NMPO: Near-Memory Computing Profiling and Offloading

TL;DR

This paper introduces NMPO, a high-level framework that uses machine learning to efficiently predict the suitability of applications for near-memory computing, significantly reducing simulation overhead.

Contribution

NMPO is a novel framework that accurately predicts NMC offloading potential using hardware-dependent features, accelerating NMC system evaluation.

Findings

Predicts NMC suitability with 85.6% accuracy.

Reduces prediction time by up to 1000 times compared to prior methods.

Enables faster early-stage NMC system design.

Abstract

Real-world applications are now processing big-data sets, often bottlenecked by the data movement between the compute units and the main memory. Near-memory computing (NMC), a modern data-centric computational paradigm, can alleviate these bottlenecks, thereby improving the performance of applications. The lack of NMC system availability makes simulators the primary evaluation tool for performance estimation. However, simulators are usually time-consuming, and methods that can reduce this overhead would accelerate the early-stage design process of NMC systems. This work proposes Near-Memory computing Profiling and Offloading (NMPO), a high-level framework capable of predicting NMC offloading suitability employing an ensemble machine learning model. NMPO predicts NMC suitability with an accuracy of 85.6% and, compared to prior works, can reduce the prediction time by using hardware-dependent applications features by up to 3 order of magnitude.