TDO-CIM: Transparent Detection and Offloading for Computation In-memory

TL;DR

This paper introduces TDO-CIM, an LLVM-based compilation flow that automatically detects and offloads suitable kernels for in-memory computing, demonstrating significant performance benefits over traditional architectures.

Contribution

It presents an end-to-end compiler framework for in-memory computing that bridges the gap between architecture and compiler support.

Findings

Automatic detection and optimization of kernels for in-memory acceleration

Performance improvements over von Neumann architecture in simulations

Effective integration of in-memory computing in existing compilation workflows

Abstract

Computation in-memory is a promising non-von Neumann approach aiming at completely diminishing the data transfer to and from the memory subsystem. Although a lot of architectures have been proposed, compiler support for such architectures is still lagging behind. In this paper, we close this gap by proposing an end-to-end compilation flow for in-memory computing based on the LLVM compiler infrastructure. Starting from sequential code, our approach automatically detects, optimizes, and offloads kernels suitable for in-memory acceleration. We demonstrate our compiler tool-flow on the PolyBench/C benchmark suite and evaluate the benefits of our proposed in-memory architecture simulated in Gem5 by comparing it with a state-of-the-art von Neumann architecture.