Hongtao Bai

MAX-MIN Ant System on GPU with CUDA

We propose a parallel MAX-MIN Ant System (MMAS) algorithm that is suitable for an implementation on graphics processing units (GPUs). Multi ant colonies with respective parameter settings are whole offloaded to the GPU in parallel. We have implemented this GPU-based MMAS on the GPU with compute unified device architecture (CUDA). Some performance optimization means for kernel program of GPU are introduced. Experimental results that are based on simulations for the traveling salesperson problem are presented to evaluate the proposed techniques.

A parallel FP-growth algorithm on World Ocean Atlas data with multi-core CPU

Abstract According to the complexity of ocean data, this paper adopts a parallel mining algorithm of association rules to explore the correlation and regularity of oxygen, temperature, phosphate, nitrate and silicate in the ocean. After the marine data is interpolated, this paper utilizes the parallel FP-growth algorithm to mine the data and then briefly analyzes the mining results of the frequent itemsets and association rules. The relationship between the parallel efficiency and the core number of CPU is analyzed through datasets with different scales. The experimental results indicate that the acceleration effect is ideal when each thread scored 200,000–300,000 data, which leads to more than 1.2 times of performance improvement.

Revised simplex algorithm for linear programming on GPUs with CUDA

The revised simplex algorithm (RSA) is a typical algorithm for solving linear programming problems. Many theoretical modifications have been done to make the algorithm more efficient, but almost all of them were based on single-instruction single-data architecture processors (CPUs), which could not make full use of the inherent parallel characteristics of RSAs. We propose a novel single-instruction multiple-data architecture processor (GPU) based on the RSA in this paper. The intensive matrix manipulations of a traditional RSA are offloaded to the GPU, which helps to make full use of its powerful parallel processing ability. We implemented the GPU-based RSA on compute unified device architecture (CUDA). Numerical experiments on randomly generated linear programs show that the GPU-based RSA can not only find the correct optimal solutions, but can also reach a speed of up to 100 times as fast as that of a CPU-based RSA: it also runs 3 to 11 times as fast as MATLAB.

Construction of High Resolution Thermocline Grid Data Sets

Thermocline has always been the emphasis of marine research. In this paper, we propose a method to construct high resolution marine grid data sets on the basis of MLP. Data used in the article is from World Ocean Atlas 2013. The experiments show that high resolution data can calculate the depth, thickness and strength of thermocline precisely. The method is vital to thermocline gridding.

Linear Programming Computation Model Based on DPVM

Matrix manipulation of Linear Programming (LP) problems is a performance bottleneck in Single Instruction Single Data (SIMD) pattern. While, GPU is specialized for this compute-intensive and highly parallel computation, which is exactly what graphics rendering is about, due to the Single Instruction Multiple Data (SIMD) architecture. This paper introduces a Revised Simplex Method (RSM) on a GPU–Data Parallel Virtual Machine (DPVM). It assigns different routines for CPU and GPU according to respective characteristics: Iteration control and minimum value obtained are completed by CPU and Matrix multiplication by DPVM. In detail, we carefully organize the data as 4-channel textures, and efficiently implement the computation using Fetch4 technology of pixel shader. Numerical experiments are presented to verify the practical value and performance of this algorithm. The results are very promising. In particular, they reveal that our method not only can get correct optimal solution, but also is sixty-six faster than a traditional method on CPU, near 2.5 times faster than a lasted released MATLAB when LP problem size reaches 1200.

Coarse-Grained 2.5-D CSAMT Parallel Inversion Method Based on Multi-core CPU

In this paper, a 2.5-D controlled source audio-frequency magnetotelluric (CSAMT) reversed parallel method based on multi-core CPU is achieved, aiming to the low efficiency of the large-scale computation in geophysical exploration. This parallel algorithm allocates the different waves to each thread of cores under the coarse-gained mode. The master thread deal with all parallel tasks and other slave threads compute the electromagnetic field values of each wave in parallel Fork-join model. The experiments demonstrate that our parallel algorithm can not only acquire the effective data accuracy, but also obtain about three times than the serial version.