Skip to main content
Video s3
    Details
    Presenter(s)
    Dinesh Kushwaha Headshot
    Display Name
    Dinesh Kushwaha
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Country
    Country
    India
    Author(s)
    Display Name
    Dinesh Kushwaha
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Display Name
    Aditya Sharma
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Display Name
    Neha Gupta
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Display Name
    Ritik Raj
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Display Name
    Ashish Joshi
    Affiliation
    Affiliation
    Intel Technology India Pvt. Ltd.
    Display Name
    Jwalant Mishra
    Affiliation
    Affiliation
    NXP Semiconductor Bangalore
    Display Name
    Rajat Kohli
    Affiliation
    Affiliation
    NXP Semiconductor Bangalore
    Display Name
    Sandeep Miryala
    Affiliation
    Affiliation
    Brookhaven national Laboratory
    Display Name
    Rajiv Joshi
    Affiliation
    Display Name
    Sudeb Dasgupta
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Display Name
    Anand Bulusu
    Affiliation
    Affiliation
    Indian Institute of Technology Roorkee
    Abstract

    In this work, we propose an energy-efficient 64×64 compute-in-memory (CIM) SRAM macro using a 7T bit-cell in 65nm CMOS UMC PDK. It supports 4-bit inputs, 4-bit weights & 4-bit outputs and performs 4-bit MAC operations. It also supports multiple row activations performing 1024 4b×4b multiply and accumulate (MAC) operations in one clock cycle. Inputs are realized by the number of pulses on the read wordline (RWL), which discharges read bitline (RBL) according to bitwise multiplication of weights & inputs. Outputs of 4 columns storing 4-bit weights are then combined via charge sharing to perform a binary-weighted average representing MAC operation, further quantized by a flash analog to digital converter (ADC) giving 4-bit output. The proposed CIM macro achieves an energy efficiency of 28.9 TOPS/W and throughput of 212.9 GOPS operating at supply voltage 1 V with a 2 GHz clock frequency.

    Slides
    • A 65nm Compute-in-Memory 7T SRAM Macro Supporting 4-Bit Multiply and Accumulate Operation by Employing Charge Sharing (application/pdf)