Burst Command Encoding for NoC Link Width Adaptation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current packet-based communication systems in Network on Chip (NoC) architectures face inefficiencies due to the need for re-encoding and latency incurred during link width changes, as they require packet inspection and modification to adapt to varying link widths, especially when transitioning between different link widths.

Innovation Solution

The implementation of burst command encoding that is independent of data word width, allowing packets to be upsized or downsized without re-encoding, using static burst size and wrapping border information, which enables seamless communication across links of different widths without the need for packet inspection or modification, utilizing upconversion and downconversion logic within routers and network interfaces.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If packet inspection and modification is performed to adapt to varying link widths, then communication compatibility across different link widths is achieved, but system complexity and latency increase

Engineering Contradiction:
Improvecommunication compatibilityVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The packet is segmented into header and payload portions, with the header containing fixed-width field identifiers. This segmentation allows the header to be processed independently of the payload width, enabling width adaptation without re-inspecting the entire packet structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Fixed-width field identifiers are embedded in the packet header before transmission. This preliminary encoding of width-independent addressing information allows receiving devices to automatically route packets across links of varying widths without requiring runtime inspection or modification of the packet contents.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If packet inspection and modification is performed to adapt to varying link widths, then communication compatibility across different link widths is achieved, but transmission latency increases

Engineering Contradiction:
Improvecommunication compatibilityVSAvoidtransmission latency
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

Width-independent field identifiers are encoded into the packet header in advance at the source device. This preliminary action eliminates the need for runtime packet inspection and modification at intermediate routers, thereby reducing transmission latency while maintaining compatibility across varying link widths.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The packet header contains self-describing field identifiers that enable receiving devices to automatically interpret and route packets without requiring external control signals or inspection protocols. This self-service mechanism reduces latency by eliminating handshaking and inspection overhead.

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If re-encoding is performed during link width changes, then packet adaptability to different link widths is achieved, but additional storage requirements increase

Engineering Contradiction:
Improvepacket adaptabilityVSAvoidstorage requirements
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The packet structure is segmented into a header portion containing fixed-width field identifiers and a payload portion. This segmentation allows the header to remain unchanged during width transitions, eliminating the need for storage buffers to hold modified packet contents while enabling payload width adaptation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of modifying packet contents to adapt to width changes, the system inverts the approach by using fixed-width identifiers that inherently describe the packet structure. This inversion eliminates the need for re-encoding and associated storage requirements, as the identifiers themselves carry the adaptation information.

Inventive Principle:
Principle #13The other way round (Inversion)

Data Source

PatentEP3231143B1A method, apparatus and system for encoding command information in a packet-based network
Publication Date: 2021.08.11 INTEL CORP
  • EP3231143B1 patent drawingFigure 1
  • EP3231143B1 patent drawingFigure 2
  • EP3231143B1 patent drawingFigure 3

AI summary

In one embodiment, an apparatus comprises: a source agent including at least one logic unit to perform instructions; an encoder to encode a burst command portion of a packet having a first field to indicate a burst size and a second field to indicate a data width of one of the source agent and a destination agent, where the burst size and the data width are to remain fixed when the packet is to be re-sized one or more times during transmission from the source agent to the destination agent; and transmission logic to transmit the packet including the burst command portion. Other embodiments are described and claimed.