OISMA: On-the-fly In-memory Stochastic Multiplication Architecture for Matrix-Multiplication Workloads
Shady Agwa, Yihan Pan, Georgios Papandroulidakis, Themis Prodromakis

TL;DR
OISMA is an energy-efficient in-memory computing architecture leveraging stochastic multiplication for matrix workloads, demonstrating significant improvements in energy and area efficiency over traditional IMC architectures.
Contribution
This work introduces OISMA, a novel IMC architecture that uses stochastic computing to enhance energy efficiency and scalability for matrix multiplication tasks.
Findings
Achieves 0.789 TOPS/W energy efficiency at 50 MHz.
Demonstrates two orders of magnitude energy efficiency improvement at 22-nm scaling.
Supports matrix multiplication with high area and energy efficiency.
Abstract
Artificial intelligence (AI) models are currently driven by a significant upscaling of their complexity, with massive matrix-multiplication workloads representing the major computational bottleneck. In-memory computing (IMC) architectures are proposed to avoid the von Neumann bottleneck. However, both digital/binary-based and analog IMC architectures suffer from various limitations, which significantly degrade the performance and energy efficiency gains. This work proposes OISMA, an energy-efficient IMC architecture that utilizes the computational simplicity of a quasi-stochastic computing (SC) domain (bent-pyramid (BP) system) while keeping the same efficiency, scalability, and productivity of digital memories. OISMA converts normal memory read operations into in situ stochastic multiplication operations with a negligible cost. An accumulation periphery then accumulates the output…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
