Automated monitoring and debugging of large scale manycore heterogeneous systems

通信和计算基础设施多年来不断发展，变得更加高效、复杂、一体化和网络化。较新的移动设备(包括智能机器人或自动驾驶汽车)和服务器通常在其中央处理单元中包含8个或更多核心。这些系统基于不同的处理器，具有高效的传统中央处理器，但也具有针对图形(具有数千个内核的GPGPU)、网络、信号处理甚至机器学习而优化的协处理器。这些协同处理单元是高度并行的，每个单元通常包含超过80亿个逻辑元件(晶体管)。增加这种复杂性的是对虚拟化的日益依赖，这隐藏了硬件的特殊性，允许应用程序在几个不同的处理器型号上运行，但使性能更难分析。*因此，即使是发起电话、进行Web搜索、路由数据包或显示视频帧等简单操作，也可能涉及多个处理器上的多个并行核心，可能位于多个服务器上。此外，在几秒钟之后，不同的核心和物理服务器可能以不同的方式为相同的操作提供服务。因此，了解这些行动的执行情况变得极其困难，而且严重缺乏实现这一目的的工具。在这个项目中，将扩展多核系统的跟踪、监测、剖析和调试工具，以便有效地从所有层的所有单元提取信息，从硬件到应用程序，并处理大量(数千个)核心。此外，还将开发新的方法和算法，以自动分析提取的监测数据。因此，移动设备、云服务器和其他异类计算系统上的分布式应用程序的设计者和操作员将拥有快速分析其系统性能、自动或手动发现问题并优化操作的工具。*

The communication and computing infrastructure has evolved through the years, getting more efficient, sophisticated, integrated and networked. Newer mobile devices (including smart robots or autonomous cars) and servers often contain 8 or more cores in their central processing unit. These systems are based on heterogeneous processors, with efficient traditional central processing units, but also with co-processing units optimised for graphics (GPGPUs with thousands of cores), networking, signal processing or even for Machine Learning. These co-processing units are highly parallel and often contain over 8 billion logic elements (transistors) each. Adding to this complexity is the increasing reliance on virtualisation, which hides the specificities of the hardware, allowing an application to run on several different processor models, but makes the performance more difficult to analyse. ****As a result, even a simple operation such as initiating a phone call, making a Web search, routing a packet or displaying a video frame, can involve many parallel cores on more than one processing unit, possibly on several servers. Moreover, the same operation, a few seconds later, may be served in a different way by different cores and physical servers. Therefore, understanding the performance of these operations has become extremely difficult and the tools for that purpose are severely lacking. In this project, the tracing, monitoring, profiling and debugging tools for manycore systems will be extended to efficiently extract information from all units in all layers, from the hardware to the application, and cope with the large number (several thousands) of cores. Furthermore, new methods and algorithms will be developed to automate the analysis of the extracted monitoring data. As a result, the designers and operators of distributed applications on mobile devices, cloud servers and other heterogeneous computing systems, will have the tools in hand to quickly analyse their system performance, automatically or manually find problems, and optimise operations.******