the correlations between the... more
We present selected recent results of the QCDSF collaboration on the localization and dimensionality of low overlap eigenmodes and of the topological density in the quenched SU(3) vacuum. We discuss the correlations between the topological structure revealed by overlap fermions without filtering and the confining monopole and P-vortex structure obtained in the Indirect Maximal Center Gauge.
Overlap fermions are particularly well suited to study the finite temperature dynamics of the chiral symmetry restoration transition of QCD, which might be just an analytic crossover. Using gauge field configurations on a 24^3x10 lattice... more
Overlap fermions are particularly well suited to study the finite temperature dynamics of the chiral symmetry restoration transition of QCD, which might be just an analytic crossover. Using gauge field configurations on a 24^3x10 lattice with N_f=2 flavours of dynamical Wilson-clover quarks generated by the DIK collaboration, we compute the lowest 50 eigenmodes of the overlap Dirac operator and try to locate the transition by fermionic means. We analyse the spectral density, local chirality and localisation properties of the low-lying modes and illustrate the changing topological and (anti-) selfdual structure of the underlying gauge fields across the transition.
Research Interests:
Research Interests:
These are the proceedings of the workshop on " Lattice QCD, Chiral Perturbation Theory and Hadron Phenomenology " held at the European Centre for Theoretical Studies in Nuclear Physics and Related Areas from October 2 to 6, 2006. The... more
These are the proceedings of the workshop on " Lattice QCD, Chiral Perturbation Theory and Hadron Phenomenology " held at the European Centre for Theoretical Studies in Nuclear Physics and Related Areas from October 2 to 6, 2006. The workshop concentrated on bringing together researchers working in lattice QCD and chiral perturbation theory with the aim of improving our understanding how hadron properties can be calculated and analyzed from first principles. Included are a short contribution per talk.