PAPS: Power budget-Aware Pipeline Scheduling for an Embedded ReRAM-based Accelerator

被引：0

作者：

Shuai, Changchi ^{[1
]}

Qiu, Keni ^{[1
]}

机构：

[1] Capital Normal Univ, Beijing, Peoples R China

来源：

SEC'19: PROCEEDINGS OF THE 4TH ACM/IEEE SYMPOSIUM ON EDGE COMPUTING | 2019年

基金：

中国国家自然科学基金;

关键词：

D O I：

10.1145/3318216.3363455

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Many recent works have proposed to embed the emerging resistive switching random-access memory crossbar (ReRAM for short) to an edge device in Internet of Things (IoTs), such that those IoT nodes themselves are smart to process data rather than merely transmit raw data [1]. To acquire high throughput, multiple ReRAMs are scheduled in a pipeline fashion in the execution, but the peak power consumption of the ReRAM pipeline may probably exceed its power budget. Addressing the power issue, this paper proposes a Power budget-Aware Pipeline Scheduling (PAPS) scheme, which can achieve a well balanced and low power ReRAM pipeline.

引用

页码：352 / 353

页数：2