English

Mixed-Precision Training and Compilation for RRAM-based Computing-in-Memory Accelerators

Machine Learning 2026-03-20 v3 Emerging Technologies

Abstract

Computing-in-Memory (CIM) accelerators are a promising solution for accelerating Machine Learning (ML) workloads, as they perform Matrix-Vector Multiplications (MVMs) on crossbar arrays directly in memory. Although the bit widths of the crossbar inputs and cells are very limited, most CIM compilers do not support quantization below 8 bit. As a result, a single MVM requires many compute cycles, and weights cannot be efficiently stored in a single crossbar cell. To address this problem, we propose a mixed-precision training and compilation framework for CIM architectures. The biggest challenge is the massive search space, that makes it difficult to find good quantization parameters. This is why we introduce a reinforcement learning-based strategy to find suitable quantization configurations that balance latency and accuracy. In the best case, our approach achieves up to a 2.48x speedup over existing state-of-the-art solutions, with an accuracy loss of only 0.086 %.

Keywords

Cite

@article{arxiv.2601.21737,
  title  = {Mixed-Precision Training and Compilation for RRAM-based Computing-in-Memory Accelerators},
  author = {Rebecca Pelke and Joel Klein and Jose Cubero-Cascante and Nils Bosbach and Jan Moritz Joseph and Rainer Leupers},
  journal= {arXiv preprint arXiv:2601.21737},
  year   = {2026}
}

Comments

PREPRINT - Accepted for publication at the Design, Automation & Test in Europe Conference & Exhibition (DATE), April 20-22, 2026, in Verona, Italy V2 - fixed typos

R2 v1 2026-07-01T09:25:44.422Z