Memory Interface Sparse Encoding for Low-Energy PAM Transmission

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Pulse Amplitude Modulation (PAM) interfaces are vulnerable to crosstalk and power noise due to smaller voltage differences between adjacent symbols, leading to increased energy consumption and inefficiency.

Innovation Solution

Implementing multi-modal sparse encoding techniques that exploit idle periods in channels to apply longer, energy-efficient codes, utilizing 2-level or 3-level sparse encodings with Maximum Transition Avoidance (MTA) to mitigate performance loss and reduce energy consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If PAM uses multiple voltage levels to represent different data symbols, then communication bandwidth is increased, but the interface becomes more vulnerable to crosstalk and power noise

Engineering Contradiction:
Improvecommunication bandwidthVSAvoidsignal robustness
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent changes the voltage level parameters by using 2-level or 3-level sparse encodings instead of conventional 4-level PAM encoding. This reduces the number of voltage transitions and minimizes the impact of crosstalk and power noise, thereby improving signal robustness while maintaining communication bandwidth through efficient use of idle periods

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent exploits idle periods in the communication channel to transmit encoded data. By periodically utilizing these idle periods with sparse encoding, the system achieves reliable communication without continuous high-voltage signaling, reducing vulnerability to crosstalk and power noise while maintaining effective bandwidth

Inventive Principle:
Principle #19Periodic action

2Productivity

If PAM divides voltage into more symbols, then more bits of data are transferred simultaneously, but the voltage difference between adjacent symbols becomes smaller

Engineering Contradiction:
Improvedata transfer rateVSAvoidvoltage discrimination
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent changes the encoding parameters from conventional PAM-4 (4 voltage levels) to sparse encoding with 2 or 3 levels. This increases the voltage difference between adjacent symbols, improving voltage discrimination and reducing errors, while maintaining data transfer rate through efficient exploitation of idle periods in the communication channel

Inventive Principle:
Principle #35Parameter changes

3Productivity

If conventional PAM coding is used, then communication is achieved, but excess energy is consumed and excess power noise is produced

Engineering Contradiction:
Improvecommunication efficiencyVSAvoidenergy consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent changes the coding parameters by implementing sparse encoding that uses fewer voltage levels (2-level or 3-level) compared to conventional PAM-4. This reduces the energy required for voltage transitions and minimizes power noise generation, achieving over 20% energy reduction while maintaining communication efficiency through intelligent use of channel idle periods

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent utilizes idle periods periodically to transmit data using sparse encoding. By concentrating transmissions during these idle periods and avoiding continuous signaling, the system reduces overall energy consumption and power noise production while maintaining effective communication throughput

Inventive Principle:
Principle #19Periodic action

Data Source

PatentUS12602335B2Memory interface with reduced energy transmit mode
Publication Date: 2026.04.14 NVIDIA CORP
  • US12602335B2 patent drawing
  • US12602335B2 patent drawing
  • US12602335B2 patent drawing

AI summary

PAM encoding techniques that leverage unused idle periods in channels between data transmissions to apply longer but more energy-efficient codes. To improve energy savings, multiple sparse encoding schemes may be utilized selectively to fit different sized gaps in the traffic. These approaches may provide energy reductions, for example with memory READ and WRITE traffic, when transferring 4-bit data using 3-symbol sequences.