Dynamic Memory Hopping for Thermal Throttling Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In datacenter and cloud environments, operating at higher ambient temperatures to reduce cooling costs and environmental impact leads to thermal stress and temperature excursions, which can adversely affect system operation, device lifetime, and performance, necessitating effective thermal management strategies to avoid thermal throttling and downtime.
Innovation Solution
Dynamic thermal management through dynamic memory hopping, where data is moved between thermally-affected primary and spare memory devices to defer or avoid thermal throttling, using integrated memory controllers within System on Chip (SoC) or other processors, allowing for balanced power consumption and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is continuously stored on primary memory devices to maintain high performance access, then productivity is improved, but temperature increases causing thermal throttling and reliability degradation
Solution Approach 1:
The patent implements dynamic memory hopping where the system continuously monitors temperature of primary memory devices and dynamically switches data storage between primary and spare memory devices based on thermal conditions. This dynamic adaptation allows the system to maintain high performance when temperatures are normal while preventing thermal throttling by switching to spare devices when temperature thresholds are exceeded, thus resolving the contradiction between maintaining productivity and controlling temperature.
2Temperature
If thermal throttling is activated to control memory temperature, then temperature is reduced, but performance penalty occurs affecting productivity
Solution Approach 1:
The patent employs preliminary action by maintaining spare memory devices in advance and proactively switching data to these spare devices before thermal throttling conditions become severe. The system monitors temperature continuously and performs preventive switching when temperature thresholds are approached, rather than waiting for thermal throttling to activate. This preliminary intervention avoids the performance penalties associated with thermal throttling while still achieving temperature control.
3Reliability
If data is transferred between memory devices to manage thermal stress, then reliability is improved by avoiding thermal throttling, but device complexity increases due to additional control logic
Solution Approach 1:
The patent applies segmentation by dividing the memory system into primary memory devices and spare memory devices, with the memory controller managing distinct temperature monitoring and data switching functions. The controller is segmented into separate modules: temperature monitoring logic, threshold comparison logic, and data switching logic. This modular segmentation makes the complex control function more manageable and maintainable while achieving reliable thermal management through coordinated operation of the segmented components.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In an embodiment, a processor includes at least one core to execute instructions and a memory controller coupled to the at least one core. In turn, the memory controller includes a spare logic to cause a dynamic transfer of data stored on a first memory device coupled to the processor to a second memory device coupled to the processor, responsive to a temperature of the first memory device exceeding a thermal threshold. Other embodiments are described and claimed.