Dynamic Thermal Budget Allocation for Memory Arrays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems with multiple memory groups allocate a fixed thermal budget, leading to inefficient use as load distribution is uneven, resulting in wasted thermal budget and limited dynamic performance.
Innovation Solution
A method and system for dynamically allocating thermal budgets to memory groups based on workload, using dynamic thermal budget logic to detect power consumption and adjust credits, thereby redistributing thermal resources to optimize performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If a fixed thermal budget is allocated to each memory group, then thermal stability is ensured, but thermal budget efficiency deteriorates due to uneven workload distribution
Solution Approach 1:
The patent implements dynamic thermal budget allocation where the thermal budget for each memory group is adjusted in real-time based on its current workload. The memory controller monitors workload metrics (such as read/write operations) and dynamically redistributes thermal budget from low-utilization memory groups to high-utilization memory groups, transforming the static fixed allocation into a dynamic adaptive system that maintains thermal stability while improving efficiency
Solution Approach 2:
The system employs feedback mechanisms where the memory controller continuously monitors workload distribution across memory groups and uses this information to adjust thermal budget allocation. The feedback loop detects when a memory group exceeds its thermal budget due to high workload and dynamically reallocates thermal budget from other groups, ensuring thermal stability is maintained while optimizing overall thermal budget utilization
2Productivity
If thermal budget is evenly distributed to all memory groups, then thermal fairness is maintained, but system performance deteriorates due to unused thermal budget in low-load groups
Solution Approach 1:
The patent changes the allocation parameter from fixed equal distribution to dynamic proportional distribution based on workload. The memory controller calculates the thermal budget allocation for each memory group as a proportion of the total thermal budget based on its relative workload, allowing high-load groups to receive more thermal budget and low-load groups to receive less, thereby eliminating thermal budget waste while improving system performance
Solution Approach 2:
The system transitions from static even distribution to dynamic workload-proportional distribution. The thermal budget allocation is continuously adjusted based on real-time workload monitoring, enabling the system to adapt to changing conditions and maximize performance by allocating thermal resources to where they are most needed
3Reliability
If worst-case thermal predictions are used for design, then thermal safety is ensured, but thermal budget utilization deteriorates due to over-provisioning
Solution Approach 1:
The patent implements feedback-based dynamic allocation that replaces worst-case static provisioning with real-time adaptive allocation. The memory controller monitors actual thermal conditions and workload distribution, allocating thermal budget based on actual needs rather than worst-case predictions. This feedback mechanism ensures thermal safety is maintained while dramatically improving thermal budget utilization by eliminating over-provisioning
Solution Approach 2:
The system enables memory groups to effectively self-regulate their thermal budget consumption based on their actual workload. Rather than being constrained by conservative worst-case allocations, memory groups dynamically receive thermal budget according to their real-time needs, with the memory controller automatically adjusting allocations to maintain thermal safety while optimizing utilization
Data Source
AI summary
Embodiments of the present inventive concept relate to systems and methods for dynamically allocating and/or redistributing thermal budget to each memory group in a memory array from a total memory thermal budget based on the workload of each memory group. In this manner, the memory groups having a higher workload can receive a higher thermal budget. The allocation can be dynamically adjusted over time. Thus, the individual and overall memory group performance increases while efficiently allocating the total thermal budget. By dynamically sharing the total thermal budget of the system, the performance of the system as a whole is increased, thereby lowering, for example, the total cost of ownership (TCO) of datacenters.