A Review of SRAM-based Compute-in-Memory Circuits

TL;DR

This paper presents a tutorial and review of Static Random Access Memory-based compute-in-memory (CIM) circuits, with a focus on both digital CIM (DCIM) and analog CIM (ACIM) implementations, examining their respective advantages and challenges.

Abstract

This paper presents a tutorial and review of SRAM-based Compute-in-Memory (CIM) circuits, with a focus on both Digital CIM (DCIM) and Analog CIM (ACIM) implementations. We explore the fundamental concepts, architectures, and operational principles of CIM technology. The review compares DCIM and ACIM approaches, examining their respective advantages and challenges. DCIM offers high computational precision and process scaling benefits, while ACIM provides superior power and area efficiency, particularly for medium-precision applications. We analyze various ACIM implementations, including current-based, time-based, and charge-based approaches, with a detailed look at charge-based ACIMs. The paper also discusses emerging hybrid CIM architectures that combine DCIM and ACIM to leverage the strengths of both approaches.

Figures (17)

• Figure 1: Conceptional block diagram of a Compute-in-Memory Macro.
• Figure 2: Conceptional block diagram of a Compute-in-Memory Macro.
• Figure 3: Floating point (BF16) enabled DCIM macro (Figure adapted from khwa202434 © IEEE)
• Figure 4: INT12xINT12 DCIM macro (Figure adapted from fujiwara202434 © IEEE)
• Figure 5: DCIM macro with approximate DAT circuits. (Figure adapted from lin2023dimca © IEEE)