Search

The domain of energy-constrained real-time systems that are operated on modern embedded system-on-chip (SoC) platforms brings numerous novel challenges for optimal resource minimization. These modern hardware platforms offer a heterogeneous variety of features to configure the tradeoff between temporal performance and energy efficiency, which goes beyond the state-of-the-art of existing dynamic-voltage-frequency-scaling (DVFS) scheduling schemes. The control center for configuring this tradeoff on platforms are complex clock subsystems that are intertwined with requirements of the SoC’s components (e.g., transceiver/memory/sensor devices). That is, several devices have precedence constraints with respect to specific clock sources and their settings. The challenge of dynamically adapting the various clock sources to select resource-optimal configurations becomes especially challenging in the presence of asynchronous preemptions, which are inherent to systems that use devices. In this paper, we present Crêpe, an approach to clock-reconfiguration-aware preemption control: Crêpe has an understanding of the target platform’s clock subsystem, its sleep states, and penalties to reconfigure clock sources for adapting clock frequencies. Crêpe’s hardware model is combined with an awareness of the application’s device requirements for each executed task, as well as possible interrupts that cause preemptions during runtime. Using these software/hardware constraints, Crêpe employs, in its offline phase, a mathematical formalization in order to select energy-minimal configurations while meeting given deadlines. This optimizing formalization, processed by standard mathematical solver tools, accounts for potentially occurring interrupts and the respective clock reconfigurations, which are then forwarded as alternative schedules to Crêpe’s runtime system. During runtime, the dispatcher assesses these offline-determined alternative schedules and reconfigures the clock sources for ene

With the emergence of embedded system-on-chip (SoC) platforms, the development of energy-constrained real-time systems brings numerous novel challenges for optimal resource consumption. On these modern hardware platforms, complex clock subsystems make it possible to tradeoff between temporal performance and energy efficiency by reconfiguring the system, which exceeds the state-of-the-art of existing dynamic-voltage-frequency-scaling (DVFS) scheduling schemes. On embedded real-time systems, the usage of the devices (e.g., transceiver/memory/sensor devices) is an essential component to be able to interact with the surrounding world. Each device has precedence constraints with respect to specific clock sources and their settings. Therefore, to select resource-optimal configurations, we need to adapt the clock subsystem, which becomes especially challenging in the presence of asynchronous preemptions, often found during device interaction. This artifact evaluation covers the work of Crêpe, an approach to clock-reconfiguration-aware preemption control on systems with devices. Crêpe makes use of the target platform’s clock subsystem, possible idle modes, and the reconfiguration penalties for adapting the clock subsystem. By combining a hardware model for the device under investigation with an awareness of the required clock configuration for each task, as well as possible interrupts causing preemptions during runtime, Crêpe employs a mathematical formalization to determine energy-minimal configuration sequences while meeting all given deadlines. Before runtime, Crêpe solves the mathematical problem with standard mathematical solver tools and generates optimal execution strategies and clock-system reconfigurations before runtime. These offline-generated schedules are then assessed by the dispatcher during runtime, leading to an overall minimized energy consumption with minimal overhead during execution. Crêpe also consists of an implementation based on a widely-used SoC pl

Numerous embedded real-time systems have, besides their worst-case execution time (WCET) requirements, strict worst-case energy consumption (WCEC) constraints that must be satisfied. The core hardware component of modern system-on-chip (SoC) platforms to configure the tradeoff between time and energy is the system’s clock tree, which provides the necessary clock source for each connected device (i.e., memory, sensors, transceivers). Existing energy-aware scheduling approaches have limitations with regard to these modern, feature-rich clock trees: These shortcomings concern the (re-)configuration of the clock tree with the associated penalties, which are a non-negligible part of dynamic frequency/voltage scaling or power-gating devices in addition to the influence of available sleep modes. This artifact evaluation covers the work on FusionClock, an approach that exploits a fine-grained model of the system’s temporal and energetic behavior. By means of our developed clock-tree model, FusionClock processes time-triggered schedules and finally generates optimized code for a system where offline-determined and online-applied reconfigurations lead to the worst-case-optimal energy demand while still meeting given timing-related deadlines. For statically determining these energy-optimal reconfigurations on task level, FusionClock builds a mathematical optimization problem based on the tasks' specifications and the system’s resource-consumption model. Specific components like transceivers of SoCs usually have strict requirements regarding the used clock source (e.g., phase-locked loop, RC network, oscillator). FusionClock accounts for these clock-tree requirements with its ability to exploit application-specific knowledge within an optimization problem. With our resource-consumption model for a modern SoC platform and our open-source prototype of FusionClock, we are able to achieve significant energy savings while still providing guarantees for timeliness, as our evaluations

Numerous embedded real-time systems have, besides their timing requirements, strict energy constraints that must be satisfied. Examples of this class of real-time systems are implantable medical devices, where knowledge of the worst-case execution time (WCET) has the same importance as of the worst-case energy consumption (WCEC) in order to provide runtime guarantees. The core hardware component of modern system-on-chip (SoC) platforms to configure the tradeoff between time and energy is the system’s clock tree, which provides the necessary clock source to all connected devices (i.e., memory, sensors, transceivers). Existing energy-aware scheduling approaches have shortcomings with regard to these modern, feature-rich clock trees: First, with their reactive, dynamic (re-)configuration of the clock tree, they are not able to provide static guarantees of the system’s resource consumption (i.e., energy and time). Second, they only account for dynamic voltage/frequency scaling of the CPU and thereby miss the reconfiguration of clock sources and clock speed for the other connected devices on such SoCs. Third, they neglect the reconfiguration penalties of frequency scaling and clock/power gating in the presence of the CPU’s sleep modes. In this paper, we present FusionClock, an approach that exploits a fine-grained model of the system’s temporal and energetic behavior. By means of our developed clock-tree model, FusionClock processes time-triggered schedules and finally generates optimized code for a system where offline-determined and online-applied reconfigurations lead to the worst-case-optimal energy demand while still meeting given timing-related deadlines. For statically determining these energy-optimal reconfigurations on task level, FusionClock builds a mathematical optimization problem based on the tasks' specifications and the system’s resource-consumption model. Specific components like transceivers of SoCs usually have strict requirements regarding the used cl

OASIcs, Volume 114, WCET 2023, Complete Volume

Front Matter, Table of Contents, Preface, Conference Organization

Many energy-constrained cyber-physical systems require both timeliness and the execution of tasks within given energy budgets. That is, besides knowledge on worst-case execution time (WCET), the worst-case energy consumption (WCEC) of operations is essential. Unfortunately, WCET analysis approaches are not directly applicable for deriving WCEC bounds in device-driven cyber-physical systems: For example, a single memory operation can lead to a significant power-consumption increase when thereby switching on a device (e.g. transceiver, actuator) in the embedded system. However, as we demonstrate in this paper, existing approaches from microarchitecture-aware timing analysis (i.e. considering cache and pipeline effects) are beneficial for determining WCEC bounds: We extended our framework on whole-system analysis with microarchitecture-aware timing modeling to precisely account for the execution time that devices are kept (in)active. Our evaluations based on a benchmark generator, which is able to output benchmarks with known baselines (i.e. actual WCET and actual WCEC), and an ARM Cortex-M4 platform validate that the approach significantly reduces analysis pessimism in whole-system WCEC analyses.

Implementation-based benchmarking of timing and schedulability analyses requires system code that can be executed on real hardware and has defined properties, for example, known worst-case execution times (WCETs) of tasks. Traditional approaches for creating benchmarks with such characteristics often result in implementations that do not resemble real-world systems, either due to work only being simulated by means of busy waiting, or because tasks have no control-flow dependencies between each other. In this paper, we address this problem with TASKers, a generator that constructs realistic benchmark systems with predefined properties. To achieve this, TASKers composes patterns of real-world programs to generate tasks that produce known outputs and exhibit preconfigured WCETs when being executed with certain inputs. Using this knowledge during the generation process, TASKers is able to specifically introduce inter-task control-flow dependencies by mapping the output of one task to the input of another.

Although internal devices (e.g., memory, timers) and external devices (e.g., sensors, transceivers) significantly contribute to the energy consumption of an embedded real-time system, their impact on the worst-case response energy consumption (WCRE) of tasks is usually not adequately taken into account. Most WCRE analysis techniques only focus on the processor and neglect the energy consumption of other hardware units that are temporarily activated and deactivated in the system. To solve the problem of system-wide energy-consumption analysis, we present SysWCEC, an approach that addresses these problems by enabling static WCRE analysis for entire real-time systems, including internal as well as external devices. For this purpose, SysWCEC introduces a novel abstraction, the power-state--transition graph, which contains information about the worst-case energy consumption of all possible execution paths. To construct the graph, SysWCEC decomposes the analyzed real-time system into blocks during which the set of active devices in the system does not change and is consequently able to precisely handle devices being dynamically activated or deactivated. In this artifact evaluation, which accompanies our related conference paper, we present easy to reproduce WCRE analyses with the SysWCEC framework using several benchmarks. The artifact comprises the generation of the power-state--transition graph from a given benchmark system and the formulation of an integer linear program whose solution eventually yields safe WCRE bounds.

Although internal devices (e.g., memory, timers) and external devices (e.g., transceivers, sensors) significantly contribute to the energy consumption of an embedded real-time system, their impact on the worst-case response energy consumption (WCRE) of tasks is usually not adequately taken into account. Most WCRE analysis techniques, for example, only focus on the processor and therefore do not consider the energy consumption of other hardware units. Apart from that, the typical approach for dealing with devices is to assume that all of them are always activated, which leads to high WCRE overestimations in the general case where a system switches off the devices that are currently not needed in order to minimize energy consumption. In this paper, we present SysWCEC, an approach that addresses these problems by enabling static WCRE analysis for entire real-time systems, including internal as well as external devices. For this purpose, SysWCEC introduces a novel abstraction, the power-state-transition graph, which contains information about the worst-case energy consumption of all possible execution paths. To construct the graph, SysWCEC decomposes the analyzed real-time system into blocks during which the set of active devices in the system does not change and is consequently able to precisely handle devices being dynamically activated or deactivated.

Engineering related research, such as research on worst-case execution time, uses experimentation to evaluate ideas. For these experiments we need example programs. Furthermore, to make the research experimentation repeatable those programs shall be made publicly available. We collected open-source programs, adapted them to a common coding style, and provide the collection in open-source. The benchmark collection is called TACLeBench and is available from GitHub in version 1.9 at the publication date of this paper. One of the main features of TACLeBench is that all programs are self-contained without any dependencies on standard libraries or an operating system.

The fact that many benchmarks for evaluating worst-case execution time (WCET) analysis tools are based on real-world applications greatly increases the value of their results. However, at the same time, the complexity of these programs makes it difficult, sometimes even impossible, to obtain all corresponding flow facts (i.e., loop bounds, infeasible paths, and input values triggering the WCET), which are essential for a comprehensive evaluation. In this paper, we address this problem by presenting GenE, a benchmark generator that in addition to source code also provides the flow facts of the benchmarks created. To generate a new benchmark, the tool combines code patterns that are commonly found in real-time applications and are challenging for WCET analyzers. By keeping track of how patterns are put together, GenE is able to determine the flow facts of the resulting benchmark based on the known flow facts of the patterns used. Using this information, it is straightforward to synthesize the accurate WCET, which can then serve as a baseline for the evaluation of WCET analyzers.