Development tools are important in all stages of the embedded design process, and research is no exception.
Limited budgets can create extra challenges for universities, and the lack of inexpensive, but high-quality tools is a challenge faced by many researchers, including Dr. Luciano Ost, University of Leicester Lecturer, Embedded Systems and Communications Research Group.
“The lack of electronic design automation (EDA) tools combining model flexibility, and fast and accurate evaluation of performance, power, and reliability is one of the major challenges currently faced by embedded researchers,” Ost says, adding that even expensive, commercially available tools don’t often meet modeling and simulation needs for emerging technologies.
Free, open source tools – especially those available to the research community – can be highly beneficial for students learning the embedded ropes and for researchers looking to advance the field. When companies and universities work together to share tools and knowledge in an open source format, good things can happen for the future of embedded computing.
Imperas provides tools and solutions for embedded software development, debug and test, and has been working with universities around the world with its Imperas University Program, which provides academic and research institutions access to tools and virtual platform models.
“The Imperas University Program encourages participation in the embedded systems community in three ways: use on research projects, use in the classroom, and sharing of virtual platform models through the Open Virtual Platforms (OVP) Library,” says Duncan Graham, University Program Manager at Imperas.
The OVP initiative provides researchers access to a library of more than 150 CPU models and more than 200 peripheral and platform models, as well as Extendable Platform Kits (EPKs) – all open source and distributed using the Apache open source license. Libraries like this are very helpful to researchers like Ost who are working with the cutting edge of processing technology.
“The description of processors – i.e., register or gate-level – is rarely available to universities, and commercial licenses are quite expensive,” Ost says. “Having free tools with different state-of-the-art processor models allows the exploration of new system architectures.”
Imperas launched OVP in 2008, providing free access for academic users to the model libraries in addition to APIs, the OVPsim simulator, and to the OVP Forum for technical questions and discussions for academic users. Users can also share their models, platforms, and tools they develop.
In 2010, the Imperas University Program formalized and expanded the OVP program, providing access to Imperas’ Multiprocessor/Multicore Software Design Kits (M*SDK) and QuantumLeap Parallel Simulation Accelerator tools.
Ost has found the tools provided through the Imperas University Program to be helpful for his research. He and his research team focus on multi- and many-core embedded systems, specifically runtime management techniques for performance, energy efficiency, and reliability improvement that must be conducted at the application, operating system, or architectural level. He uses system modeling tools and languages like SystemC as well as virtual platform simulation tools like Imperas’ OVPsim, which help Ost and his team work with increasingly complex processors and software architectures.
“The high-speed simulation and debugging capabilities of Imperas tools facilitate us to extend the popular FreeRTOS to support novel runtime techniques as well as to promote instruction-driven performance and power models, which were incorporated within the original OVPsim distribution,” Ost says.
Ost also notes that free tools also facilitate collaborative projects between universities and external industry partners. In collaboration with the Federal University of Rio Grande do Sul (UFRGS) in Brazil, a fast and flexible fault injector framework, called OVPSim-FIM, was developed on the basis of OVPsim and presented at DFT 2015. OVPSim-FM enables users to identify and critical soft errors in multi- and many-core systems at an early design phase.
More than 1,000 worldwide Imperas University Program members have formed a community where researchers and students can get support, leverage each other’s contributions, and work together to create new tools, like Ost and the OVPSim-FIM.
FlexTiles is an independent consortium of member universities, research institutes, and commercial companies which recently collaborated to create another program: the FlexTiles self-adaptive heterogeneous multicore 3D system-on-chip (SoC) architecture.
“The FlexTiles virtual platform for the FlexTiles adaptive multicore SoC program is based on Imperas and OVP simulators and models,” Graham says.
“The FlexTiles project has developed a high-performance, energy-efficient, programmable many-core platform with self-adaptive capabilities, along with an innovative virtualization layer and a dedicated tool flow. The project successfully used the Imperas simulation and tool technologies, with OVP models, to create a many-core virtual platform, which is available at no cost from Karlsruhe Institute of Technology (KIT), and from the OVP Library page for KIT.”
FlexTiles is positioned to help other researchers by providing a standard, adaptive architecture that can be used as a jumping off point for further development, especially for Internet of Things (IoT) applications.
Other work to come out of the Imperas University Program focuses on many-core architectures, performance and power estimation tools, and techniques for security and safety-critical systems.
“These are all critical issues in the embedded systems community, and advancing the state of the art in any of those areas is beneficial,” Graham says.
Graham hopes the next step for the Imperas University Program involves more collaboration with researchers.
“Imperas is currently looking at taking a more proactive role in working with the academic users, even a collaborative role, in several research projects in the user community,” Graham says. “We are hopeful that our deep knowledge of our tools and technology working in tandem with researchers can more quickly achieve results that push forward the leading edge of technology.”