Simon Holmbacka

Power-aware HEVC decoding with tunable image quality

A high pressure is put on mobile devices to support increasingly advanced applications requiring more processing capabilities. Among those, the emerging High Efficiency Video Coding (HEVC) provides a better video quality for the same bit rate than the previous H.264 standard. A limitation in the usability of a mobile video playing device is the lack of support for guaranteeing stand-by time and up time for battery driven devices. The Green Metadata initiative within the MPEG standard was launched to address the power saving issues of the decoder and defines the technology requirements. In this paper, we propose a HEVC decoder with tunable decoding quality levels for maximum power savings as suggested in the scope of the Green Metadata initiative. Our experiments reveal that the modified HEVC video decoder can save up to 28% of power consumption in real-world platforms while keeping better quality than decoding with H.264.

Task Migration for Dynamic Power and Performance Characteristics on Many-Core Distributed Operating Systems

Spatial locality of task execution will become more important on future hardware platforms since the number of cores are steadily increasing. The large amount of cores requires more intelligent power management due to the notion of spatial locality, and the high chip density requires an increased thermal awareness in order to avoid thermal hotspots on the chip. At the same time, high performance of the CPU is only achieved by parallelizing tasks over the chip in order to fully utilize the hardware. This paper presents a task migration mechanism for distributed operating systems running on many-core platforms. In this work, we evaluate the performance and energy efficiency of an implemented task migration mechanism. This is shown by parallelizing tasks as the performance of a single core is not sufficient, and by collecting tasks to as few cores as possible as CPU load is low. The task migration mechanism is implemented as a library for FreeRTOS using 1300 lines of code, and introduced a total task migration overhead of 100 ms on a shared memory platform. With the presented task migration mechanism, we intend to improve the dynamism of power and performance characteristics in distributed many-core operating systems.

Energy efficiency and performance management of parallel dataflow applications

Parallelizing software is a popular way of achieving high energy efficiency since parallel applications can be mapped on many cores and the clock frequency can be lowered. Perfect parallelism is, however, not often reached and different program phases usually contain different levels of parallelism due to data dependencies. Applications have currently no means of expressing the level of parallelism, and the power management is mostly done based on only the workload. In this work, we provide means of expressing QoS and levels of parallelism in applications for more tight integration with the power management to obtain optimal energy efficiency in multi-core systems. We utilize the dataflow framework PREESM to create and analyze program structures and expose the parallelism in the program phases to the power management. We use the derived parameters in a NLP (Non Linear Programming) solver to determine the minimum power for allocating resources to the applications.

Thermal influence on the energy efficiency of workload consolidation in many-core architectures

Webserver farms and datacenters currently use workload consolidation to match the dynamic workload with the available resources since switching off unused machines has been shown to save energy. The workload is placed on the active servers until the servers are saturated. The idea of workload consolidation can be brought also to chip level by the OS scheduler to pack as much workload to as few cores as possible in a many-core system. In this case all idle cores in the system are placed in a sleep state, and are woken up on-demand. Due to the relationship between static power dissipation and temperature, this paper investigates the thermal influence on the energy efficiency of chip level workload consolidation and its potential impact on the scheduling decisions. This work lay down the foundation for the development of a model for energy efficient OS scheduling for many-core processors taking into account external factors such as ambient and core level temperatures.

Energy-Awareness and Performance Management with Parallel Dataflow Applications

Applications have traditionally been executed as fast as possible (Race-to-Idle) and mapped to as many cores as possible (Fair scheduling) to minimize the energy consumption. With modern hardware, this method has become inefficient because of the power characteristics of the platforms. Instead, applications should utilize an optimal combination of clock frequency and number of cores to balance the dynamic and static power. Such approaches have been difficult to achieve since resource allocation is based only on CPU utilization. Resources are then allocated to prohibit over utilization rather than following software performance requirements. By adjusting the clock frequency directly according to software requirements and activating CPU cores according to the application parallelism, significant energy can be saved by lowering the average power dissipation. To enforce these recommendations, this paper provides means of expressing performance and parallelism in applications for more tight integration with the power management to balance the execution speed and mapping on multi-core systems. An interface between the applications and the hardware resources is provided in combination with a novel power management runtime system called Bricktop. A signal processing case study demonstrates real-world energy savings up to 50 % without performance degradation.

A PID-Controlled Power Manager for Energy Efficient Web Clusters

Large data centers using high-end processors operating continuously around the clock are major energy consumers. Long periods of idling due to low workload will cause a waste in energy because the processors are active but not doing any useful work. A cluster of low-end embedded processors could continuously match its computational capacity with the workload at a much finer granularity than a server-grade processor by changing the power states of the CPUs. This paper introduces a framework simulating a system level power manager for many-core clusters targeting server cards used in warehouse-sized data centers. The power management system uses sleep states to switch on or off processing elements in a cluster of low power boards to match the capacity of the whole system with the workload, and thus save energy. A PID-controller is implemented in the system, a component already well known with established methods in the industrial control domain. We intend to use this component to effectively determine the number of active processing elements in the used many-core cluster. The proposed power manager can save up to 62 percent in energy compared to a system which only uses dynamic voltage and frequency scaling as power management.

A task migration mechanism for distributed many-core operating systems

Spatial locality of task execution is becoming important in future hardware platforms since the number of cores is steadily increasing. The large amount of cores requires an intelligent power manager and the high chip and core density requires increased thermal awareness to avoid thermal hotspots on the chip. This paper presents a lightweight task migration mechanism explicitly for distributed operating systems running on many-core platforms. As the distributed OS runs one scheduler on each core, the tasks are migrated between OS kernels within the same shared memory platform. The benefits, such as performance and energy efficiency, of task migration are achieved by re-locating running tasks on the most appropriate cores and keeping the overhead of executing such a migration sufficiently low. We investigate the overhead of migrating tasks on a distributed OS running both on a bus-based platform and a many-core NoC—with these means of measures, we can predict the task migration overhead and pinpoint the emerging bottlenecks. With the presented task migration mechanism, we intend to improve the dynamism of power and performance characteristics in distributed many-core operating systems.

Datacenters powered by renewable energy: A case study for 60 degrees latitude north

Attention paid to energy efficiency in datacenters has been increasing significantly in the last decade. One of the latest trends on this issue is supplying datacenters with renewable energy, and reducing the carbon footprint of datacenters is a major step towards green computing. Following this trend, more and more datacenters are built in cold climate areas for cheaper cooling and increased energy efficiency of the facilities. However, such geographical locations have a very varying availability of renewable energy (especially solar energy), depending on the annual month and fitting the datacenters completely with renewable energy is hence more difficult. This paper analyzes the feasibility of using renewable energies for a datacenter located on the 60° latitude north. We analyze the energy consumption caused by the server workload and the cost-to-energy trade-off related to available wind and solar energy sources. A wind and solar power model is built based on real weather data for the city of Turku, Finland, and a feasibility study is conducted for fitting a datacenter with renewable energy sources on such a location. The available monthly and annual renewable energy is analyzed for different scenarios and compared with the energy consumption of a simulated datacenter. We also analyze the ratio between number of wind turbines and m2 solar panel to achieve a desired fraction of renewable energy sources for a given datacenter.

Accurate Energy Modelling for Many-Core Static Schedules

Static schedules can be a preferable alternative for applications with timing requirements and predictable behavior since the processing resources can be more precisely allocated for the given workload. Unused resources are handled by power management systems to either scale down or shut off parts of the chip to save energy. In order to efficiently implement power management, especially in many-core systems, an accurate model is important in order to make the appropriate power management decisions at the right time. For making correct decisions, practical issues such as latency for controlling the power saving techniques should be considered when deriving the system model, especially for fine timing granularity. In this paper we present an accurate energy model for many-core systems which includes switching latency of modern power saving techniques. The model is used when calculating an optimal static schedule for many-core task execution on systems with dynamic frequency levels and sleep state mechanisms. We create the model parameters for an embedded processor, and we validate it in practice with synthetic benchmarks on real hardware.

QoS Manager for Energy Efficient Many-Core Operating Systems

The oncoming many-core platforms is a hot topic these days, and this next generation hardware sets new focus on energy and thermal awareness. With a more and more dense packing of transistors, the system must be made energy aware to not suffer from overheating and energy waste. As a step towards increased energy efficiency, we intend to add the notion of QoS handling to the OS level and to applications. We suggest the design of a QoS manager as a plug-in OS extension capable of providing applications with the necessary resources leading to better energy efficiency.