Understanding Resource Selection Requirements for Computationally Intensive Tasks on Heterogeneous Computing Infrastructure

The Liquid Argon Time Projection Chamber (LArTPC) technology plays an essential role in many current and future neutrino experiments. Accurate and fast simulation is critical to developing efficient analysis algorithms and precise physics model projections. The speed of simulation becomes more important as Deep Learning algorithms are getting more widely used in LArTPC analysis and their training requires a large simulated dataset. Heterogeneous computing is an efficient way to delegate computationally intensive tasks to specialized hardware. However, as the landscape of compute accelerators quickly evolves, it becomes increasingly difficult to manually adapt the code to the latest hardware or software environments. A solution which is portable to multiple hardware architectures without substantially compromising performance would thus be very beneficial, especially for long-term projects such as the LArTPC simulations. In search of a portable, scalable and maintainable software solution for LArTPC simulations, we have started to explore high-level portable programming frameworks that support several hardware backends. In this paper, we present our experience porting the LArTPC simulation code in the Wire-Cell Toolkit to NVIDIA GPUs, first with the CUDA programming model and then with a portable library called Kokkos. Preliminary performance results on NVIDIA V100 GPUs and multi-core CPUs are presented, followed by a discussion of the factors affiecting the performance and plans for future improvements.

Download Full-text

CRIC: Computing Resource Information Catalogue as a unified topology system for a large scale, heterogeneous and dynamic computing infrastructure

EPJ Web of Conferences ◽

10.1051/epjconf/202024503032 ◽

2020 ◽

Vol 245 ◽

pp. 03032

Author(s):

Alexey Anisenkov ◽

Julia Andreeva ◽

Alessandro Di Girolamo ◽

Panos Paparrigopoulos ◽

Boris Vasilev

Keyword(s):

Large Scale ◽

Heterogeneous Computing ◽

Third Party ◽

Resource Information ◽

Independent Information ◽

Recent Developments ◽

Level Information ◽

Suitable Framework ◽

High Level ◽

Computing Infrastructure

CRIC is a high-level information system which provides flexible, reliable and complete topology and configuration description for a large scale distributed heterogeneous computing infrastructure. CRIC aims to facilitate distributed computing operations for the LHC experiments and consolidate WLCG topology information. It aggregates information coming from various low-level information sources and complements topology description with experimentspecific data structures and settings required by the LHC VOs in order to exploit computing resources. Being an experiment-oriented but still experiment-independent information middleware, CRIC offers a generic solution, in the form of a suitable framework with appropriate interfaces implemented, which can be successfully applied on the global WLCG level or at the level of a particular LHC experiment. For example there are CRIC instances for CMS[11] and ATLAS[10]. CRIC can even be used for a special task. For example, a dedicated CRIC instance has been built to support transfer tests performed by DOMA Third Party Copy working group. Moreover, extensibility and flexibility of the system allow CRIC to follow technology evolution and easily implement concepts required to describe new types of computing and storage resources. The contribution describes the overall CRIC architecture, the plug-in based implementation of the CRIC components as well as recent developments and future plans.

Download Full-text

CLEVER

Achieving Federated and Self-Manageable Cloud Infrastructures ◽

10.4018/978-1-4666-1631-8.ch012 ◽

2012 ◽

pp. 219-241

Author(s):

Francesco Tusa ◽

Maurizio Paone ◽

Massimo Villari

Keyword(s):

Internet Of Things ◽

Heterogeneous Computing ◽

Electronic Device ◽

The Internet ◽

Use Case ◽

Grid Infrastructures ◽

The Internet Of Things ◽

Computing Infrastructure ◽

Cloud Resources

This chapter describes both the design and architecture of the CLEVER cloud middleware, pointing out the possibilities it offers towards enlarging the concept of federation in more directions. CLEVER is able to accomplish such an enlargement enabling the interaction among whatever type of electronic device connected to Internet, thus offering the opportunity of implementing the Internet of Things. Together with this type of perspective, CLEVER aims to “aggregate” heterogeneous computing infrastructure by putting together Cloud and Grid, as an example. The chapter starts with a description of the cloud projects related to CLEVER, followed by a discussion on the middleware components that mainly focuses on the innovative features they have, in particular the communication mechanisms adopted. The second part of the chapter presents a real use case that exploits the CLEVER features that allow easy creation of federated clouds’ infrastructures that can be also based on integration with existing Grids; it is demonstrated thanks to the “oneshot” CLEVER deploying mechanism. It is possible to scale dynamically the cloud resources by taking advantage of the existing Grid infrastructures, and minimizing the changes needed at the involved management middleware.

Download Full-text

A Cloud Computing Infrastructure on Heterogeneous Computing Resources

Journal of Computers ◽

10.4304/jcp.6.8.1789-1796 ◽

2011 ◽

Vol 6 (8) ◽

Cited By ~ 6

Author(s):

Baomin XU ◽

Ning Wang ◽

Chunyan Li

Keyword(s):

Cloud Computing ◽

Heterogeneous Computing ◽

Computing Infrastructure

Download Full-text

Geocomputation over the Emerging Heterogeneous Computing Infrastructure

Transactions in GIS ◽

10.1111/tgis.12108 ◽

2014 ◽

Vol 18 ◽

pp. 3-24 ◽

Cited By ~ 11

Author(s):

Xuan Shi ◽

Chenggang Lai ◽

Miaoqing Huang ◽

Haihang You

Keyword(s):

Heterogeneous Computing ◽

Computing Infrastructure

Download Full-text

Real-Time Task Assignment Approach Leveraging Reinforcement Learning with Evolution Strategies for Long-Term Latency Minimization in Fog Computing

Sensors ◽

10.3390/s18092830 ◽

2018 ◽

Vol 18 (9) ◽

pp. 2830 ◽

Cited By ~ 11

Author(s):

Long Mai ◽

Nhu-Ngoc Dao ◽

Minho Park

Keyword(s):

Reinforcement Learning ◽

Real Time ◽

Heterogeneous Computing ◽

Learning Algorithm ◽

Fog Computing ◽

Task Assignment ◽

Evolution Strategies ◽

Time Task ◽

Computing Infrastructure

The emerging fog computing technology is characterized by an ultralow latency response, which benefits a massive number of time-sensitive services and applications in the Internet of things (IoT) era. To this end, the fog computing infrastructure must minimize latencies for both service delivery and execution phases. While the transmission latency significantly depends on external factors (e.g., channel bandwidth, communication resources, and interferences), the computation latency can be considered as an internal issue that the fog computing infrastructure could actively self-handle. From this view point, we propose a reinforcement learning approach that utilizes the evolution strategies for real-time task assignment among fog servers to minimize the total computation latency during a long-term period. Experimental results demonstrate that the proposed approach reduces the latency by approximately 16.1% compared to the existing methods. Additionally, the proposed learning algorithm has low computational complexity and an effectively parallel operation; therefore, it is especially appropriate to be implemented in modern heterogeneous computing platforms.

Download Full-text

Computational Simulation of Gas Turbines: Part 2—Extensible Domain Framework

Journal of Engineering for Gas Turbines and Power ◽

10.1115/1.1287489 ◽

2000 ◽

Vol 122 (3) ◽

pp. 377-386 ◽

Cited By ~ 5

Author(s):

John A. Reed ◽

Abdollah A. Afjeh

Keyword(s):

Gas Turbine ◽

Gas Turbines ◽

Design Patterns ◽

Heterogeneous Computing ◽

Computational Simulation ◽

Object Oriented ◽

Turbine Engine ◽

Engine Model ◽

Computationally Intensive ◽

High Level

This paper describes the design concepts and object-oriented architecture of Onyx, an extensible domain framework for computational simulation of gas turbine engines. Onyx provides a flexible environment for defining, modifying, and simulating the component-based gas turbine models described in Part 1 of this paper. Using advanced object-oriented technologies such as design patterns and frameworks, Onyx enables users to customize and extend the framework to add new functionality or adapt simulation behavior as required. A customizable visual interface provides high-level symbolic control of propulsion system construction and execution. For computationally-intensive analysis, components may be distributed across heterogeneous computing architectures and operating systems. A distributed gas turbine engine model is developed and simulated to illustrate the use of the framework. [S0742-4795(00)02403-0]

Download Full-text

Computational Simulation of Gas Turbines: Part II — Extensible Domain Framework

Volume 1: Aircraft Engine; Marine; Turbomachinery; Microturbines and Small Turbomachinery ◽

10.1115/99-gt-347 ◽

1999 ◽

Cited By ~ 2

Author(s):

John A. Reed ◽

Abdollah A. Afjeh

Keyword(s):

Gas Turbine ◽

Gas Turbines ◽

Design Patterns ◽

Heterogeneous Computing ◽

Computational Simulation ◽

Object Oriented ◽

Turbine Engine ◽

Engine Model ◽

Computationally Intensive ◽

High Level

This paper describes the design concepts and object-oriented architecture of Onyx, an extensible domain framework for computational simulation of gas turbine engines. Onyx provides a flexible environment for defining, modifying and simulating the component-based gas turbine models described in Part 1 of this paper. Using advanced object-oriented technologies such as design patterns and frameworks, Onyx enables users to customize and extend the framework to add new functionality or adapt simulation behavior as required. A customizable visual interface provides high-level symbolic control of propulsion system construction and execution. For computationally-intensive analysis, components may be distributed across heterogeneous computing architectures and operating systems. A distributed gas turbine engine model is developed and simulated to illustrate the use of the framework.

Download Full-text

Metaheuristics Based Optimization Technique for Protein-Ligand Docking

International Journal of Recent Technology and Engineering - 2 ◽

10.35940/ijrte.c6834.098319 ◽

2019 ◽

Vol 8 (3) ◽

pp. 4617-4622

Keyword(s):

Molecular Docking ◽

Virtual Screening ◽

Heterogeneous Computing ◽

Optimization Technique ◽

Ligand Docking ◽

Screening Methods ◽

Scoring Functions ◽

Parallel Metaheuristics ◽

Novel Approach ◽

Computationally Intensive

Virtual screening using molecular docking requires optimization, which can be solved by using metaheuristics methods. Typically the interaction between two compounds is calculated using computationally intensive Scoring Functions (SF) which is computed in several spots which are called as binding surfaces. In this paper we present a novel approach for molecular docking which is based on parameterized and parallel metaheuristics which is useful in leveraging heterogeneous computing based on heterogeneous architectures. The approach decides on the optimization technique at running time by setting up a new configuration schema that allows parallel offloading of the data intensive sections of the docking. Hence the docking process is carried out in parallel efficiently while performing the metaheuristics execution. The approach carries out docking and computations of molecular interactions required for SF in parallel so that the time efficiency is improved. This opens a new path for further developments in virtual screening methods in heterogeneous platform.

Download Full-text