Use of shared memory in the context of embedded multi-core processors: exploration of the technology and its limits

Modern embedded systems embrace many-core shared-memory designs. Due to constrained power and area budgets, most of them feature software-managed scratchpad memories instead of data caches to increase the data locality. It is therefore programmers’ responsibility to explicitly manage the memory transfers, and this make programming these platform cumbersome. Moreover, complex modern applications must be adequately parallelized before they can the parallel potential of the platform into actual performance. To support this, programming languages were proposed, which work at a high level of abstraction, and rely on a runtime whose cost hinders performance, especially in embedded systems, where resources and power budget are constrained. This dissertation explores the applicability of the shared-memory paradigm on modern many-core systems, focusing on the ease-of-programming. It focuses on OpenMP, the de-facto standard for shared memory programming. In a first part, the cost of algorithms for synchronization and data partitioning are analyzed, and they are adapted to modern embedded many-cores. Then, the original design of an OpenMP runtime library is presented, which supports complex forms of parallelism such as multi-level and irregular parallelism. In the second part of the thesis, the focus is on heterogeneous systems, where hardware accelerators are coupled to (many-)cores to implement key functional kernels with orders-of-magnitude of speedup and energy efficiency compared to the “pure software” version. However, three main issues rise, namely i) platform design complexity, ii) architectural scalability and iii) programmability. To tackle them, a template for a generic hardware processing unit (HWPU) is proposed, which share the memory banks with cores, and the template for a scalable architecture is shown, which integrates them through the shared-memory system. Then, a full software stack and toolchain are developed to support platform design and to let programmers exploiting the accelerators of the platform. The OpenMP frontend is extended to interact with it.

Nano-Power Integrated Circuits for Energy Harvesting

The energy harvesting research field has grown considerably in the last decade due to increasing interests in energy autonomous sensing systems, which require smart and efficient interfaces for extracting power from energy source and power management (PM) circuits. This thesis investigates the design trade-offs for minimizing the intrinsic power of PM circuits, in order to allow operation with very weak energy sources. For validation purposes, three different integrated power converter and PM circuits for energy harvesting applications are presented. They have been designed for nano-power operations and single-source converters can operate with input power lower than 1 μW. The first IC is a buck-boost converter for piezoelectric transducers (PZ) implementing Synchronous Electrical Charge Extraction (SECE), a non-linear energy extraction technique. Moreover, Residual Charge Inversion technique is exploited for extracting energy from PZ with weak and irregular excitations (i.e. lower voltage), and the implemented PM policy, named Two-Way Energy Storage, considerably reduces the start-up time of the converter, improving the overall conversion efficiency. The second proposed IC is a general-purpose buck-boost converter for low-voltage DC energy sources, up to 2.5 V. An ultra-low-power MPPT circuit has been designed in order to track variations of source power. Furthermore, a capacitive boost circuit has been included, allowing the converter start-up from a source voltage VDC0 = 223 mV. A nano-power programmable linear regulator is also included in order to provide a stable voltage to the load. The third IC implements an heterogeneous multisource buck-boost converter. It provides up to 9 independent input channels, of which 5 are specific for PZ (with SECE) and 4 for DC energy sources with MPPT. The inductor is shared among channels and an arbiter, designed with asynchronous logic to reduce the energy consumption, avoids simultaneous access to the buck-boost core, with a dynamic schedule based on source priority.

High performance and energy-efficient instruction cache design and optimisation for ultra-low-power multi-core clusters

High Energy efficiency and high performance are the key regiments for Internet of Things (IoT) end-nodes. Exploiting cluster of multiple programmable processors has recently emerged as a suitable solution to address this challenge. However, one of the main bottlenecks for multi-core architectures is the instruction cache. While private caches fall into data replication and wasting area, fully shared caches lack scalability and form a bottleneck for the operating frequency. Hence we propose a hybrid solution where a larger shared cache (L1.5) is shared by multiple cores connected through a low-latency interconnect to small private caches (L1). However, it is still limited by large capacity miss with a small L1. Thus, we propose a sequential prefetch from L1 to L1.5 to improve the performance with little area overhead. Moreover, to cut the critical path for better timing, we optimized the core instruction fetch stage with non-blocking transfer by adopting a 4 x 32-bit ring buffer FIFO and adding a pipeline for the conditional branch. We present a detailed comparison of different instruction cache architectures' performance and energy efficiency recently proposed for Parallel Ultra-Low-Power clusters. On average, when executing a set of real-life IoT applications, our two-level cache improves the performance by up to 20% and loses 7% energy efficiency with respect to the private cache. Compared to a shared cache system, it improves performance by up to 17% and keeps the same energy efficiency. In the end, up to 20% timing (maximum frequency) improvement and software control enable the two-level instruction cache with prefetch adapt to various battery-powered usage cases to balance high performance and energy efficiency.

Planning and flexible operation of storage systems in power grids: from transmission to distribution networks

The first part of the thesis has been devoted to the transmission planning with high penetration of renewable energy sources. Both stationary and transportable battery energy storage (BES, BEST) systems have been considered in the planning model, so to obtain the optimal set of BES, BEST and transmission lines that minimizes the total cost in a power network. First, a coordinated expansion planning model with fixed transportation cost for BEST devices has been presented; then, the model has been extended to a planning formulation with a distance-dependent transportation cost for the BEST units, and its tractability has been proved through a case study based on a 190-bus test system. The second part of this thesis is then devoted to the analysis of planning and management of renewable energy communities (RECs). Initially, the planning of photovoltaic and BES systems in a REC with an incentive-based remuneration scheme according to the Italian regulatory framework has been analysed, and two planning models, according to a single-stage, or a multi-stage approach, have been proposed in order to provide the optimal set of BES and PV systems allowing to achieve the minimum energy procurement cost in a given REC. Further, the second part of this thesis is devoted to the study of the day-ahead scheduling of resources in renewable energy communities, by considering two types of REC. The first one, which we will refer to as “cooperative community”, allows direct energy transactions between members of the REC; the second type of REC considered, which we shall refer to as “incentive-based”, does not allow direct transactions between members but includes economic revenues for the community shared energy, according to the Italian regulation framework. Moreover, dispatchable renewable energy generation has been considered by including producers equipped with biogas power plants in the community.