Special Sessions

Share it now


FDL 2016
Forum on specification & Design Languages

September 14-16, 2016
Bremen, Germany
 
FDL 2016 will feature several Special Sessions with the purpose of introducing the FDL community to relevant hot topics that were not covered by previous editions of the conference.

Designing Reliable Cyber-Physical Systems
Chair : Goerschwin Fey - German Aerospace Center
 
Abstract :
Cyber-Physical Systems (CPS), that consist of a cyber part – a computing system – and a physical part – the system in the physical environment, are omnipresent in our daily lives. The application in the physical environment drives the overall requirements that must be respected when designing the computing system. Here, reliability is a core aspect to ensure dependable operation where some of the most pressing design challenges are
• ensuring correctness of the computing system with respect to application-driven requirements rooted in the physical environment,
• monitoring failures throughout the computing system, and
• determining the impact of failures on the application constraints.
The tutorial proposes techniques to tackle these challenges throughout the system stack of the computing system while tightly coupling the design methodology to the physical requirements.
 
Resilient Embedded Electronic Systems
Chair : Pablo Sánchez - University of Cantabria
 
Abstract :
Current electronic systems are driven by complexity and heterogeneity: complexity as they are based on many functions working together that must be designed in a very short time, with limited resources and facing first-attempt success as a major goal; heterogeneity as they contain not only custom hardware (mainly implemented on FPGA), but many microprocessor units running embedded software and communications that are attached to the system, resulting in networked embedded elements.
 
These systems have opened up new application fields based on the possibilities offered by new devices, communication mechanisms, application algorithms and everyday smarter systems. For example, in the Smart City context these systems bring up important challenges in different areas related with the sustainable development of the city and the provision of services to citizens. Among these areas, urban security is particularly important specially when considered for large public spaces and/or under the celebration of major events. Despite the fact that under these circumstances specific and ad-hoc infrastructures are commonly deployed, recent advances enable enhanced functionalities, able to adapt systems to changing needs and to inter-operate naturally and collaboratively with each other and, of course, with the already existing infrastructure. These resilient systems could modify their normal behaviour to face specific situation such as major events, urban security risks (e.g. terrorism attacks), natural disasters or unusual urban situations. In these situations, image-based HW/SW systems could play an essential role.
 
The session will explore design methodologies, use of languages and tools that support resilient HW/SW systems and image-based applications.
 
Reliability and Safety in VP-based Embedded System Development
Chair : Daniel Große - University of Bremen
 
Abstract :
Although embedded systems have witnessed a reduction of their development time in the past decades, their complexity has been increasing steadily. Hence, it has become a necessity for their development to raise the level of abstraction beyond the traditional Register Transfer Level (RTL). By abstracting out the hardware details and following the concepts of Transaction Level Modeling (TLM) Virtual Prototypes (VPs) emerged for the system development.
 
Today VPs are heavily used in industry for early SW development, verification/validation, and design space exploration. However, once in the field embedded systems also have to preserve a well-defined level of reliability and fault-tolerance to ensure operational safety such as it is prescribed by several industry standards like ISO26262, IEC 60730, and DO-254. This special session focusses in particular on design and verification solutions to ensure high reliability and safety already at the system-level avoiding costly iterations due to unnecessary actions in subsequent development steps.
 
Tools and techniques – Compiler support for multi-core/many-core architectures as well as GPUs, Accelerators in heterogeneous computing platforms
Chair : Nicola Bombieri - University of Verona
 
Abstract :
Workload partitioning is a key aspect to face when implementing any efficient application for heterogeneous computing platforms. While it has been largely explored in the context of multi-core processing elements, it has returned subject of investigation since Graphic Processing Units (GPUs) have spread as accelerators in such platforms. Indeed, application developers have now to address workload partitioning at different levels: At GPU level and at platform level. Several and different techniques have been proposed to deal with workload partitioning and distribution GPU level. They range from the computationally simplest static to the most complex dynamic ones. Each of them finds the best use depending on the workload characteristics (static for more regular workloads, dynamic for irregular workloads). Nevertheless, no one of them provides a sound trade-off when applied in both cases. Static approaches lead to load unbalancing with irregular problems, while the computational overhead introduced by the dynamic or semi-dynamic approaches often worsens the overall application performance when run on regular problems. The first contribution of the session is an efficient dynamic technique for workload partitioning and work item-to-thread mapping whose complexity is sensibly reduced with respect to the other dynamic approaches in literature. The paper shows how the partitioning and mapping algorithm has been implemented by fully taking advantage of the GPU device characteristics with the aim of minimizing the involved computational overhead. The paper shows, compares, and analyses the experimental results obtained by applying the proposed approach and several static, dynamic, and semi-dynamic techniques at the state of the art to different benchmarks and over different GPU technologies (i.e., NVIDIA Maxwell GTX 980 device, NVIDIA Jetson Kepler TK1 low-power embedded system) to understand when and how each technique best applies.
 
On the other hand, making use of platform-level heterogeneity still bring additional and significant performance improvements, as it essentially means using additional resources for the same
computation. In this context, one significant question emerges: is the performance gained by using these additional resources worth the effort to program and deploy these applications? The second contribution of the session addresses this question by demonstrating (1) the performance improvement of using platform heterogeneity for different applications, (2) a model-based approach that enables (near-)optimal workload partitioning for different hardware platforms, and (3) a simple way to leverage this model for real applications. The proposed method is based on static partitioning, and can be applied to large variety of data parallel applications. The authors have added a simple diagnosis tool, which evaluates the application and decides whether static partitioning is suitable. If it is, the proposed model is applied and the workload is partitioned. If not, dynamic runtime-based partitioning can be employed. The author demonstrate the proposed solution on several applications, including typical kernels like matrix multiplication, imbalanced
applications such as ray tracing, and irregular applications such as graph processing. The results demonstrate that heterogeneous computing is an asset not only because of the additional resources, but also because these resources are specialized.
 
Finally, the special session deals with runtime task distribution and management in parallel heterogeneous platforms. The impact of process variations in current and next generation technology nodes is becoming relevant and cannot be compensated at the device or architectural level. Intra-die process variations raising at the core level and platform level makes even multicore platforms intrinsically heterogeneous, because the various cores are clocked at different operational frequencies. Power consumption becomes heterogeneous too, both considering dynamic and leakage consumption. In this context, to fully exploit the computational capability of the platform parallelism, variability aware task allocation strategies must be developed. The third contribution of the special session discusses on techniques to perform task allocation, by showing how they can be integrated in a software toolchain.
 
System Performance Modeling & Analysis Based on Extra-Functional Properties
Chair : Adam Morawiec - ECSI
 
Abstract :
The integration of heterogeneous electronic systems composed of SW and HW requires not only a proper handling of system functionality, but also an appropriate expression and analysis of various extra-functional properties: timing, energy consumption, thermal behavior, reliability, cost and others as well as performance aspects related to caching, non-determinism, probabilistic effects.
 
The session addresses cross-domain aspects related to the design and verification framework covering methodology, interoperable tools, flows, interfaces and standards that enable formalization, specification, annotation and refinement of functional and extra-functional properties of a system. Special emphasis will be given to formalization and expression of power, temperature, reliability, degradation and aging. Several research and industry efforts address (parts of) the problem. However, there is a need for community-wide cooperation to establish a holistic vision on extra-functional property treatment, and to agree on research and development directions and further on validation of applicable solutions and standardization.
 
The session invites submissions on industrial and scientific work in progress and practical solution and experiences.

Share it now