Switched CXL Interconnect Abstraction for Any-to-Any Resource Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current PCIe and CXL switches are limited in their ability to modify protocol or physical addresses of packets, and impose restrictions on communication paths, constraining flexibility and resource management in computing systems.
Innovation Solution
Introduce an abstraction for compute nodes and accelerators in a switched CXL interconnect, enabling dynamic modification of protocol and physical addresses, and facilitating any-to-any connectivity between hosts and devices through advanced switching mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If standard PCIe and CXL switches forward packets based on original headers without modification, then protocol compatibility and routing simplicity are maintained, but flexibility in resource management and dynamic communication paths are limited
Solution Approach 1:
The patent introduces an intermediary switching mechanism that can modify packet headers and protocol types between different domains (e.g., converting CXL.mem requests to CXL.cache requests). This intermediary layer enables flexible resource management while maintaining protocol compatibility at each domain boundary, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The switching mechanism dynamically adapts packet routing based on resource availability and communication requirements rather than following fixed predefined paths. The system can dynamically allocate and share resources among different entities coupled to the fabric, enabling flexible resource management while maintaining manageable operation through automated decision-making.
2Productivity
If hierarchical interconnect topologies with predefined communication paths are used, then routing simplicity and protocol adherence are maintained, but dynamic allocation and sharing of resources among different entities is constrained
Solution Approach 1:
The system replaces static hierarchical routing with dynamic packet routing that adapts to real-time resource needs. Packets can be routed through different paths depending on resource availability, enabling efficient dynamic allocation and sharing while maintaining ease of operation through automated routing decisions.
Solution Approach 2:
The switching mechanism serves multiple functions: it can forward packets traditionally, modify protocol types (e.g., CXL.mem to CXL.cache), remap physical addresses, and enable communication between any entities in the fabric regardless of their hierarchical position. This multi-functionality improves resource sharing efficiency while maintaining operational simplicity.
3Adaptability or versatility
If switches cannot modify protocol types or physical addresses, then protocol compliance and address accuracy are maintained, but cross-domain communication and dynamic address remapping are restricted
Solution Approach 1:
The switching mechanism acts as an intermediary that can translate between different protocols (e.g., converting CXL.mem requests to CXL.cache requests) and remap physical addresses while maintaining protocol compliance at each domain boundary. This enables protocol adaptation without compromising reliability, as each protocol conversion is handled carefully to maintain compatibility with the target domain.
Solution Approach 2:
The system modifies packet parameters such as protocol type and physical address while maintaining the structural integrity and compliance of each protocol. By carefully changing parameters like destination address and protocol identifier, the system achieves adaptability across domains while preserving protocol compliance through controlled parameter transformation.
Data Source
AI summary
Embodiments for communicating using a switch having multiple sets of switch ports, which can generate abstractions for compute nodes and accelerators in a switched Compute Express Link (CXL) interconnect, enabling any-to-any connectivity between hosts and devices. A first set of switch ports couples hosts and/or devices with a first virtual to physical binding. A second set couples a Resource Provisioning Unit (RPU) with the first virtual to physical binding. The first binding binds at least some of the first and/or second sets of switch ports to a first set of Virtual CXL Switches (VCSs). A third set couples the RPU to a second virtual to physical binding. A fourth set couples additional hosts and/or devices with the second virtual to physical binding. And the second virtual to physical binding binds at least some of the third and/or fourth sets of switch ports to a second set of VCSs.