Title | ||
---|---|---|
PULP-NN: A Computing Library for Quantized Neural Network inference at the edge on RISC-V Based Parallel Ultra Low Power Clusters |
Abstract | ||
---|---|---|
We present PULP-NN, a multicore computing library for a parallel ultra-low-power cluster of RISC-V based processors. The library consists of a set of kernels for Quantized Neural Network (QNN) inference on edge devices, targeting byte and sub-byte data types, down to INT-1. Our software solution exploits the digital signal processing (DSP) extensions available in the PULP RISC-V processors and the cluster's parallelism, improving performance by up to 63× with respect to a baseline implementation on a single RISC-V core implementing the RV32IMC ISA. Using the PULP-NN routines, the inference of a CIFAR-10 QNN model runs in 30× and 19.6× less clock cycles than the current state-of-the-art ARM CMSIS-NN library, running on an STM32L4 and an STM32H7 MCUs, respectively. By running the library kernels on the GAP-8 processor at the maximum efficiency operating point, the energy efficiency on GAP-8 is 14.1× higher than STM32L4 and 39.5× than STM32H7. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/ICECS46596.2019.8965067 | 2019 26th IEEE International Conference on Electronics, Circuits and Systems (ICECS) |
Keywords | Field | DocType |
PULP-NN routines,CIFAR-10 QNN model,STM32H7 MCUs,GAP-8 processor,quantized neural network inference,RISC-V based parallel ultra low power clusters,multicore computing library,RISC-V based processors,edge devices,sub-byte data types,digital signal processing extensions,PULP RISC-V processors,single RISC-V core,RV32IMC ISA,DSP,cluster parallelism,ARM CMSIS-NN,STM32L4 MCUs | RISC-V,Byte,Digital signal processing,Computer science,Operating point,Parallel computing,Electronic engineering,Data type,Edge device,Software,Artificial neural network | Conference |
ISBN | Citations | PageRank |
978-1-7281-0997-8 | 2 | 0.45 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Angelo Garofalo | 1 | 3 | 0.81 |
Manuele Rusci | 2 | 2 | 1.47 |
Francesco Conti 0001 | 3 | 125 | 18.24 |
Davide Rossi | 4 | 416 | 47.47 |
Luca Benini | 5 | 13116 | 1188.49 |