本文選題：離散元　切入點(diǎn)：顆粒　出處：《大連理工大學(xué)》2013年碩士論文

【摘要】：離散元方法(Discrete Element Method)是將計(jì)算的對(duì)象離散成大量相互獨(dú)立的單元,并定義單元間相互作用的方法。通過離散元的方法可以解決很多現(xiàn)實(shí)中的問題,如模擬泥石流、滑坡等地質(zhì)災(zāi)害,對(duì)于傳統(tǒng)的計(jì)算機(jī)系統(tǒng),在解決DEM問題上存在著局限性。尤其是在大規(guī)模DEM計(jì)算上,近些年發(fā)展緩慢。 本文是基于GPU實(shí)現(xiàn)大規(guī)模DEM的并行計(jì)算。目前,并行計(jì)算是提高算法性能的主要方式,無論是大規(guī)模集群系統(tǒng)還是個(gè)人電腦,甚至是智能手機(jī),都將性能的提升寄托于并行計(jì)算。曾主要用于計(jì)算機(jī)可視化和圖形圖像處理的GPU芯片,內(nèi)部具有大量的并行計(jì)算單元,有著天然的并行優(yōu)勢(shì)。 并行計(jì)算在大規(guī)模離散元計(jì)算方面有著重要的意義。本文的并行計(jì)算過程中是基于大規(guī)模的顆粒離散元基于GPU平臺(tái),通過CUDA編程工具,完成了包括顆粒與三角形邊界面碰撞檢測(cè)和數(shù)據(jù)更新的并行判斷,整個(gè)系統(tǒng)的實(shí)現(xiàn)是由CPU和GPU來實(shí)現(xiàn)的,對(duì)于計(jì)算前的數(shù)據(jù)準(zhǔn)備和數(shù)據(jù)的輸出都是有CPU來完成的,其中主要的接觸判斷和力學(xué)計(jì)算是由GPU來完成。顆粒數(shù)量可達(dá)到百萬級(jí)別的,通過與串行系統(tǒng)的比較,計(jì)算出加速比,通過顆粒系統(tǒng)的動(dòng)態(tài)顯示系統(tǒng)軟件對(duì)系統(tǒng)計(jì)算的結(jié)果進(jìn)行了顯示。 本文通過多種方法進(jìn)行對(duì)并行計(jì)算系統(tǒng)進(jìn)行了優(yōu)化,通過使用NVIDIA Profiler對(duì)并行計(jì)算的三個(gè)主要的內(nèi)核函數(shù)進(jìn)行系統(tǒng)的分析,針對(duì)不同結(jié)構(gòu)的函數(shù)使用不同的優(yōu)化方式,在顆粒與顆粒的接觸判斷核函數(shù)中,鄰居的搜索是關(guān)鍵,通過對(duì)網(wǎng)格法的計(jì)算,確定出最優(yōu)的并行模式下網(wǎng)格劃分,優(yōu)化后的數(shù)據(jù)結(jié)構(gòu)適用于數(shù)據(jù)的合并訪存,并且對(duì)共享存儲(chǔ)模式下的訪問沖突做出了優(yōu)化。在力學(xué)算法上的優(yōu)化,減少內(nèi)部分支,并分別對(duì)優(yōu)化前和優(yōu)化后的結(jié)果進(jìn)行性能上的分析,在優(yōu)化比上顯示出優(yōu)化后的整個(gè)程序計(jì)算性能的穩(wěn)定性和隨著數(shù)目的增多優(yōu)勢(shì)越明顯的特性,進(jìn)一步證明GPU離散元計(jì)算的有效性、穩(wěn)定性、高效性。
[Abstract]:Discrete Element method is a method to discretize computing objects into a large number of independent units and to define the interaction between units.The discrete element method can solve many practical problems, such as simulating debris flow, landslide and other geological disasters. For the traditional computer system, there are some limitations in solving the DEM problem.Especially in large-scale DEM computing, the development is slow in recent years.This paper implements parallel computing of large scale DEM based on GPU.At present, parallel computing is the main way to improve the performance of the algorithm. Both large-scale cluster systems, personal computers, and even smart phones, all rely on parallel computing to improve their performance.The GPU chip, which was mainly used in computer visualization and graphic image processing, has a large number of parallel computing units, which has the natural advantages of parallelism.Parallel computing plays an important role in large-scale discrete element computation.In this paper, the parallel computing process is based on the large-scale particle discrete element based on the GPU platform, through CUDA programming tools, including particle and triangular boundary plane collision detection and data update of the parallel judgment.The implementation of the whole system is implemented by CPU and GPU, and the data preparation and data output before calculation are accomplished by CPU, in which the main contact judgment and mechanical calculation are accomplished by GPU.If the number of particles can reach the level of millions, the speedup ratio is calculated by comparing with the serial system, and the results of the system calculation are displayed by the dynamic display system software of the particle system.In this paper, the parallel computing system is optimized by a variety of methods. By using NVIDIA Profiler, the three main kernel functions of parallel computing are systematically analyzed, and different optimization methods are used for the functions with different structures.In the kernel function of particle contact judgment, neighbor search is the key. Through the calculation of grid method, the optimal grid division in parallel mode is determined. The optimized data structure is suitable for data merging and storage.And the access conflicts in the shared storage mode are optimized.Optimization in mechanics algorithm, reducing internal branch, and performance analysis of the results before and after optimization,The optimization ratio shows the stability of the whole program and the more obvious advantages with the increase of the number, which further proves the effectiveness, stability and efficiency of the GPU discrete element calculation.
【學(xué)位授予單位】：大連理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2013
【分類號(hào)】：TP338.6

【參考文獻(xiàn)】

相關(guān)期刊論文 前8條

1 廖秋林;李曉;郝釗;王思敬;吳碩;赫建明;;土石混合體的研究現(xiàn)狀及研究展望[J];工程地質(zhì)學(xué)報(bào);2006年06期

2 唐德翔,葛蔚,王小偉,麻景森,郭力,李靜海;顆粒流體系統(tǒng)宏觀擬顆粒模擬的并行算法[J];中國(guó)科學(xué)(B輯 化學(xué));2004年03期

3 陳飛國(guó);葛蔚;李靜海;;復(fù)雜多相流動(dòng)分子動(dòng)力學(xué)模擬在GPU上的實(shí)現(xiàn)[J];中國(guó)科學(xué)(B輯:化學(xué));2008年12期

4 王小偉,郭力,楊章遠(yuǎn);N-body算法及其并行化[J];計(jì)算機(jī)與應(yīng)用化學(xué);2003年Z1期

5 王健;許明;葛蔚;李靜海;;單相流動(dòng)數(shù)值模擬的SIMPLE算法在GPU上的實(shí)現(xiàn)[J];科學(xué)通報(bào);2010年20期

6 孫其誠(chéng);程曉輝;季順迎;金峰;;巖土類顆粒物質(zhì)宏-細(xì)觀力學(xué)研究進(jìn)展[J];力學(xué)進(jìn)展;2011年03期

7 黃綿松;安雪暉;;顆粒離散元的HACell檢索算法用于SCC模擬[J];清華大學(xué)學(xué)報(bào)(自然科學(xué)版);2010年09期

8 包南森;李正杰;柴亞輝;徐煒民;;鄰居搜索問題在CUDA上基于KD-TRIE方法的優(yōu)化與實(shí)現(xiàn)[J];上海大學(xué)學(xué)報(bào)(自然科學(xué)版);2012年03期

，

本文編號(hào)：1706873