Analog memory stores multiple vectors in single array for ai models
Superposed analog memory for vector storage in neural networks
Emerging TechnologiesHardware Architecture
Summary
Artificial intelligence models need a lot of memory to store their data, which can be energy-consuming. The authors show a new kind of analog memory that can hold many data vectors all mixed together in one place and still recover each one accurately. This means more data can fit in the same physical space, saving energy and resisting noise. They tested this on special memory chips using examples from language and recommendation models and found it worked as well as traditional digital methods.
What this means in practice
- •For ai hardware engineers: Design energy-efficient memory chips that store multiple neural network vectors compactly using analog superposition to reduce chip area and power usage.
- •For data center infrastructure teams: Integrate dense analog memory to handle large language and recommendation model embeddings with less energy and space compared to conventional digital memory.
Authors
Athanasios Vasilopoulos, Patrick Diener, Michael Hersche, Abbas Rahimi, Abu Sebastian
Abstract
The growing memory demands of artificial neural networks pose a major challenge for energy-efficient computing, particularly for models that rely on large collections of high-dimensional weight and embedding vectors. Conventional memory architectures store these vectors individually, requiring many memory cells per vector and limiting achievable storage density. Here we demonstrate an analog memory architecture that encodes multiple vectors in a distributed superposition within a single memory array. Individual vectors are recovered through in-memory matrix-vector multiplication, allowing the degree of superposition to be tuned to balance storage density against recall fidelity. This approach introduces a new scaling dimension for analog memory, in which effective storage density can be increased without a corresponding increase in the number of physical memory cells. The distributed representation also provides resilience to device-level noise. We implement the architecture on a prototype phase-change memory chip and experimentally evaluate it using embedding tables from deep learning recommendation models and large language models. The results demonstrate inference accuracy comparable to full-precision software baselines, establishing analog superposition as a promising route towards dense, energy-efficient memory for artificial intelligence.