high performance computing on graphics processing units: hgpu.org

Computer vision research enables machines to understand the world. Humans usually interpret and analyze the world through what they see – the objects they capture with their eyes. Similarly, machines can better understand the world by recognizing objects in images. Object recognition is therefore a major branch of computer vision. To achieve the highest accuracy, […]

Radiosity is a popular global illumination algorithm capable of achieving photorealistic rendering results. However, its use in interactive environments is limited by its computational complexity. This paper presents a GPGPU-based implementation of the gathering radiosity approach using texture-based discretisation and the OpenCL framework. Hemicubes are rendered to a texture array and are processed by OpenCL […]

This thesis explores the design and experimental implementation of GPF, a novel protocol-independent, multi-match packet classification framework. This framework is targeted and optimised for flexible, efficient execution on NVIDIA GPU platforms through the CUDA API, but should not be difficult to port to other platforms, such as OpenCL, in the future. GPF was conceived and […]

This paper presents a comparison of parallelization frameworks for efficient execution of computational finance workloads. We use a Value-at-Risk (VaR) workload to evaluate OpenCL and OpenMP parallelization frameworks on multi-core CPUs as opposed to GPUs. In addition, we study the impact of SMT on performance using GCC (4.4) and IBM XLC (11.01) compilers for both […]

In this paper we present an implementation of the H.264/AVC Inverse Discrete Cosine Transform (IDCT) optimized for Graphics Processing Units (GPUs) using OpenCL. By exploiting that most of the input data of the IDCT for real videos are zero valued coefficients a new compacted data representation is created that allows for several optimizations. Experimental evaluations […]

In this paper we present an overview of our implementation of a fluid simulation technique called "Smoothed Particle Hydrodynamics". Our implementation uses a hybrid CPU+GPU hash based data structure to provide quick lookups of particle nearest neighbors and improve memory access patterns.In our discussion we begin with a brief overview of the Navier Stokes equations […]

Mapping algorithms in an efficient way to the target hardware poses a challenge for algorithm designers. This is particular true for heterogeneous systems hosting accelerators like graphics cards. While algorithm developers have profound knowledge of the application domain, they often lack detailed insight into the underlying hardware of accelerators in order to exploit the provided […]

Using GPUs for general purpose programming is, nowadays, much easier than the previous years. In the very beginning were Brook-GPU or Close To Metal the approaches used for exploring the new possibilities of hardware accelerators. After that, CUDA and OpenCL were released. They had been adopted by many programmers due to theirs advantages but, however, […]

When modeling pedestrian movement, it is necessary to find a path to the target point. It is possible to use a distance matrix or derived gradient map for this purpose. Calculations of distance matrix for large areas and multiple targets are very time-consuming. Therefore this article focuses on acceleration of these calculations utilizing Graphics Processing […]

The exploitation of heterogeneous resources is becoming increasingly important for general purpose computing. Unfortunately, heterogeneous systems require much more effort to be programmed than the traditional single or even multi-core computers most programmers are familiar with. Not only new concepts, but also new tools with different restrictions must be learned and applied. Additionally, many of […]

The aim of the paper is to show how to design fast parallel algorithms for linear congruential and lagged Fibonacci pseudorandom numbers generators. The new algorithms employ the divide-and-conquer approach for solving linear recurrence systems and can be easily implemented on GPU-accelerated hybrid systems using CUDA or OpenCL. Numerical experiments performed on a computer system […]

We present a technique for verifying race- and divergencefreedom of GPU kernels that are written in mainstream kernel programming languages such as OpenCL and CUDA. Our approach is founded on a novel formal operational semantics for GPU programming termed synchronous, delayed visibility (SDV) semantics. The SDV semantics provides a precise definition of barrier divergence in […]