Energy Efficient Low Latency Multi-issue Cores for Intelligent Always-On IoT Applications
- PDF / 2,042,089 Bytes
- 17 Pages / 595.224 x 790.955 pts Page_size
- 68 Downloads / 182 Views
Energy Efficient Low Latency Multi-issue Cores for Intelligent Always-On IoT Applications Joonas Multanen1
1 ¨ ¨ · Heikki Kultala1 · Kati Tervo1 · Pekka Ja¨ askel ainen
Received: 2 May 2019 / Revised: 29 June 2020 / Accepted: 6 July 2020 © The Author(s) 2020
Abstract Advanced Internet-of-Things applications require control-oriented codes to be executed with low latency for fast responsivity while their advanced signal processing and decision making tasks require computational capabilities. For this context, we propose three multi-issue core designs featuring an exposed datapath architecture with high performance, while retaining energy-efficiency. These features are achieved with exploitation of instruction-level parallelism, fast branching and the use of an instruction register file. With benchmarks in control-flow and signal processing application domains we measured in the best case 64% reduced energy consumption compared to a state-of-the-art RISC core, while consuming less silicon area. A high-performance design point reaches nearly 2.6 GHz operating frequency in the best case, over 2× improvement, while simultaneously achieving a 14% improvement in system energy-delay product. Keywords Low power · Instruction stream · Energy-efficiency · Instruction register file · IoT · Always-on · RISC-V · TTA · Exposed datapath · Transport Triggered Architecture
1 Introduction It is estimated, that the information and communication technology (ICT) sector will consume up to 20% of global energy production by 2025 [1]. From an environmental point of view, there are estimates that around 14% of total greenhouse gas emissions emerge from the ICT sector by 2040 [2]. The era of Internet-of-Things (IoT) and its increasing demands on computational complexity are expected to result in the introduction of billions of compute devices. Many of these small form factor devices
Joonas Multanen
[email protected] Heikki Kultala [email protected] Kati Tervo [email protected] Pekka J¨aa¨ skel¨ainen [email protected] 1
Faculty of Information Technology and Communication Sciences, Tampere University, Tampere, Finland
are battery-powered or use energy harvesting for their power supply, requiring energy-efficient and low power operation. While maintaining low energy consumption, devices such as always-on surveillance cameras, small drones, and sensor nodes, are required to react to events and perform demanding signal processing and artificial intelligence tasks, and also to handle external events with low control code execution latency. Besides their low power and energy consumption requirements, this calls for the devices to be highly performance scalable. For maximal energy-efficiency, fixed function accelerators are typically used. Compared to programmable devices, their hardware is optimized at design-time to match predefined requirements. This allows removing instruction delivery overheads and tailoring the datapath, resulting in high computational capability and energy efficiency in small chip area. The clear
Data Loading...
