Processing Unit Platform Research Articles

Accelerating the Lagrangian Particle Tracking in Hydrologic Modeling to Continental‐Scale

AbstractUnprecedented climate change and anthropogenic activities have induced increasing ecohydrological problems, motivating the development of large‐scale hydrologic modeling for solutions. Water age/quality is as important as water quantity for understanding the terrestrial water cycle. However, scientific progress in tracking water parcels at large‐scale with high spatiotemporal resolutions is far behind that in simulating water balance/quantity owing to the lack of powerful modeling tools. EcoSLIM is a particle tracking model working with ParFlow‐CLM that couples integrated surface‐subsurface hydrology with land surface processes. Here, we demonstrate a parallel framework on distributed, multi‐Graphics Processing Unit platforms with Compute Unified Device Architecture‐Aware Message Passing Interface for accelerating EcoSLIM to continental‐scale. In tests from catchment‐, to regional‐, and then to continental‐scale using 25‐million to 1.6‐billion particles, EcoSLIM shows significant speedup and excellent parallel performance. The parallel framework is portable to atmospheric and oceanic particle tracking models, where the parallelization is inadequate, and a standard parallel framework is also absent. The parallelized EcoSLIM is a promising tool to accelerate our understanding of the terrestrial water cycle and the upscaling of subsurface hydrology to Earth System Models.

A Fast Parallel GPS Acquisition Algorithm Based on Hybrid GPU and Multi-core CPU

Over the last decade, to improve the necessary accuracy and reliability of the existing global positioning system (GPS), incorporation with new constellations such as European Galileo navigation system and the update of the Russian Global Navigation Satellite System (GLONASS) attract a lot of attentions. In software integrations, the most important challenge is the calculation load which should be solved using suitable techniques. The parallel nature of this application causes graphic processing unit (GPU) and multi-core platform as a low cost and effective solution for the implementation of a software receiver. In this paper, in order to accelerate GPS acquisition, a hybrid parallel algorithm based on GPU and multi-core CPU is designed and implemented on two systems with GT630 and GT630M GPU. Experimental results show that the proposed algorithm is 2.2 times faster than sequential algorithm in one-core implementation and 4 times faster in quad core. On the other hand, the proposed parallel acquisition algorithm on GPU-based 2-core is 15% faster than acquisition on 4-core without GPU, which shows the effective role of GPU in parallelism of this application.

HIGH SPEED COMPUTING OF ICE THICKNESS EQUATION FOR ICE SHEET MODEL

Two-dimensional (2-D) ice flow thermodynamics coupled model acts as a vital role for visualizing the ice sheet behaviours of the Antarctica region and the climate system. One of the parameters used in this model is ice thickness. Explicit method of finite difference method (FDM) is used to discretize the ice thickness equation. After that, the equation will be performed on Compute Unified Device Architecture (CUDA) programming by using Graphics Processing Unit (GPU) platform. Nowadays, the demand of GPU for solving the computational problem has been increasing due to the low price and high performance computation properties. This paper investigates the performance of GPU hardware supported by the CUDA parallel programming and capable to compute a large sparse complex system of the ice thickness equation of 2D ice flow thermodynamics model using multiple cores simultaneously and efficiently. The parallel performance evaluation (PPE) is evaluated in terms of execution time, speedup, efficiency, effectiveness and temporal performance.

Performance evaluation of a 3D multi-view-based particle filter for visual object tracking using GPUs and multicore CPUs

This paper presents a deep and extensive performance analysis of the particle filter (PF) algorithm for a very compute intensive 3D multi-view visual tracking problem. We compare different implementations and parameter settings of the PF algorithm in a CPU platform taking advantage of the multithreading capabilities of the modern processors and a graphics processing unit (GPU) platform using NVIDIA CUDA computing environment as developing framework. We extend our experimental study to each individual stage of the PF algorithm, and evaluate the quality versus performance trade-off among different ways to design these stages. We have observed that the GPU platform performs better than the multithreaded CPU platform when handling a large number of particles, but we also demonstrate that hybrid CPU/GPU implementations can run almost as fast as only GPU solutions.

FPGA-based Personal Authentication Using Fingerprints

The current technological age demands the deployment of biometric security systems not only in those stringent and highly reliable fields (forensic, government, banking, etc.) but also in a wide range of daily use consumer applications (internet access, border control, health monitoring, mobile phones, laptops, etc.) accessible worldwide to any user. In order to succeed in the exploitation of biometric applications over the world, it is needed to make research on power-efficient and cost-effective computational platforms able to deal with those demanding image and signal operations carried out in the biometric processing. The present work deals with the evaluation of alternative system architectures to those existing PC (personal computers), HPC (high-performance computing) or GPU-based (graphics processing unit) platforms in one specific scenario: the physical implementation of an AFAS (automatic fingerprint-based authentication system) application. The development of automated fingerprint-based personal recognition systems in the way of compute-intensive and real-time embedded systems under SoPC (system-on-programmable-chip) devices featuring one general-purpose MPU (microprocessor unit) and one run-time reconfigurable FPGA (field programmable gate array) proves to be an efficient and cost-effective solution. The provided flexibility, not only in terms of software but also in terms of hardware thanks to the programmability and run-time reconfigurability performance exhibited by the suggested FPGA device, permits to build any application by means of hardware-software co-design techniques. The parallelism and acceleration performances inherent to the hardware design and the ability of reusing hardware resources along the application execution time are key factors to improve the performance of existing systems.

Multi-prediction particle filter for efficient parallelized implementation

Abstract Particle filter (PF) is an emerging signal processing methodology, which can effectively deal with nonlinear and non-Gaussian signals by a sample-based approximation of the state probability density function. The particle generation of the PF is a data-independent procedure and can be implemented in parallel. However, the resampling procedure in the PF is a sequential task in natural and difficult to be parallelized. Based on the Amdahl's law, the sequential portion of a task limits the maximum speed-up of the parallelized implementation. Moreover, large particle number is usually required to obtain an accurate estimation, and the complexity of the resampling procedure is highly related to the number of particles. In this article, we propose a multi-prediction (MP) framework with two selection approaches. The proposed MP framework can reduce the required particle number for target estimation accuracy, and the sequential operation of the resampling can be reduced. Besides, the overhead of the MP framework can be easily compensated by parallel implementation. The proposed MP-PF alleviates the global sequential operation by increasing the local parallel computation. In addition, the MP-PF is very suitable for multi-core graphics processing unit (GPU) platform, which is a popular parallel processing architecture. We give prototypical implementations of the MP-PFs on multi-core GPU platform. For the classic bearing-only tracking experiments, the proposed MP-PF can be 25.1 and 15.3 times faster than the sequential importance resampling-PF with 10,000 and 20,000 particles, respectively. Hence, the proposed MP-PF can enhance the efficiency of the parallelization.

SU-GG-T-255: Development of a Computational Infrastructure for GPU-Based Online Adaptive Radiation Therapy

Purpose: To develop a computational infrastructure to enable a streamlined clinical workflow for online adaptive radiation therapy (ART). Method and Materials: Our computational infrastructure consists of modules for deformable image registration based on the demons algorithm, and fluence map and direct aperture re‐optimization based on finite size pencil beam dose calculation. These modules have been implemented on a graphics processing unit (GPU) platform. Around these highly efficient implementations we have built modules for DICOM file handling, and graphical user interface based on the Qt framework. Furthermore, we have built a communication infrastructure for web‐based plan reviewing and approval using a PDA device such as the iPhone. Results: We measured a net re‐planning computation time of <10 seconds for a 9‐field prostate IMRT plan. A software infrastructure with a clean yet functional user interface has been built to seamlessly guide clinicians through the online re‐planning process. Conclusion: We integrated new computational tools into real clinical environment. We achieved this in two ways. First, by turning prototype high performance codes into a product‐alike software solution to be used by clinicians. Second, by developing a web‐based communication infrastructure to allow faster treatment plan review and approval by attending.