文件名称：The C11 and C++11 Concurrency Model.pdf
文件大小：2.93MB
文件格式：PDF
更新时间：2021-05-11 04:50:52
C11 and C++11 Concurrency Model
The advent of pervasive concurrency has caused fundamental design changes throughout computer systems. In a bid to offer faster and faster machines, designers had been pro- ducing hardware with ever higher clock frequencies, leading to extreme levels of power dissipation. This approach began to give diminishing returns, and in order to avoid phys- ical limitations while maintaining the rate of increase in performance, processor vendors embraced multi-core designs. Multi-core machines contain several distinct processors that work in concert to complete a task. The individual processors can operate at lower fre- quencies, while collectively possessing computing power that matches or exceeds their single core counterparts. Unfortunately, multi-core processor performance is sensitive to the sort of work they are given: large numbers of wholly independent tasks are ideal, whereas monolithic tasks that cannot be split up are pathological. Most tasks require some communication between cores, and the cost of this communication limits perfor-mance on multi-core systems.