Dynamic Memory Frequency Voltage Scaling via H-States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory power management in electronic devices is inefficient due to static configuration of memory frequency and voltage, which fails to adapt to workload sensitivity, leading to suboptimal energy efficiency and performance.
Innovation Solution
Implementing dynamic memory frequency/voltage scaling techniques, known as H-states, within the memory controller to adjust operational states based on workload sensitivity, using a Memory Scaling Factor (MSF) to select the appropriate frequency and voltage settings, and providing an interface for operating system interaction to support adaptive power management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If memory frequency and voltage are statically configured, then device complexity is reduced and ease of operation is improved, but energy efficiency deteriorates and performance adaptability worsens
Solution Approach 1:
The patent implements dynamic memory frequency and voltage scaling by introducing multiple H-states (memory operation states) that allow the memory system to transition between different frequency-voltage combinations based on workload demands. The memory controller dynamically selects appropriate H-states using a Memory Scaling Factor (MSF) that captures workload sensitivity to memory latency, enabling the system to adapt memory operational parameters in real-time rather than using static configuration.
Solution Approach 2:
The patent changes physical parameters (frequency and voltage) of the memory system to optimize energy efficiency. By defining H-states with different frequency-voltage pairs and dynamically transitioning between them based on workload characteristics captured by the MSF, the system adjusts operational parameters to match demand, reducing energy consumption during low-demand periods while maintaining performance when needed.
2Use of energy by moving object
If memory frequency is reduced to save energy, then energy consumption decreases, but memory latency increases and performance deteriorates
Solution Approach 1:
The system dynamically transitions between H-states with different frequency-speed characteristics based on real-time workload assessment. When workloads are latency-tolerant, the system operates at lower frequencies to save energy; when workloads are latency-sensitive, the system transitions to higher frequency H-states to maintain performance, thus dynamically balancing power consumption and speed.
Solution Approach 2:
The patent utilizes frequency-voltage scaling by changing the operational frequency parameter of the memory system. Each H-state represents a specific frequency-voltage combination, allowing the system to select appropriate frequency levels based on workload requirements, thereby adjusting speed parameter dynamically to optimize the trade-off between power consumption and performance.
3Loss of energy
If aggressive power-down states are used, then idle memory power is reduced, but memory responsiveness and system performance worsen
Solution Approach 1:
The system dynamically manages memory power states by transitioning between active H-states and power-down states based on workload activity and latency sensitivity. Rather than using aggressive static power-down states, the system assesses workload characteristics using MSF and selectively transitions to power-saving states only when appropriate, maintaining responsiveness for latency-sensitive workloads while achieving power savings for tolerant workloads.
Data Source
AI summary
Described herein are techniques for dynamic memory frequency/voltage scaling to augment existing memory power management techniques and further improve memory power efficiency. Each operating point is defined as an operational state for the memory.