Cryogenic 3T CIMC Macros for Reliable Writes and Boolean Computing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing cryogenic enhanced dynamic random access memory (eDRAM) is not optimal for reliable write operations at cryogenic temperatures, and its bitcell topology needs to be redesigned to meet the requirements of different computing operations, particularly in energy-efficient Boolean logic and convolutional operations.

Innovation Solution

The design of an energy-efficient cryogenic-in-memory-computing (CIMC) accelerator using cryogenic 3T (C3T) macros, which include a C3T array with bitcells, a digital timing sequence converter, and a sense amplifier, enabling efficient charging and discharging on read bit lines, and adaptive reconfigurable sense amplifiers for Boolean and convolutional operations, along with a cryogenically optimized flash analog-to-digital converter.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If existing cryogenic eDRAM is used, then memory storage is achieved, but write operation reliability deteriorates at cryogenic temperatures

Engineering Contradiction:
Improvewrite operation reliabilityVSAvoidcryogenic temperature
Core Design Contradiction:
ReliabilityVSTemperature

Solution Approach 1:

The patent changes the bitcell topology from conventional 6T to a specialized 3T design optimized for cryogenic temperatures. This structural parameter change enables reliable write operations at cryogenic temperatures by using a pull-up transistor configuration that maintains proper voltage levels during write operations, overcoming the reliability issues of standard eDRAM at low temperatures.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent implements dynamic control mechanisms including word line voltage boosting and adaptive timing sequences that adjust operation parameters based on temperature conditions. The sense amplifier is dynamically configured with adjustable reference voltages to maintain optimal sensitivity across varying cryogenic temperatures, ensuring reliable write and read operations.

Inventive Principle:
Principle #15Dynamics

2Use of energy by moving object

If cryogenic computing architecture is implemented, then energy efficiency is improved, but computational versatility deteriorates due to limited operation types

Engineering Contradiction:
Improveenergy efficiencyVSAvoidcomputational operation versatility
Core Design Contradiction:
Use of energy by moving objectVSAdaptability or versatility

Solution Approach 1:

The patent designs a universal C3T bitcell that can perform multiple computational functions including Boolean logic operations (AND, OR, NOT), convolutional operations, and standard memory operations. The same hardware structure supports different computing modes by adjusting control signals and timing sequences, eliminating the need for separate specialized circuits for each operation type.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent implements a reconfigurable architecture where the computational mode is dynamically selected through control signals. The system can switch between Boolean logic mode, convolutional mode, and memory mode by adjusting word line activation sequences and sense amplifier configuration, providing computational versatility while maintaining energy-efficient cryogenic operation.

Inventive Principle:
Principle #15Dynamics

3Reliability

If C3T bitcell topology is used, then write operation reliability is improved, but device complexity increases compared to conventional eDRAM

Engineering Contradiction:
Improvewrite operation reliabilityVSAvoidbitcell topology complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies local quality optimization by placing the pull-up transistor specifically at the bit line intersection point where it is most needed for write operation reliability. This localized structural modification provides the necessary voltage boosting only where required, rather than complicating the entire memory array structure, thus balancing reliability improvement with acceptable device complexity.

Inventive Principle:
Principle #3Local quality

4Use of energy by moving object

If cryogenic computing is implemented, then cooling cost is reduced through high energy efficiency, but manufacturing precision requirements worsen due to cryogenic environment constraints

Engineering Contradiction:
Improveenergy efficiencyVSAvoidcryogenic environment manufacturing precision
Core Design Contradiction:
Use of energy by moving objectVSManufacturing precision

Solution Approach 1:

The patent optimizes transistor sizing parameters specifically for cryogenic operation, adjusting width-to-length ratios to compensate for changed carrier mobility characteristics at low temperatures. The pull-up transistor is sized to provide appropriate pull-up strength without requiring excessive precision in fabrication, making the design more tolerant of manufacturing variations while maintaining reliability at cryogenic temperatures.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS20240221811A1Energy-efficient cryogenic-in-memory-computing (CIMC) accelerator
Publication Date: 2024.07.04 SHANGHAI TECH UNIV
  • US20240221811A1 patent drawing
  • US20240221811A1 patent drawing
  • US20240221811A1 patent drawing

AI summary

An energy-efficient cryogenic-in-memory-computing (CIMC) accelerator includes cryogenic 3T (C3T) macros. Each of the C3T macros comprises a C3T array containing M rows×N columns of bitcells. An input signal is converted into a timing sequence signal of a corresponding pulse width by using a digital timing sequence converter array. A C3T bitcell of a corresponding row in the C3T macro is controlled to perform charging and discharging on a read bit line (RBL) of a corresponding column. A voltage on the RBL of the corresponding column is sampled by a sense amplifier configured in each C3T macro to obtain a final result. With adaptive reference voltage configuration and storage on the chip, this design can achieve fast and low-power boolean/convolutional computing.