PCIe Shunt Circuit for Direct Accelerator Linking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing configuration of connecting operation accelerators to a central processing unit (CPU) through PCIe links leads to severe traffic congestion and performance bottlenecks due to shared communication pathways among PCIe devices, such as operation accelerators, network controllers, and storage medium access controllers.
Innovation Solution
The implementation of a shunt circuit directly connected to PCIe devices through PCIe links, which translates addresses and transmits data through an internal bus, reducing the need for data to pass through the CPU and allowing direct communication between operation accelerators and PCIe devices, thereby alleviating traffic congestion and bandwidth conflicts on the PCIe link.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple PCIe devices (operation accelerator, network controller, storage medium access controller) communicate through a common PCIe link, then system integration is achieved, but severe traffic congestion occurs on the PCIe link
Solution Approach 1:
The patent segments the PCIe communication path by introducing a shunt circuit that creates separate direct connection paths between PCIe devices and the operation accelerator, bypassing the congested common PCIe link. This segmentation allows different data traffic to flow through different paths, resolving the traffic congestion while maintaining system integration.
Solution Approach 2:
The shunt circuit acts as an intermediary component that enables direct communication between PCIe devices and the operation accelerator without requiring all traffic to pass through the CPU. This intermediary structure reduces the burden on the common PCIe link and improves overall system throughput.
2Ease of operation
If all PCIe devices communicate through the CPU, then centralized control is maintained, but the CPU becomes a performance bottleneck
Solution Approach 1:
The communication architecture is segmented into two paths: centralized CPU control for management tasks and direct peer-to-peer paths for data transfer. The shunt circuit enables the latter, allowing data-intensive operations to bypass the CPU and proceed directly between PCIe devices and the operation accelerator.
Solution Approach 2:
The patent extracts the data transmission function from the CPU by introducing direct communication paths through the shunt circuit. This extraction allows the CPU to focus on control functions while data transfer is handled more efficiently through dedicated paths.
3Reliability
If data transmission passes through multiple components (CPU, PCIe link), then system coordination is achieved, but transmission delay increases
Solution Approach 1:
The patent extracts unnecessary intermediate components from the critical data transmission path. By enabling direct communication between PCIe devices and the operation accelerator through the shunt circuit, the path is shortened and transmission delay is reduced while system coordination is maintained through the controller's arbitration logic.
Data Source
AI summary
Examples in this application disclose an operation accelerator, a switch, and a processing system. One example operation accelerator includes a shunt circuit directly connected to a first peripheral component interconnect express (PCIe) device through a PCIe link. The shunt circuit is configured to receive first data sent by the first PCIe device through the PCIe link, and transmit the first data through an internal bus. A first address carried in the first data is located in a first range. In some examples of this application, the first PCIe device directly communicates with the operation accelerator through the shunt circuit in the operation accelerator.