Jon M. Hjelmervik

State-of-the-art in heterogeneous computing

Node level heterogeneous architectures have become attractive during the last decade for several reasons: compared to traditional symmetric CPUs, they offer high peak performance and are energy and/or cost efficient. With the increase of fine-grained parallelism in high-performance computing, as well as the introduction of parallelism in workstations, there is an acute need for a good overview and understanding of these architectures. We give an overview of the state-of-the-art in heterogeneous computing, focusing on three commonly found architectures: the Cell Broadband Engine Architecture, graphics processing units (GPUs), and field programmable gate arrays (FPGAs). We present a review of hardware, available software tools, and an overview of state-of-the-art techniques and algorithms. Furthermore, we present a qualitative and quantitative comparison of the architectures, and give our view on the future of heterogeneous computing.

Visual simulation of shallow-water waves

A commodity-type graphics card (GPU) is used to simulate nonlinear water waves described by a system of balance laws called the shallow-water system. To solve this hyperbolic system we use explicit high-resolution central-upwind schemes, which are particularly well suited for exploiting the parallel processing power of the GPU. In fact, simulations on the GPU are found to run 15‐30 times faster than on a CPU. The simulated cases involve dry-bed zones and non-trivial bottom topographies, which are real challenges to the robustness and accuracy of the discretization.

How to Solve Systems of Conservation Laws Numerically Using the Graphics Processor as a High-Performance Computational Engine

The paper has two main themes: The first theme is to give the reader an introduction to modern methods for systems of conservation laws. To this end, we start by introducing two classical schemes, the Lax-Friedrichs scheme and the Lax-Wendroff scheme. Using a simple example, we show how these two schemes fail to give accurate approximations to solutions containing discontinuities. We then introduce a general class of semi-discrete finite-volume schemes that are designed to produce accurate resolution of both smooth and nonsmooth parts of the solution. Using this special class we wish to introduce the reader to the basic principles used to design modern high-resolution schemes. As examples of systems of conservation laws, we consider the shallow-water equations for water waves and the Euler equations for the dynamics of an ideal gas.

An Introduction to General-Purpose Computing on Programmable Graphics Hardware

Using graphics hardware for general-purpose computations (GPGPU) has for selected applications shown a performance increase of more than one order of magnitude compared to traditional CPU implementations. The intent of this paper is to give an introduction to the use of graphics hardware as a computational resource. Understanding the architecture of graphics hardware is essential to comprehend GPGPU-programming. This paper first addresses the fixed functionality graphics pipeline, and then explains the architecture and programming model of programmable graphics hardware. As the CPU is instruction driven, while a graphics processing unit (GPU) is data stream driven, a good CPU algorithm is not necessarily well suited for GPU implementation. We will illustrate this with some commonly used GPU algorithms. The paper winds up with examples of GPGPU-research at SINTEF within simulation, visualization, image processing, and geometry processing.

GPU-Accelerated Shape Simplification for Mechanical-Based Applications

In this paper we present a GPU-based method for removing shape details of 3D models. 3D models used in finite element analysis (FEA) are often either constructed for the purpose of manufacturing, or a result of 3D scanning. The models therefore contain shape details that are neither important for FEA nor compatible with the mechanical hypotheses. Vertex removal is a popular method for removing geometrical details where vertices are removed one by one, provided certain constraints are satisfied. The constraints can either be based purely on geometrical properties, or also on mechanical ones. The computations required in this process can be time consuming, especially if mechanical constraints are involved. The main idea behind our method is to perform the computations for all the vertices in parallel using graphics hardware, and then use the CPU to maintain the data structure representing the triangulation. As a result, simplification functions can stay interactive while incorporating complementary mechanically-based criteria in addition to the geometric ones involved in shape transformation.

A framework for OpenGL client-server rendering

We present a software framework that facilitates the development of OpenGL applications utilizing the limited GPU capacities of a portable client in combination with the high-end rendering hardware on a server. The resulting web-application uses standard technologies and can be run on a wide variety of devices, such as smart phones, tablets and laptops. The framework is designed to make it simple changing an existing OpenGL application into a web-application, gradually adding client-side rendering. Furthermore, it provides automatic network scaling to provide interactivity even on poor connections.

Interactive Isogeometric Volume Visualization with Pixel-Accurate Geometry

A recent development, called isogeometric analysis, provides a unified approach for design, analysis and optimization of functional products in industry. Traditional volume rendering methods for inspecting the results from the numerical simulations cannot be applied directly to isogeometric models. We present a novel approach for interactive visualization of isogeometric analysis results, ensuring correct, i.e., pixel-accurate geometry of the volume including its bounding surfaces. The entire OpenGL pipeline is used in a multi-stage algorithm leveraging techniques from surface rendering, order-independent transparency, as well as theory and numerical methods for ordinary differential equations. We showcase the efficiency of our approach on different models relevant to industry, ranging from quality inspection of the parametrization of the geometry, to stress analysis in linear elasticity, to visualization of computational fluid dynamics results.

Interactive Exploration of Big Scientific Data: New Representations and Techniques

Although splines have been in popular use in CAD for more than half a century, spline research is still an active field, driven by the challenges we are facing today within isogeometric analysis and big data. Splines are likely to play a vital future role in enabling effective big data exploration techniques in 3D, 4D, and beyond.

Interactive Pixel-Accurate Rendering of LR-Splines and T-Splines

Flexible surface types on irregular grids, such as T-splines and LR-splines, are gaining popularity in science and industry due to the possibility for local grid refinement. We present a novel rendering algorithm for those surface types that guarantees pixel-accurate geometry and water-tight tessellation (no drop-outs). Before rendering, we extract the Bezier coefficients. The resulting irregular grids of Bezier patches are then rendered using a multistage algorithm, that decouples the tesselator and the patch geometry. The implementation using OpenGL utilizes compute shaders and hardware tessellation functionality. We showcase interactive rendering achieved by our approach on three representative use cases.

Flexible integration of cloud-based engineering services using semantic technologies

Cloud-based engineering services and applications can enable a more flexible, dynamic and integrated usage of software solutions such as CAD, CAM, CFD and PLM. On-demand availability of extensive computing power can lead to significantly improved products, especially for small- and medium-sized companies. In addition to these technical advantages, offering such integrated solutions in the Cloud opens the door to new business models for the software vendors, who can e.g. sell their software on a pay-per-use basis, thus attracting more possible customers. However, to exploit these benefits, such Cloud-based services must be easily accessible and combinable with minimal manual effort - even if they are provided by different software vendors. In this work, we propose a workflow (which is typically a chain of single services) description and execution system based on semantic technologies which abstracts from vendor-specific interfaces and thus allows an effective usage of Cloud-based engineering services in common workflows.