ID |
原文 |
译文 |
52847 |
直接按照文件数据的存储粒度(如4 kB)在应用和文件系统之间传输数据。 |
and they directly transfer data between the application and file system according to the storage size of the file data(4 kB). |
52848 |
然而,文件系统中的数据读写性能仍然受到操作粒度的影响。 |
However, the data read and write performance in the file system is still affected by the granularity of operations. |
52849 |
本文分析了文件系统在真实NVMM硬件上的性能,发现大粒度操作会降低文件系统性能,并针对数据读写操作粒度提出了优化策略。 |
The performance of file systems on real NVMM hardware is analyzed, it is found that large granularity operations reduce the performance of the file system, and optimization strategies for data operation granularity are proposed. |
52850 |
实验结果表明,本文提出的优化策略使得NVMM文件系统性能提升30.1%。 |
The experimental results show that these strategies can improve the performance of NVMM file system by 30. 1%. |
52851 |
智能天车倒垛优化是提高钢卷库堆场利用率的重要手段,同时对提升钢铁仓库物流效率具有重要意义。 |
The optimization of smart crane shuffle operations is an important means to improve the utilization rate of the steel roll yard, and it is significance to promote the logistics efficiency of the steel roll warehouse to address this problem. |
52852 |
针对该问题,建立最小倒垛次数为目标的天车作业负荷数学模型。 |
This paper develops a mathematical model of crane operation load with the goal of minimizing the number of shuffle operations. |
52853 |
在对模型求解过程中,借鉴了 Alpha Go-Zero中树搜索方法,设计了蒙特卡洛钢卷搜索树(MCRST)。 |
In the process of solving the model, using the tree search method in Alpha Go-Zero for reference, this paper designs a Monte Carlo roll search tree(MCRST). |
52854 |
为了提升搜索树的收敛速度和结果的准确性,将树的置信度上界(UCT)改为快速动作值估计(RAVE), |
In order to improve the convergence speed and accuracy of search tree results, the upper confidence bounds for trees(UCT) is changed to rapid action value estimation(RAVE). |
52855 |
同时引入绝对剪枝策略避免节点盲目扩展。 |
Meanwhile, the algorithm introduces absolute pruning strategy to avoid blind expansion of nodes. |
52856 |
通过不同规模算例实验,将改进算法与原树搜索和粒子群算法(PSO)进行比较,证明了该算法在大规模问题上的优越性; |
Through experiments of different scales, the improved algorithm is compared with the original. tree search and particle swarm optimization(PSO), which proves the superiority of the algorithm in large-scale problems. |