NDN Workload Offloading Across Specialized Accelerators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing NFD technologies perform workload requests using only the core accelerator without considering the capabilities of other accelerators, leading to inefficiencies in memory speed, processing time, and power consumption.
Innovation Solution
An NFD module offloads workload requests to other accelerators based on their capabilities, utilizing APIs and different interfaces to enhance efficiency by leveraging specialized hardware features such as caching and interest forwarding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If workload requests are performed solely by the core accelerator, then the system maintains simplicity in task distribution, but memory speed, processing time, and power consumption are suboptimal
Solution Approach 1:
The patent segments the workload requests into different types (caching operations, interest forwarding, data transmission) and assigns them to specialized accelerators based on their capabilities. The NFD module divides the processing tasks and distributes them across multiple accelerators including GPU, FPGA, ASIC, and CPU, allowing each to operate in its optimal performance regime.
Solution Approach 2:
The system dynamically selects which accelerator handles each workload request based on real-time capability matching. The NFD module evaluates the specific requirements of each interest packet and dynamically routes it to the most suitable accelerator, enabling adaptive optimization of processing speed, memory access, and power consumption for different task types.
2Loss of time
If workload requests are performed solely by the core accelerator, then the system maintains simple processing logic, but processing time and power consumption increase
Solution Approach 1:
The NFD module acts as an intermediary between incoming interest packets and the various accelerators. It receives workload requests, determines the appropriate accelerator based on capability matching, and forwards requests accordingly. This intermediary layer handles the complexity of task distribution while presenting a simplified interface to both the network protocol and the accelerators.
Solution Approach 2:
The system implements a universal task distribution mechanism that can handle multiple types of workload requests (caching, forwarding, transmission) across different accelerator types. The NFD module provides multi-functional capability by supporting various NDN protocols and interfacing with diverse hardware accelerators through a unified framework.
3Speed
If the core accelerator handles all workload requests, then the system structure remains simple, but memory speed and power efficiency are reduced
Solution Approach 1:
The patent applies local quality by matching specific workload characteristics with corresponding accelerator capabilities. Caching operations are directed to accelerators with high memory access speed, interest forwarding to those with low latency, and data transmission to those with high bandwidth. Each accelerator is configured with the specific quality needed for its designated task type, optimizing overall system performance.
Solution Approach 2:
The system changes operational parameters by selecting different accelerators based on the specific requirements of each workload request. Parameters such as memory access speed, processing throughput, and power consumption are optimized by matching the appropriate accelerator type (GPU, FPGA, ASIC, CPU) to the task at hand, rather than using a single core accelerator for all operations.
Data Source
AI summary
A device may include a processor that includes a named data network (NDN) forwarding daemon (NFD) module. The NFD module may be configured to identify an accelerator on which to configure an offloaded NFD module based on a configuration table indicating a current configuration of the accelerator. The NFD module may be configured to configure the offloaded NFD module on the identified accelerator. The NFD module may be configured to receive an interest packet that includes a workload request. The interest packet may be configured according to an NDN protocol. The NFD module may be configured to determine that the offloaded NFD module is configured to perform the workload request using the identified accelerator based on the configuration table. The NFD module may be configured to offload, via an application programming interface, the workload request to the offloaded NFD module to perform the workload request using the identified accelerator.