首页
学习
活动
专区
工具
TVP
发布
精选内容/技术社群/优惠产品,尽在小程序
立即前往
您找到你想要的搜索结果了吗?
是的
没有找到

python数据结构之quick_sor

Quick sort , also known as partition-exchange sort, divides the data to be sorted into two separate parts by a single sort, in which all the data of one part is smaller than all the other parts. Then, according to this method, the two parts of the data are sorted separately, and the whole sorting process can be recursively, so that the entire data becomes an ordered sequence. The steps are: 1. Pick an element from the series, called "pivot", 2. Reorder the series, all elements are placed in front of the reference smaller than the reference value, and all elements are placed larger than the reference value. Behind the benchmark (the same number can go to either side). After the end of this partition, the benchmark is in the middle of the sequence. This is called a partition operation. 3. Recursively sorts sub-columns that are smaller than the reference value element and sub-columns that are larger than the reference value element. The bottommost case of recursion is that the size of the series is zero or one, that is, it has always been sorted. Although it has been recursively, this algorithm will always end, because in each iteration, it will at least put an element to its final position.

01

P&R | 如何在实现全流程中考虑IR-Drop

随着工艺进步,芯片上的线宽越来越窄,单位电阻也越来越大,而同时设计的复杂度也越来越高,芯片尺寸非但没有减小还长得更大了,以至于绕线越来密集,这对电源完整性提出了新的挑战,维基百科上对电源完整性的定义是:Power integrity: or PI is the analysis to check whether the desired voltage and current are met from source to destination. 对应于数字实现就是IR 跟EM 的分析。对于老工艺,IR 跟EM 在设计末期去修干净即可,但到了新工艺点,如果把IR 跟EM 留到最后再看,是在玩火,因为极可能修不掉需要重头再来。所以亟需在实现早期就去考虑PI, 此处就需要了解一下C 记的IR-Aware 全家桶 —— Innovus + PVS + voltus + Tempus —— 从IR-Aware Placement 到 IR-Aware CCOPT 到reinforce_pg 到Tempus ECO IR drop fixing 到PVS TBF-PG 到Power Grid Optimization.

04

tf.train.batch

在张量中创建多个张量。参数张量可以是张量的列表或字典。函数返回的值与张量的类型相同。这个函数是使用队列实现的。队列的QueueRunner被添加到当前图的QUEUE_RUNNER集合中。 如果enqueue_many为False,则假定张量表示单个示例。一个形状为[x, y, z]的输入张量将作为一个形状为[batch_size, x, y, z]的张量输出。如果enqueue_many为真,则假定张量表示一批实例,其中第一个维度由实例索引,并且张量的所有成员在第一个维度中的大小应该相同。如果一个输入张量是shape [*, x, y, z],那么输出就是shape [batch_size, x, y, z]。容量参数控制允许预取多长时间来增长队列。返回的操作是一个dequeue操作,将抛出tf.errors。如果输入队列已耗尽,则OutOfRangeError。如果该操作正在提供另一个输入队列,则其队列运行器将捕获此异常,但是,如果在主线程中使用该操作,则由您自己负责捕获此异常。

01
领券