Hybrid High Performance Computing Research Articles

Induced homomorphism: Kirchhoff’s law in photonics

Abstract When solving, modeling or reasoning about complex problems, it is usually convenient to use the knowledge of a parallel physical system for representing it. This is the case of lumped-circuit abstraction, which can be used for representing mechanical and acoustic systems, thermal and heat-diffusion problems and in general partial differential equations. Integrated photonic platforms hold the prospective to perform signal processing and analog computing inherently, by mapping into hardware specific operations which relies on the wave-nature of their signals, without trusting on logic gates and digital states like electronics. Here, we argue that in absence of a straightforward parallelism a homomorphism can be induced. We introduce a photonic platform capable of mimicking Kirchhoff’s law in photonics and used as node of a finite difference mesh for solving partial differential equation using monochromatic light in the telecommunication wavelength. Our approach experimentally demonstrates an arbitrary set of boundary conditions, generating a one-shot discrete solution of a Laplace partial differential equation, with an accuracy above 95% with respect to commercial solvers. Our photonic engine can provide a route to achieve chip-scale, fast (10 s of ps), and integrable reprogrammable accelerators for the next generation hybrid high-performance computing. Summary A photonic integrated platform which can mimic Kirchhoff’s law in photonics is used for approximately solve partial differential equations noniteratively using light, with high throughput and low-energy levels.

Formation of an Individual Modeling Environment in a Hybrid High-Performance Computing System

In this paper, we consider the problem of solving scientific problems in the field of materials science in the environment of high-performance computing systems. Mathematical modeling methods implemented by specialized modeling systems is used to solve a certain kind of problems in materials science. The modeling systems show the greatest efficiency when deployed in hybrid high-performance computing systems (HHPC) that allow solving problems in a reasonable time period with sufficient accuracy. However, there are a number of restrictions that affect the work of research teams with modeling systems in the HHPC computing environment: the need to access graphics accelerators at the stage of developing and debugging algorithms in the modeling system, the need to use several modeling systems in order to obtain the optimal solution, and the need to dynamically change the settings of modeling systems for solving problems. The solution to the problem of these restrictions is assigned to the individual modeling environment operating in the HHPC computing environment. The best solution for creating an individual modeling environment is virtual containerization technology. We propose an algorithm for the formation of an individual modeling environment in a hybrid high-performance computing complex based on the Docker virtual containerization system. An individual simulation environment is created by installing the required software in the base container, setting environment variables, and installing custom software and licenses. A feature of the algorithm is the ability to form a library image from a base container with a customized individual modeling environment. The direction for further research work is indicated in the conclusion. The algorithm presented here is independent of the implementation of the problems’ control system and may be used for any high-performance computing system.

On the Use of Large Intel Xeon Phi Clusters for GEANT4-Based Simulations

Abstract GEANT4 is the basic software for fast and precise simulation of particle interactions with matter. Along the way towards enabling the execution of GEANT4 based simulations on hybrid High Performance Computing (HPC) architectures with large clusters of Intel Xeon Phi co-processors, we study the performance of this software suit on the supercomputer system Avitohol@BAS, Some practical scripts are collected in the supplementary material shown in the appendix.

Job placement advisor based on turnaround predictions for HPC hybrid clouds

Several companies and research institutes are moving their CPU-intensive applications to hybrid High Performance Computing (HPC) cloud environments. Such a shift depends on the creation of software systems that help users decide where a job should be placed considering execution time and queue wait time to access on-premise clusters. Relying blindly on turnaround prediction techniques will affect negatively response times inside HPC cloud environments. This paper introduces a tool to make job placement decisions in HPC hybrid cloud environments taking into account the inaccuracy of execution and waiting time predictions. We used job traces from real supercomputing centers to run our experiments, and compared the performance between environments using real speedup curves. We also extended a state-of-the-art machine learning based predictor to work with data from the cluster scheduler. Our main findings are as follows: (i) depending on workload characteristics, there is a turning point where predictions should be disregarded in favor of a more conservative decision to minimize job turnaround times and (ii) scheduler data plays a key role in improving predictions generated with machine learning using job trace data—our experiments showed around 20% prediction accuracy improvements.

Energy Study of Monte Carlo and Quasi-Monte Carlo Algorithms for Solving Integral Equations

In the past few years the development of exascale computing technology necessitated to obtain an estimate for the energy consumption when large-scale problems are solved with different high-performance computing (HPC) systems. In this paper we study the energy efficiency of a class of Monte Carlo (MC) and Quasi-Monte Carlo (QMC) algorithms for a given integral equation using hybrid HPC systems. The algorithms are applied to solve quantum kinetic integral equations describing ultra-fast transport in quantum wire. We compare the energy performance of the algorithms using a GPU-based computer platform and CPU-based computer platform both with and without hyper-threading (HT) technology. We use SPRNG library and CURAND generator to produce parallel pseudo-random (PPR) sequences for the MC algorithms on CPU-based and GPU -based platforms, respectively. For our QMC algorithms Sobol and Halton sequences are used to produce parallel quasi-random (PQR) sequences.We compare the obtained results of the tested algorithms with respect to the given energy metric. The results of our study demonstrate the importance of taking into account not only scalability of the HPC intensive algorithms but also their energy efficiency. They also show the need for further optimisation of the QMC algorithms when GPU-based computing platforms are used.

A hybrid solution method for CFD applications on GPU-accelerated hybrid HPC platforms

Heterogeneous multiprocessor systems, where commodity multicore processors are coupled with graphics processing units (GPUs), have been widely used in high performance computing (HPC). In this work, we focus on the design and optimization of Computational Fluid Dynamics (CFD) applications on such HPC platforms. In order to fully utilize the computational power of such heterogeneous platforms, we propose to design the performance-critical part of CFD applications, namely the linear equation solvers, in a hybrid way. A hybrid linear solver includes both one CPU version and one GPU version of code for solving a linear equations system. When a hybrid linear equation solver is invoked during the CFD simulation, the CPU portion and the GPU portion will be run on corresponding processing devices respectively in parallel according to the execution configuration. Furthermore, we propose to build functional performance models (FPMs) of processing devices and use FPM-based heterogeneous decomposition method to distribute workload between heterogeneous processing devices, in order to ensure balanced workload and optimized communication overhead. Efficiency of this approach is demonstrated by experiments with numerical simulation of lid-driven cavity flow on both a hybrid server and a hybrid cluster.

A hybrid HPC/cloud distributed infrastructure: Coupling EC2 cloud resources with HPC clusters to run large tightly coupled multiscale applications

In this paper, we report on the experimental results of running a large, tightly coupled, distributed multiscale computation over a hybrid High Performance Computing (HPC) infrastructures. We connected EC2 based cloud clusters located in USA to university clusters located in Switzerland. We ran a concurrent multiscale MPI based application on this infrastructure and measured the overhead induced by extending our HPC clusters with EC2 resources. Our results indicate that accommodating some parts of the multiscale computation on cloud resources can lead to low performance without a proper adjustment of CPUs power and workload. However, by enforcing a load-balancing strategy one can benefit from the extra Cloud resources.

Large-scale, high-resolution agricultural systems modeling using a hybrid approach combining grid computing and parallel processing

The solution of complex global challenges in the land system, such as food and energy security, requires information on the management of agricultural systems at a high spatial and temporal resolution over continental or global extents. However, computing capacity remains a barrier to large-scale, high-resolution agricultural modeling. To model wheat production, soil carbon, and nitrogen dynamics in Australia's cropping regions at a high resolution, we developed a hybrid computing approach combining parallel processing and grid computing. The hybrid approach distributes tasks across a heterogeneous grid computing pool and fully utilizes all the resources of computers within the pool. We simulated 325 management scenarios (nitrogen application rates and stubble management) at a daily time step over 122 years, for 12,707 climate–soil zones using the Windows-based Agricultural Production Systems SIMulator (APSIM). These simulations would have taken over 30 years on a single computer. Our hybrid high performance computing (HPC) approach completed the modeling within 10.5 days—a speed-up of over 1000 times—with most jobs finishing within the first few days. The approach utilizes existing idle organization-wide computing resources and eliminates the need to translate Windows-based models to other operating systems for implementation on computing clusters. There are however, numerous computing challenges that need to be addressed for the effective use of these techniques and there remain several potential areas for further performance improvement. The results demonstrate the effectiveness of the approach in making high-resolution modeling of agricultural systems possible over continental and global scales.

Hybrid Computing—Where HPC meets grid and Cloud Computing

We introduce a hybrid High Performance Computing (HPC) infrastructure architecture that provides predictable execution of scientific applications, and scales from a single resource to multiple resources, with different ownership, policy, and geographic locations. We identify three paradigms in the evolution of HPC and high-throughput computing: owner-centric HPC (traditional), Grid computing, and Cloud computing. After analyzing the synergies among HPC, Grid and Cloud computing, we argue for an architecture that combines the benefits of these technologies. We call the building block of this architecture, Elastic Cluster. We describe the concept of Elastic Cluster and show how it can be used to achieve effective and predictable execution of HPC workloads. Then we discuss implementation aspects, and propose a new distributed information system design that combines features of distributed hash tables and relational databases.