ROM Architecture Selective Encoding Bit Line Merging

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional ROM bitcells and memory arrays face inefficiencies in area usage, cost, and power consumption due to high current leakage, making them unsuitable for applications requiring significant on-chip read-only memory.

Innovation Solution

The implementation of improved ROM bitcells and arrays that store multiple logic states per memory cell using a transistor with program nodes connected to bit lines, allowing for selective encoding and decoding methods based on dominant logic states, reducing power consumption and area requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If conventional ROM bitcells are used to store data, then the memory array can be implemented, but the chip area is large and cost is high

Engineering Contradiction:
Improvechip areaVSAvoiddata storage capacity
Core Design Contradiction:
Area of stationary objectVSQuantity of substance

Solution Approach 1:

Multiple memory cells are merged to share common bit lines. Specifically, four memory cells share two common bit lines, allowing the same bit lines to be used for reading data from multiple cells by selectively activating different word lines. This merging reduces the total number of bit lines required, thereby reducing chip area while maintaining data storage capacity.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The common bit lines serve multiple functions: they are shared by multiple memory cells for data reading operations. The same bit lines can be used to read from different memory cells at different times by selecting different word lines, making the bit lines universal resources that increase storage capacity without proportionally increasing area.

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

2Use of energy by moving object

If conventional ROM bitcells are used, then the memory array can operate, but power consumption is high due to current leakage

Engineering Contradiction:
Improvepower consumptionVSAvoidcurrent leakage
Core Design Contradiction:
Use of energy by moving objectVSReliability

Solution Approach 1:

The memory array is segmented into multiple blocks, each with its own set of word lines and shared bit lines. This segmentation allows selective activation of only the necessary memory blocks during read operations, reducing the overall current leakage by keeping other blocks in a low-power state rather than having the entire array active simultaneously.

Inventive Principle:
Principle #1Segmentation

3Quantity of substance

If more bit lines are added to increase storage capacity, then data storage capacity increases, but chip area and cost increase

Engineering Contradiction:
Improvedata storage capacityVSAvoidchip area
Core Design Contradiction:
Quantity of substanceVSArea of stationary object

Solution Approach 1:

Multiple memory cells are merged to share common bit lines. Specifically, four memory cells share two common bit lines, allowing the same bit lines to be used for reading data from multiple cells by selectively activating different word lines. This merging reduces the total number of bit lines required, thereby reducing chip area while maintaining data storage capacity.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces a hierarchical addressing scheme that adds a block selection dimension. Instead of directly addressing individual cells, the system first selects a block and then addresses cells within that block. This dimensional change allows the same physical bit lines to access multiple logical memory locations across different blocks, effectively increasing storage capacity without adding proportional physical infrastructure.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS9799409B2Read-only memory (ROM) architecture with selective encoding
Publication Date: 2017.10.24 AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LTD
  • US9799409B2 patent drawing
  • US9799409B2 patent drawing
  • US9799409B2 patent drawing

AI summary

Embodiments provide improved memory bitcells, memory arrays, and memory architectures. In an embodiment, a memory array includes a plurality of memory cells to store data bits. Each of the plurality of memory cells includes a transistor having drain, source, and gate terminals, and a plurality of program nodes, each of the program nodes charged to a predetermined voltage and coupled to a respective one of a plurality of bit lines. For each memory cell in a subset of the plurality of memory cells, none of the plurality of program nodes is coupled to the drain terminal of the transistor to program the each memory cell in the subset of the plurality of memory cells to store at least one data bit, the at least one data bit is most occurred between the data bits.