CuLD: Current-Limiting Differential Reading Circuit for Current-Based Compute-in-Memory
Seiji Uenohara, Satoshi Awamura, Norio Hattori

TL;DR
This paper introduces CuLD, a circuit that automatically adjusts MAC results in current-based compute-in-memory systems to reduce deviation during high parallelism operations, enhancing accuracy.
Contribution
The paper presents a novel current-limiting differential reading circuit that improves MAC accuracy in current-based CiM by dynamically adjusting for parallelism-induced deviations.
Findings
Reduces MAC deviation during high parallelism
Improves accuracy of current-based CiM circuits
Effective for charge integration methods
Abstract
This paper proposes a circuit configuration that addresses the issue of deviation in the multiply-accumulate (MAC) results when numerous word lines are simultaneously opened in current-based compute-in-memory (CiM) circuits. The proposed circuit solves this problem by automatically shrinking the product value according to the degree of parallelism. This circuit configuration is effective for circuit methods that calculate MAC through time integration of charge.
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.
Taxonomy
TopicsAdvanced Memory and Neural Computing · Ferroelectric and Negative Capacitance Devices · Quantum Computing Algorithms and Architecture
