XpulpNN: Enabling Energy Efficient and Flexible Inference of Quantized Neural Network on RISC-V based IoT End Nodes

TL;DR

This paper presents ISA extensions and a parallel RISC-V cluster design to significantly improve energy-efficient quantized neural network inference on IoT microcontrollers, achieving near-linear speedups and high efficiency.

Contribution

It introduces nibble and crumb SIMD instructions and a new execution paradigm for RISC-V, enabling efficient QNN inference on microcontrollers with a scalable multi-core architecture.

Findings

6x to 8x faster QNN kernels with 2-4 bit data operands

Peak efficiency of 2.22 TOPs/s/W comparable to dedicated accelerators

Up to 1000x better energy efficiency than ARM Cortex-M microcontrollers

Abstract

This work introduces lightweight extensions to the RISC-V ISA to boost the efficiency of heavily Quantized Neural Network (QNN) inference on microcontroller-class cores. By extending the ISA with nibble (4-bit) and crumb (2-bit) SIMD instructions, we are able to show near-linear speedup with respect to higher precision integer computation on the key kernels for QNN computation. Also, we propose a custom execution paradigm for SIMD sum-of-dot-product operations, which consists of fusing a dot product with a load operation, with an up to 1.64x peak MAC/cycle improvement compared to a standard execution scenario. To further push the efficiency, we integrate the RISC-V extended core in a parallel cluster of 8 processors, with near-linear improvement with respect to a single core architecture. To evaluate the proposed extensions, we fully implement the cluster of processors in GF22FDX technology. QNN convolution kernels on a parallel cluster implementing the proposed extension run 6 x and 8 x faster when considering 4- and 2-bit data operands, respectively, compared to a baseline processing cluster only supporting 8-bit SIMD instructions. With a peak of 2.22 TOPs/s/W, the proposed solution achieves efficiency levels comparable with dedicated DNN inference accelerators, and up to three orders of magnitude better than state-of-the-art ARM Cortex-M based microcontroller systems such as the low-end STM32L4 MCU and the high-end STM32H7 MCU.