Disaggregated Accelerator Resource Threshold Control for Shared Workloads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing accelerator devices face inefficiencies in dynamically reallocating shared resources like high bandwidth memory, leading to delays and suboptimal performance due to static allocation methods, which hinder the ability to adapt to changing workload demands.
Innovation Solution
Implementing a system with an orchestrator server and dynamic resource allocation logic within accelerators that monitor and adjust resource utilization thresholds in real-time, allowing for the efficient redistribution of resources among logic portions without interrupting workload execution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If static allocation of shared resources is used among logic portions, then resource allocation is simple and stable, but resource utilization efficiency deteriorates when workload demands change
Solution Approach 1:
The patent implements dynamic resource allocation by allowing logic portions to access shared resources allocated to other logic portions when those resources are not currently being used. This is achieved through a resource allocation mechanism that monitors resource usage and dynamically grants access permissions without requiring system reconfiguration, thereby improving resource utilization efficiency while maintaining operational stability.
2Adaptability or versatility
If reconfiguration of logic portions is performed to reallocate shared resources, then resource allocation adapts to workload needs, but execution time increases due to stopping and resuming workloads
Solution Approach 1:
The patent establishes resource allocation rules and access permissions in advance, before workload execution begins. The system pre-configures which logic portions can access which shared resources under specific conditions, allowing dynamic resource sharing during runtime without interrupting workload execution. This eliminates reconfiguration delays while maintaining adaptability to workload demands.
Solution Approach 2:
The patent enables continuous workload execution by implementing a resource allocation mechanism that operates transparently during runtime. Logic portions can dynamically access shared resources without stopping or pausing their workload execution, ensuring continuous useful action while adapting resource allocation to changing workload demands.
3Productivity
If more shared resources are allocated to a logic portion, then performance of that logic portion improves, but available resources for other logic portions decrease
Solution Approach 1:
The patent implements a universal shared resource pool where resources serve multiple logic portions simultaneously based on dynamic needs. Instead of dedicating exclusive resources to each logic portion, the system allows any logic portion to access any shared resource when it is not currently being used by another logic portion, maximizing the utility of available resources while maintaining high performance for all logic portions.
Data Source
AI summary
Technologies for dynamically managing resources in disaggregated accelerators include an accelerator. The accelerator includes acceleration circuitry with multiple logic portions, each capable of executing a different workload. Additionally, the accelerator includes communication circuitry to receive a workload to be executed by a logic portion of the accelerator and a dynamic resource allocation logic unit to identify a resource utilization threshold associated with one or more shared resources of the accelerator to be used by a logic portion in the execution of the workload, limit, as a function of the resource utilization threshold, the utilization of the one or more shared resources by the logic portion as the logic portion executes the workload, and subsequently adjust the resource utilization threshold as the workload is executed. Other embodiments are also described and claimed.