Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
139 tokens/sec
GPT-4o
47 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

X-HEEP: An Open-Source, Configurable and Extendible RISC-V Microcontroller for the Exploration of Ultra-Low-Power Edge Accelerators (2401.05548v2)

Published 10 Jan 2024 in cs.AR

Abstract: The field of edge computing has witnessed remarkable growth owing to the increasing demand for real-time processing of data in applications. However, challenges persist due to limitations in performance and power consumption. To overcome these challenges, heterogeneous architectures have emerged that combine host processors with specialized accelerators tailored to specific applications, leading to improved performance and reduced power consumption. However, most of the existing platforms lack the necessary configurability and extendability options for integrating custom accelerators. To overcome these limitations, we introduce in this paper the eXtendible Heterogeneous Energy-Efficient Platform (X-HEEP). X-HEEP is an open-source platform designed to natively support the integration of ultra-low-power edge accelerators. It provides customization options to match specific application requirements by exploring various core types, bus topologies, addressing modes, memory sizes, and peripherals. Moreover, the platform prioritizes energy efficiency by implementing low-power strategies, such as clock-gating and power-gating. We demonstrate the real-world applicability of X-HEEP by providing an integration example tailored for healthcare applications that includes a coarse-grained reconfigurable array (CGRA) and in-memory computing (IMC) accelerators. The resulting design, called HEEPocrates, has been implemented both in field programmable gate array (FPGA) on the Xilinx Zynq-7020 chip and in silicon with TSMC 65nm low-power CMOS technology. We run a set of healthcare applications and measure their energy consumption to demonstrate the alignment of our chip with other state-of-the-art microcontrollers commonly adopted in this domain. Moreover, we present the energy benefits of 4.9x and 4.8x gained by exploiting the integrated CGRA and IMC accelerators compared to running on the host CPU.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (35)
  1. “Chipyard: Integrated Design, Simulation, and Implementation Framework for Custom SoCs” In IEEE Micro 40.4, 2020, pp. 10–21 DOI: 10.1109/MM.2020.2996616
  2. David E Bellasi and Luca Benini “Smart energy-efficient clock synthesizer for duty-cycled sensor socs in 65 nm/28nm cmos” In IEEE Transactions on Circuits and Systems I: Regular Papers 64.9 IEEE, 2017, pp. 2322–2333
  3. Andrea Bocco, Yves Durand and Florent De Dinechin “SMURF: Scalar Multiple-Precision Unum Risc-V Floating-Point Accelerator for Scientific Computing” In Proc. of the ACM Conference for Next Generation Arithmetic, CoNGA’19, 2019 DOI: 10.1145/3316279.3316280
  4. Rubén Braojos, Giovanni Ansaloni and David Atienza “A Methodology for Embedded Classification of Heartbeats Using Random Projections” In DATE IEEE, 2013, pp. 899–904 DOI: 10.7873/DATE.2013.189
  5. “A 12.4TOPS/W @ 136GOPS AI-IoT System-on-Chip with 16 RISC-V, 2-to-8b Precision-Scalable DNN Acceleration and 30%-Boost Adaptive Body Biasing” In IEEE ISSCC, 2023, pp. 21–23 DOI: 10.1109/ISSCC42615.2023.10067643
  6. “core-v-xif” URL: https://github.com/openhwgroup/core-v-xif
  7. “Modular Design and Optimization of Biomedical Applications for Ultralow Power Heterogeneous Platforms” In IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems 39.11, 2020, pp. 3821–3832 DOI: 10.1109/TCAD.2020.3012652
  8. “Implementation and Integration of Keccak Accelerator on RISC-V for CRYSTALS-Kyber” In Proc. of the 20th ACM Int. Conf. on Computing Frontiers, CF ’23, 2023, pp. 381–382 DOI: 10.1145/3587135.3591432
  9. “A multi-core reconfigurable architecture for ultra-low power bio-signal analysis” In IEEE BioCAS, 2016, pp. 416–419 DOI: 10.1109/BioCAS.2016.7833820
  10. “ESP” URL: https://github.com/sld-columbia/esp
  11. Tim Fritzmann, Georg Sigl and Johanna Sepúlveda “RISQ-V: Tightly Coupled RISC-V Accelerators for Post-Quantum Cryptography” In IACR Transactions on Cryptographic Hardware and Embedded Systems 2020.4, 2020, pp. 239–280 DOI: 10.13154/tches.v2020.i4.239-280
  12. “FuseSoC” URL: https://github.com/olofk/fusesoc
  13. “Near-threshold RISC-V core with DSP extensions for scalable IoT endpoint devices” In IEEE Transactions on Very Large Scale Integration (VLSI) Systems 25.10 IEEE, 2017, pp. 2700–2713
  14. “Automatic seizure detection based on imaged-EEG signals through fully convolutional networks” In Scientific reports 10.1 Nature Publishing Group, 2020, pp. 1–13
  15. “Towards a Truly Integrated Vector Processing Unit for Memory-Bound Applications Based on a Cost-Competitive Computational SRAM Design Solution” In J. Emerg. Technol. Comput. Syst. 18.2 New York, NY, USA: Association for Computing Machinery, 2022 DOI: 10.1145/3485823
  16. Kai Li, Wei Yin and Qiang Liu “A Portable DSP Coprocessor Design Using RISC-V Packed-SIMD Instructions” In IEEE ISCAS, 2023, pp. 1–5 DOI: 10.1109/ISCAS46773.2023.10181681
  17. “Litex” URL: https://github.com/enjoy-digital/litex
  18. LowRISC “OpenTitan” URL: https://github.com/lowRISC/opentitan
  19. David Mallasén, Alberto A. del Barrio and Manuel Prieto-Matias “Big-PERCIVAL: Exploring the Native Use of 64-Bit Posit Arithmetic in Scientific Computing”, 2023 arXiv:2305.06946
  20. “Big biomedical image processing hardware acceleration: A case study for K-means and image filtering” In IEEE ISCAS, 2016, pp. 1134–1137 DOI: 10.1109/ISCAS.2016.7527445
  21. “Open Bus Interface Protocol” URL: https://github.com/pulp-platform/obi
  22. “Cheshire: A Lightweight, Linux-Capable RISC-V Host Platform for Domain-Specific Accelerator Plug-In” In IEEE Transactions on Circuits and Systems II: Express Briefs, 2023, pp. 1–1 DOI: 10.1109/TCSII.2023.3289186
  23. “BlackParrot: An Agile Open-Source RISC-V Multicore for Accelerator SoCs” In IEEE Micro 40.4, 2020, pp. 93–102 DOI: 10.1109/MM.2020.2996145
  24. “A Hardware/Software Co-Design Vision for Deep Learning at the Edge” In IEEE Micro 42.6 Los Alamitos, CA, USA: IEEE Computer Society, 2022, pp. 48–54 DOI: 10.1109/MM.2022.3195617
  25. “Mr. Wolf: An energy-precision scalable parallel ultra low power SoC for IoT edge processing” In IEEE Journal of Solid-State Circuits 54.7 IEEE, 2019, pp. 1970–1981
  26. “Rocket” URL: https://github.com/chipsalliance/rocket-chip
  27. “Arnold: An eFPGA-augmented RISC-V SoC for flexible and low-power IoT end nodes” In IEEE Transactions on Very Large Scale Integration (VLSI) Systems 29.4 IEEE, 2021, pp. 677–690
  28. “Quentin: an Ultra-Low-Power PULPissimo SoC in 22nm FDX” In 2018 IEEE SOI-3D-Subthreshold Microelectronics Technology Unified Conference (S3S), 2018, pp. 1–3 DOI: 10.1109/S3S.2018.8640145
  29. “Slow and steady wins the race? A comparison of ultra-low-power RISC-V cores for Internet-of-Things applications” In Int. Symp. on Power and Timing Modeling, Optimization and Simulation (PATMOS), 2017, pp. 1–8 IEEE
  30. “X-HEEP: An Open-Source, Configurable and Extendible RISC-V Microcontroller” In Proc. of Int. Conf. on Computing Frontiers, CF ’23 New York, NY, USA: ACM, 2023, pp. 379–380 DOI: 10.1145/3587135.3591431
  31. “BLADE: An in-cache computing architecture for edge devices” In IEEE Transactions on Computers 69.9 IEEE, 2020, pp. 1349–1363
  32. “Echoes: a 200 GOPS/W Frequency Domain SoC with FFT Processor and I2S DSP for Flexible Data Acquisition from Microphone Arrays”, 2023 arXiv:2305.07325
  33. “Vortex: Extending the RISC-V ISA for GPGPU and 3D-Graphics” In IEEE/ACM Int. Symp. on Microarchitecture (MICRO), 2021, pp. 754–766 DOI: 10.1145/3466752.3480128
  34. “Architectural design of a complex arithmetic signal processor (CASP)” In Region 5 Conference: Annual Technical and Leadership Workshop, 2004, pp. 69–76 DOI: 10.1109/REG5.2004.1300163
  35. “The cost of application-class processing: Energy and performance analysis of a Linux-ready 1.7-GHz 64-bit RISC-V core in 22-nm FDSOI technology” In IEEE Transactions on Very Large Scale Integration (VLSI) Systems 27.11 IEEE, 2019, pp. 2629–2640
Citations (9)
View on Semantic Scholar

Summary

  • The paper presents X-HEEP, an open-source RISC-V microcontroller that enables flexible integration of ultra-low-power edge accelerators.
  • It introduces configurable architecture options including customizable cores, bus topologies, and memory configurations to optimize power and performance.
  • Experimental use case HEEPocrates demonstrates energy improvements of nearly 5× over standalone CPU execution in healthcare applications.

An Essay on "X-HEEP: An Open-Source, Configurable and Extendible RISC-V Microcontroller for the Exploration of Ultra-Low-Power Edge Accelerators"

The paper "X-HEEP: An Open-Source, Configurable and Extendible RISC-V Microcontroller for the Exploration of Ultra-Low-Power Edge Accelerators" presents a novel platform that addresses existing gaps and challenges in the design of ultra-low-power devices for edge computing. Specifically, it tackles the problem of configurability and extendability in platforms used for integrating custom accelerators, offering a solution based on the open RISC-V architecture.

Key Contributions and Architecture

The authors present X-HEEP, an open-source platform that addresses several limitations present in current ultra-low-power edge computing systems. It provides a system specifically designed for integration with ultra-low-power edge accelerators and supports customization of core types, bus topologies, and memory configurations. This configurability allows designers the flexibility to tailor the platform to meet specific application constraints in terms of area, performance, and power consumption. Key architectural elements of X-HEEP include:

  • Core Options: Integration of RISC-V cores from the OpenHW Group CORE-V family, providing a range of computational capabilities suited to varying processing demands.
  • Bus Topologies: Options for one-at-a-time or fully connected bus topologies, accommodating different bandwidth requirements and power considerations.
  • Memory Configuration: Flexible memory size and bank configurations with support for low-power modes such as retention, allowing energy efficiency improvements.

Extendible Accelerator Interface (XAIF)

The paper introduces the Extendible Accelerator Interface (XAIF), a versatile solution for integrating various accelerators while maintaining high system configurability. This interface supports:

  • Memory Mapped Ports: OBI protocol-based slave and master ports facilitating efficient data transfer between accelerators and main systems.
  • Interrupts and Power Control: Options for fine-grained power management and rapid synchronization with the host CPU, which are critical for optimizing power consumption across different operational phases.

Demonstrated Use Case: HEEPocrates

To underscore the real-world applicability and efficacy of X-HEEP, the authors present an implementation named HEEPocrates, tailored for healthcare applications. Featuring a combination of a Coarse-Grained Reconfigurable Array (CGRA) and In-Memory Computing (IMC) accelerators, HEEPocrates exemplifies the platform's flexibility in addressing domain-specific computational demands. The implementation is verified in both FPGA and silicon using TSMC's 65 nm technology, delivering promising results:

  • Operational Range: Demonstrated performance from 0.8V to 1.2V, achieving frequencies of up to 470 MHz.
  • Power Consumption: Ranges from 270 μW at lower operational frequencies to 48 mW when pushing full operational capacity, highlighting its suitability for ultra-low-power applications.
  • Energy Efficiency: The accelerators deliver energy improvements of approximately 4.9× and 4.8× over standalone execution on the host CPU, validating the effectiveness of such configurations.

Experimental Analysis and Implications

The paper positions X-HEEP as a formidable alternative to, or complement to, existing platforms like Apollo 3 Blue and GAP9. It contributes a balance of power efficiency and configurability, factors that are crucial for the flexible support of acquisition-dominated and processing-dominated tasks typical of domain-specific applications such as healthcare.

This work also sets the foundation for future developments in open-source architectures, enabling customizable, efficient edge computing systems that leverage specialized accelerators. Speculatively, X-HEEP's introduction could stimulate newer application-specific innovations by providing a foundational, configurable toolset optimized for energy efficiency in varied domains beyond healthcare, such as industrial IoT or mobile AI solutions.

In conclusion, "X-HEEP: An Open-Source, Configurable and Extendible RISC-V Microcontroller for the Exploration of Ultra-Low-Power Edge Accelerators" contributes a significant advancement in edge computing. It proposes a framework conducive to rapid development, customization, and real-world deployment of energy-efficient, highly configurable computing architectures.

File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown
Github Logo Streamline Icon: https://streamlinehq.com

GitHub

  1. GitHub - esl-epfl/x-heep: eXtendable Heterogeneous Energy-Efficient Platform based on RISC-V (193 stars)