三维片上网络离散量子粒子群布图算法研究
发布时间：2018-12-18 17:27
【摘要】：三维片上网络在多种性能上均优于二维片上网络,已成为研究热点。布图算法直接影响芯片的面积和布线长度,成为三维片上网络优化设计的重要方向。提出一种基于离散粒子群算法的三维片上网络布图优化算法,与之前常使用的模拟退火算法相比,不再使用单一解局部扰动的方式得到整个解空间,该算法采用初始化随机种群并不断迭代的进化方式,具有更优的搜索能力和更快的收敛速度。仿真结果表明,采用该算法选择布图方案可以显著降低微片延迟,节省CPU计算时间,尤其是在IP核数量众多的测试用例和高注入率情况下效果更为明显,如对于ami49测试用例当注入率为100%时,基于离散量子粒子群算法的结果和基于模拟退火算法的结果相比,平均微片延迟减少了20.63%,CPU平均时间减少了69.40%。
[Abstract]:Three-dimensional on-chip network is superior to two-dimensional on-chip network in many performance, so it has become a research hotspot. The layout algorithm directly affects the chip area and wiring length, and becomes an important direction of 3D on-chip network optimization design. In this paper, a new algorithm based on discrete particle swarm optimization (DPSO) is proposed. Compared with the previous simulated annealing algorithm, the entire solution space is obtained by using a single local perturbation method. The algorithm has better searching ability and faster convergence speed by initializing random population and iterating continuously. The simulation results show that the algorithm can significantly reduce the delay of microchip and save the computing time of CPU, especially in the case of large number of test cases and high injection rate of IP cores. For ami49 test cases, when the injection rate is 100, the average microchip delay is reduced by 20.63 and 69.40, compared with the results based on discrete Quantum Particle Swarm Optimization (DQPSO) and simulated annealing (SA).
【作者单位】： 天津工业大学计算机科学与软件学院;
【基金】：国家自然科学基金No.61272006~~
【分类号】：TP18
本文编号：2386221
[Abstract]:Three-dimensional on-chip network is superior to two-dimensional on-chip network in many performance, so it has become a research hotspot. The layout algorithm directly affects the chip area and wiring length, and becomes an important direction of 3D on-chip network optimization design. In this paper, a new algorithm based on discrete particle swarm optimization (DPSO) is proposed. Compared with the previous simulated annealing algorithm, the entire solution space is obtained by using a single local perturbation method. The algorithm has better searching ability and faster convergence speed by initializing random population and iterating continuously. The simulation results show that the algorithm can significantly reduce the delay of microchip and save the computing time of CPU, especially in the case of large number of test cases and high injection rate of IP cores. For ami49 test cases, when the injection rate is 100, the average microchip delay is reduced by 20.63 and 69.40, compared with the results based on discrete Quantum Particle Swarm Optimization (DQPSO) and simulated annealing (SA).
【作者单位】： 天津工业大学计算机科学与软件学院;
【基金】：国家自然科学基金No.61272006~~
【分类号】：TP18
【相似文献】
相关期刊论文 前5条
1 刘斌;常振超;张兴明;闫佳佳;易洪波;;一种基于遗传算法的片上网络电压岛划分方法[J];计算机应用研究;2012年10期
2 董文箫;沈海斌;全励;严晓浪;;延迟约束下的功耗与热感知NoC映射[J];电路与系统学报;2012年05期
3 张玉艳;杨潇;黄国栋;侯金奎;;基于图算法的二元组合文法分析[J];计算机应用;2008年07期
4 龙华;颜学明;;基于人工免疫调节网络的自然抠图算法及应用[J];现代计算机;2013年07期
5 张文杰;胡文_";何星;;基于CIM文件的电力系统SVG图形自动排图算法[J];微型电脑应用;2009年03期
相关硕士学位论文 前2条
1 王俊贤;基于混沌粒子群的三维片上网络映射算法设计与实现[D];西安电子科技大学;2015年
2 刘旭辉;基于遗传算法的片上网络区域分割映射算法研究[D];电子科技大学;2012年
，本文编号：2386221
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2386221.html
