UC知道计算机专业英语翻译不会？第三段，会的拜托帮忙解决下
来源：百度知道 编辑：UC知道 时间：2024/06/09 04:48:30
The second experiment generated the same result as the first experiment, but the result of the third experiment was different from those of the previous two experiments. All
values increased drastically. Two possible reasons can be
supposed, 1) data transfers between the client and servers
exceeds the capacity of the network, i.e. the network itself
becomes a bottleneck, and 2) computation on the client exceeds
the capacity of the client machine, i.e. the client machine
becomes a bottleneck. In order to investigate the reason,
we carried out an additional experiment. In the additional
experiment, we ran two client programs on the KOUME cluster. Each client program ran 750 sample simulations on the PrestoIII cluster (100 processors), the UME cluster (25 processors), and the Venus cluster (25 processors).The results are shown in the bottom in Table 3.
The performance was drastically improved when compared
to the result of the third exper
values increased drastically. Two possible reasons can be
supposed, 1) data transfers between the client and servers
exceeds the capacity of the network, i.e. the network itself
becomes a bottleneck, and 2) computation on the client exceeds
the capacity of the client machine, i.e. the client machine
becomes a bottleneck. In order to investigate the reason,
we carried out an additional experiment. In the additional
experiment, we ran two client programs on the KOUME cluster. Each client program ran 750 sample simulations on the PrestoIII cluster (100 processors), the UME cluster (25 processors), and the Venus cluster (25 processors).The results are shown in the bottom in Table 3.
The performance was drastically improved when compared
to the result of the third exper
第二实验产生那相同的结果当做那第一实验, 但是那结果那第三实验不同于早先人的二实验。 所有的
价值增加的大幅。 二可能的理由能是
假如, 1) 客户和伺候器之间的数据移动
超过网络的能力, 也就是网络本身
变成一个瓶颈, 和 2) 计算在客户上超过
客户机器的能力,也就是客户机器
变成一个瓶颈。 为了要调查那说服,
我们被运的外面一另外的实验。 在那另外的
实验, 我们在 KOUME 上进行二个客户计画群。 每位客户计画在 PrestoIII 上进行 750个样品模拟群 (100个处理器), UME 群 (25个处理器), 和维那斯群 (25个处理器).结果被显示在那底部在表 3.
表现大幅改良的当比较
对第三实验的结果。 结果表演
那一个瓶颈是由于客户机器的能力。
这指出它是可能的到奔跑大的刻度
parallelizing 的模拟客户计画。 有
多数可能性为像 MPI 的使用， OpenMP 这样的 parallelization, 多线的, 和阶层的 RPC。