DRAM Refresh Rate Control Using Battery Voltage
Original Publication Date: 2000-May-01
Included in the Prior Art Database: 2003-Jun-20
Disclosed is a system capable of controlling the data retention power of Dynamic Random Access Memory (DRAM) by changing its refresh rate as a function of battery voltage in a more intelligent and adaptive manner. Many battery-operated computer devices, such as Personal Digital Assistance (PDA) devices, recently use DRAM as their persistent storage. Because DRAM needs to be periodically refreshed to retain its contents, users have an "empty-battery" risk, that is the risk of losing the entire contents in the persistent storage when battery becomes empty. On the other hand, since the refresh operation drains some current, the battery power consumption is reduced by reducing the DRAM refresh rate at "potential retention failure" risk, the risk of losing a very small part of the contents. Our idea proposes to balance and optimize these two inherent risks in the DRAM-based persistent storage. This invention provides a system that controls DRAM refresh rate, in a risk-optimized manner, as a function of the battery voltage. A typical implementation of the proposed system consists of DRAM, a DRAM control circuit (responsible for the refresh operation), and a battery, and a battery voltage measurement unit. The DRAM control circuit periodically asks to measure the battery voltage, and chooses an optimal value of the DRAM refresh rate. When the battery voltage becomes lower than a specified value, indicating that "empty-battery" risk becomes higher than "potential retention failure" risk, the system chooses a lower DRAM refresh rate to balance the above-mentioned two risks. The system can use a table of "potential retention failure" risk as a function of the DRAM refresh rate, either pre-defined or generated through on-the-use measurement, in order to optimize the selection of the DRAM refresh rate. Instead of measuring the battery voltage periodically, the refresh rate adjustment may be triggered through interrupts, such as low-battery-alert interrupts, asserted by an intelligent battery system. In addition, users of the system may set their preference. For example, for a user who may prefer a longer battery life even though "potential retention failure" risk is higher, the system can start to reduce the refresh rate at a much higher voltage. Moreover, the system may measure and consider other factors, such as temperature, Furthermore, the present mechanism may be combined with other mechanisms, such as Error Correction Coding in order to reduce "potential retention failure" risk.