Tiered Memory Device Selection Using Weighted Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Tiered memory systems fail to differentiate between different types of memory devices within the same tier, leading to suboptimal performance due to disregarding differences in parameters like latency, bandwidth, and utilization.
Innovation Solution
A tiered memory system selects a specific memory device within a tier based on parameters such as latency, bandwidth, and utilization, using weighted distance metrics (WDM) to allocate memory pages, allowing user control over parameter importance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If memory devices in the same tier are treated uniformly without differentiation, then device complexity is reduced, but memory performance deteriorates due to ignoring latency, bandwidth, and utilization differences
Solution Approach 1:
The patent applies local quality by treating different memory devices within the same tier differently based on their specific characteristics. Instead of uniform management, the system evaluates individual device parameters (latency, bandwidth, utilization) and allocates memory pages to specific devices based on these local qualities, thereby optimizing performance while maintaining manageable complexity through automated evaluation.
2Speed
If memory pages are allocated without considering device parameters, then allocation speed is improved, but resource utilization deteriorates
Solution Approach 1:
The patent implements preliminary action by pre-evaluating and establishing performance metrics for each memory device before memory page allocation occurs. The system maintains performance data structures that contain pre-computed information about device latency, bandwidth, and utilization, allowing rapid allocation decisions without real-time computation overhead, thus maintaining both speed and utilization efficiency.
3Productivity
If device selection within tiers is based on multiple parameters with weights, then memory performance is improved, but control complexity increases
Solution Approach 1:
The patent applies parameter changes by introducing a weighted evaluation system that transforms multiple device parameters (latency, bandwidth, utilization) into a single composite score. By allowing user-configurable weights for each parameter, the system flexibly adjusts the importance of different factors without increasing structural complexity, maintaining performance optimization while keeping control mechanisms adaptable and manageable.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method may include receiving a request for a memory page in a memory tier comprising a first memory device and a second memory device, wherein the first memory device has a first parameter and the second memory device has a second parameter, selecting, based on the first parameter and the second parameter, the first memory device, and allocating, based on the request, based on the selecting, the memory page from the first memory device. The selecting may include determining a first result based on the first parameter, determining a second result based on the second parameter, and comparing the first result and the second result. The determining the first result may include combining the first parameter with a first weight. The first weight may include a first scale factor, and the combining the first parameter with the first weight may include multiplying the first parameter and the first scale factor.