Memory Drive Strength and ODT Training for DVFS Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current system-on-chip (SoC) technologies face challenges in optimizing memory device performance across varying operating frequencies, leading to inefficient power consumption and stability issues due to the lack of effective training methods for dynamic voltage and frequency scaling (DVFS) schemes.
Innovation Solution
A method for training memory devices by performing initialization operations, measuring and storing configurable operating parameters at different frequencies, and using optimized parameters based on current operation modes and frequencies to enhance stability and reduce power consumption through dynamic adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If dynamic voltage and frequency scaling (DVFS) is implemented without effective training methods, then power consumption is reduced through frequency scaling, but memory device stability deteriorates due to lack of optimization at varying frequencies
Solution Approach 1:
The patent performs training operations in advance to determine optimized operating parameters for multiple operating frequencies before actual DVFS operation. The memory device and host execute training sequences that measure data valid windows and determine optimal parameters for each frequency, storing these as training data. This preliminary action ensures that when DVFS dynamically changes frequency, the device can immediately switch to pre-determined optimized parameters, maintaining stability while achieving power savings.
2Adaptability or versatility
If operating parameters are fixed for all frequencies, then device complexity is reduced, but adaptability to different operating frequencies deteriorates
Solution Approach 1:
The patent segments the operating parameter configuration into frequency-specific optimized parameters determined through training operations. Instead of using a single fixed parameter set or complex real-time optimization, the system divides the parameter space by operating frequency, pre-determining optimal parameters for each frequency level. This segmentation allows the memory device to adapt to different frequencies by simply selecting the appropriate pre-determined parameter set, achieving high adaptability without excessive complexity.
3Reliability
If training operations are performed for all operating frequencies, then operating parameter optimization is improved, but initialization time increases
Solution Approach 1:
The patent performs training operations for multiple operating frequencies during initialization, but the training data determined through these operations is stored and reused for subsequent DVFS operations. The full training process executes once during initialization to establish optimized parameters across the frequency range, rather than performing training continuously. This partial action approach achieves comprehensive parameter optimization while limiting the time penalty to a single initialization phase.
Data Source
AI summary
In a method of training for a memory device, an initialization operation is performed on the memory device when the memory device is powered on. A training operation is performed on a plurality of operating frequencies of the memory device such that at least one of a plurality of operating parameters of the memory device is obtained as a configurable operating parameter for each of the plurality of operating frequencies. The configurable operating parameter for each of the plurality of operating frequencies is stored as training data. An optimized operating parameter for the memory device is used based on the training data, a current operation mode of the memory device, and a current operating frequency of the memory device.