Inter-Die Message Compression Using Register Field References

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing processors face challenges with single die fabrication defects leading to reduced functionality and increased yield risk, as well as limitations in scalability and flexibility in design and functionality.

Innovation Solution

The use of multiple physically separate dies connected via an interconnect to form a processor, allowing for a monolithic cache domain, reduced latency, and increased bandwidth, with options for different clocking rates and power domains, enabling redundancy and flexibility in design.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a single die is used for processor fabrication, then manufacturing simplicity is maintained, but fabrication yield risk increases due to defects and scalability is limited

Engineering Contradiction:
Improvefabrication yieldVSAvoidprocessor structure
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The processor is divided into multiple physically separate dies, each containing specific functional units (e.g., core die, I/O die, GPU die). These segmented dies are connected via an interconnect network, allowing independent fabrication and testing of each die while maintaining overall processor functionality. This segmentation reduces yield risk as defects in one die do not necessarily compromise the entire processor.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

An interconnect network acts as an intermediary between the multiple physically separate dies, enabling communication and data transfer while maintaining electrical isolation. The interconnect includes protocol converters and buffering mechanisms that facilitate coordinated operation of the distributed processor components without requiring direct physical integration.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If multiple physically separate dies are used, then scalability and flexibility are enhanced, but device complexity increases

Engineering Contradiction:
Improveprocessor configurationVSAvoidinterconnect structure
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The interconnect network is designed as a universal platform that can accommodate multiple types of dies with different functionalities (CPU cores, I/O interfaces, graphics processing). The interconnect uses standardized protocols and interfaces that allow flexible configuration and reconfiguration of processor architectures without requiring redesign of the interconnect infrastructure.

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

Solution Approach 2:

The processor configuration is made dynamic through the ability to selectively activate or deactivate specific dies based on operational requirements. The system can dynamically allocate resources across multiple dies, adjust clocking rates independently for different die, and reconfigure data flow paths through the interconnect to optimize performance for different workloads.

Inventive Principle:
Principle #15Dynamics

3Speed

If multiple dies are connected via interconnect, then bandwidth and latency are improved, but manufacturing precision requirements increase

Engineering Contradiction:
Improvedata transfer rateVSAvoidinterconnect alignment
Core Design Contradiction:
SpeedVSManufacturing precision

Solution Approach 1:

The interconnect design incorporates preliminary alignment features and reference marks that are established during die fabrication. These pre-configured alignment structures enable precise positioning of multiple dies during assembly, reducing the precision requirements for final die-to-die alignment and simplifying the manufacturing process.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The interconnect interface designs are standardized and replicated across multiple die types and generations. By using copied, proven interface designs rather than custom designs for each die pair, the patent reduces manufacturing complexity and precision requirements while maintaining high data transfer rates and low latency performance.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP4614335A1Apparatus and method for scenario-based compression in a processor
Publication Date: 2025.09.10 INTEL CORP
  • EP4614335A1 patent drawingFigure 1
  • EP4614335A1 patent drawingFigure 2A
  • EP4614335A1 patent drawingFigure 2B

AI summary

Techniques for scenario-based compression, including a processor (100), comprising: a first die (102) of a plurality of dies coupled over a first inter-die link (106) to a second die (104), the first die (102) comprising: message control registers to store message fields; message compression circuitry to compress a message transmitted over the first inter-die link (106), the first message compression circuitry to replace a message field in the message with an indication of a storage location of the message field in the first one or more message control registers; the second die (104) comprising: second message control registers to store the message fields; message decompression circuitry to decompress the message received over the first inter-die link (106) from the first die (102), the message decompression circuitry to identify the storage location of the message field in the second one or more message control registers based on the indication and to insert the message field in the message.