IP Library Granted Patent US 12,555,629
Granted Patent B2
US 12,555,629 · App. 18/499,643 · Granted Feb 17, 2026

Method and system for a programmable and generic processing-in-SRAM accelerator

Inventors: Deliang Fan (Tempe, AZ); Shaahin Angizi (Newark, NJ)
Assignee: Arizona Board of Regents on behalf of Arizona State University
G11C11/419G06N3/06G11C7/1006G11C11/412G11C11/54H03K19/20
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,555,629
App. No.
18/499,643
Granted
Feb 17, 2026
Kind
B2
Abstract

A system for in-memory computing comprises a volatile memory comprising at least a first layered subarray, wherein each subarray comprises a plurality of memory cells, and a plurality of sub-sense amplifiers connected to a read bitline of the first subarray of the memory, configured to compare a measured voltage of the read bitline to at least one threshold and provide at least one binary output corresponding to a logic operation based on whether the voltage of the read bitline is above or below the threshold. A method for in-memory computing is also disclosed.

Claims (26)

1 . A system for in-memory computing, comprising:

a volatile memory comprising at least a first layered subarray, wherein each subarray comprises a plurality of memory cells; and

a plurality of sub-sense amplifiers connected to a read bitline of the first subarray of the memory, configured to compare a measured voltage of the read bitline to a threshold corresponding to the distinct logic operation and provide a binary output corresponding to the distinct logic operation based on whether the voltage of the read bitline is above or below the threshold.

2 . The system of claim 1 , wherein the logic operation is selected from NOR, XOR, and NAND.

3 . The system of claim 1 , wherein the plurality of sub-sense amplifiers comprises three sub-sense amplifiers; and

wherein the at least one threshold comprises at least three thresholds, corresponding to NOR, XOR, and NAND logic operations.

4 . The system of claim 1 , further comprising a multiplexer connected to the binary outputs of the plurality of sub-sense amplifiers, configured to select one of the binary outputs in response to a control signal.

5 . The system of claim 1 , wherein the memory is a random-access memory.

6 . The system of claim 5 , wherein the random-access memory is an in-processing static random-access memory (SRAM).

7 . The system of claim 1 , wherein the plurality of memory cells are 8T SRAM cells.

8 . The system of claim 4 , wherein the multiplexer is selected from a 2:1 multiplexer, a 3:1 multiplexer, and a 4:1 multiplexer.

9 . A method of in-memory computing, comprising:

storing a set of data in a layered subarray of a volatile memory;

comparing, with a set of sub-sense amplifiers, an analog voltage of a read bitline of the volatile memory to a set of corresponding voltage references, each voltage reference corresponding to a digital logic operation;

providing a set of corresponding binary output values from the sub-sense amplifiers; and

selecting, with a multiplexer, one of the set of corresponding binary output values.

10 . The method of claim 9 , wherein the three memory rows are three operand vectors stored in a memory.

11 . The method of claim 9 , wherein the memory is a random-access memory.

12 . The method of claim 11 , wherein the random-access memory is an in-processing static random-access memory (SRAM).

13 . The method of claim 10 , wherein the method is performed in a single memory cycle.

14 . The method of claim 9 , wherein the set of corresponding binary output values are three-input logic operations.

15 . The method of claim 14 , wherein the three-input logic operations are selected from XOR3, AND3, NOR3, and OR3.

16 . The system of claim 1 , wherein the plurality of sub-sense amplifiers comprises three sub-sense amplifiers connected to the read bitline of the first subarray of the memory, each of the three sub-sense amplifiers dedicated to distinct logic operations and are configured to, during a single memory read cycle, compare the measured voltage of the read bitline to the threshold corresponding to the respective distinct logic operation and provide the binary output corresponding to the respective distinct logic operation based on whether the voltage of the read bitline is above or below the threshold.

17 . The system of claim 16 , wherein a first sub-sense amplifier is configured to compare the measured voltage to a first threshold, wherein a second sub-sense amplifier is configured to compare the measured voltage to a second threshold, and third sub-sense amplifier is configured to compare the measured voltage to a third threshold.

18 . The system of claim 16 , wherein a first sub-sense amplifier is configured to perform a NOR logic operation, a second sub-sense amplifier is configured to perform a XOR logic operation, and a third sub-sense amplifier is configured to perform a NAND logic operation.

19 . The system of claim 1 , wherein the plurality of sub-sense amplifiers are simultaneously connected to the read bit line and configured to derive contents of the plurality of memory cells based on a fall time of the read bit line.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Feb 8, 2024
From: FAN, DELIANG; ANGIZI, SHAAHIN
To: ARIZONA BOARD OF REGENTS ON BEHALF OF ARIZONA STATE UNIVERSITY
Reel/Frame 066413/0885 →
Continuity (2)
Provisional Application 63381936 · Nov 2, 2022
Related Publication 20240144998A1 · May 2, 2024
References Cited (16)
US 10073733B1 · Jain · 2018 [cited by examiner]
US 10706923B2 · Seo · 2020 [cited by applicant]
US 20200279597A1 · Wang · 2020 [cited by applicant]
US 20230031841A1 · Badaroglu · 2023 [cited by examiner]
US 20230037054A1 · Wang · 2023 [cited by examiner]
US 20230297331A1 · Fan · 2023 [cited by examiner]
US 20240282366A1 · Jacob · 2024 [cited by examiner]
A. Biswas et al., “Conv-sram: An energy-efficient sram with in-memory dot-product computation for low-power convolutional neural networks,” IEEE JSSC, 2018. [cited by applicant]
Amitesh Sridharan, Shaahin Angizi, Sai Kiran Cherupally, Fan Zhang, Jae-sun Seo, and Deliang Fan, “A 1.23-GHz 16-Kb Programmable and Generic Processing-in-SRAM Accelerator in 65nm,” 48th European Solid-State Circuits Co… [cited by applicant]
H. Valavi et al., “A 64-tile 2.4-mb in-memory-computing cnn accelerator employing charge-domain compute,” IEEE JSSC, 2019. [cited by applicant]
J. Faraone et al., “Syq: Learning symmetric quantization for efficient deep neural networks,” in CVPR, 2018. [cited by applicant]
J. Wang et al., “A 28-nm compute SRAM with bit-serial logic/arithmetic operations for programmable in-memory vector computing,” IEEE JSSC, 2020. [cited by applicant]
J. Yue et al., “14.3 a 65nm computing-in-memory-based cnn processor with 2.9-to-35.8 tops/w system energy efficiency using dynamic-sparsity performance-scaling architecture and energy-efficient inter/intra-macro data re… [cited by applicant]
M. Hutter et al., “A cryptographic processor for low-resource devices: Canning ecdsa and aes like sardines,” in IFIP, 2011. [cited by applicant]
S. Mathew et al., “53gbps native gf (2 4) 2 composite-field aes-encrypt/decrypt accelerator for content-protection in 45nm highperformance microprocessors,” in 2010 Symp.on VLSI. IEEE. [cited by applicant]
Y. Zhang et al., “Recryptor: A reconfigurable cryptographic cortex-m0 processor with in-memory and near-memory computing for iot security,” IEEE JSSC, 2018. [cited by applicant]