A component-based framework for the Cell Broadband Engine

Abstract

With the increasing trend of microprocessor manufacturers to rely on parallelism to increase their products' performance, there is an associated increasing need for simple techniques to leverage this hardware parallelism for good application performance. Unfortunately, many application developers do not have the benefit of long experience in programming parallel and distributed systems. While the filter-stream programming paradigm helps bridge the gap between developers of scientific applications and the performance they need, current and future high-performance multicore processor designs do not have a filter-stream programming library available. This work aims to fill that gap in the software world. This initial DataCutter-Lite implementation defines a powerful, but simple abstraction for carrying out complex computations in a filter-stream model. Additionally, the initial implementation shows that complex architectures such as the Cell Broadband Engine Architecture can make use of the filter-stream model, and give good application performance when doing so.

Generative component-based IP camera design

“…Obermaisser et al [5] showed the development of the GENESYS MPSoC and how it can be used as a base component building block, consisting of a collection of hardware-software (operating system plus application or baremetal/gateware) components connected by a trusted timetriggered network-on-chip, in a variety of application domains. Similar work has been described in [6] and [7].…”

Scheduling streaming applications on a complex multicore platform

“…For steady‐state schedules on the QS 22 to be run, a complex software framework is needed: it has to map tasks on different types of processing elements and to handle all communications. Although there already exists some frameworks dedicated to streaming applications , none of them is able to deal with complex task graphs while allowing to statically select the mapping. Thus, we have decided to develop our own framework.…”

“…The objective is to optimize the throughput of the application: for example, in the case of a video stream, we are looking for a solution that maximizes the number of images processed per time unit.Several streaming solutions have already been developed or adapted for the Cell processor. DataCutter-Lite [19] is an adaptation of the DataCutter framework for the Cell processor, but it is limited to simple streaming applications described as linear chains, so it cannot deal with complex task graphs. StreamIt [18, 31] is a language developed to model streaming applications; a version of the Streamit compiler has been developed for the Cell processor; however, it does not allow the user to specify the mapping of the application and thus to precisely control the application.…”

Scheduling complex streaming applications on the Cell processor

“…If there already exist some frameworks dedicated to streaming applications [12,13], none of them is able to deal with complex task graphs while allowing to statically select the mapping. Thus, we have decided to develop one 1 .…”

“…DataCutterLite [13] is an adaptation of the DataCutter framework for the Cell processor, but it is limited to simple streaming applications described as linear chains, so it cannot deal with complex task graphs. StreamIt [12,2] is a language developed to model streaming applications; a version of the Streamit compiler has been developed for the Cell processor, however it does not allow the user to specify the mapping of the application, and thus to precisely control the application.…”