| CPC G11C 16/26 (2013.01) [G06F 3/0604 (2013.01); G06F 3/0659 (2013.01); G06F 3/0673 (2013.01); G06F 11/1076 (2013.01); G11C 16/3404 (2013.01); G11C 16/0483 (2013.01)] | 20 Claims |
|
10. A method, comprising:
receiving, in a device having memory cells and a calibration circuit, first data representative of first signal and noise characteristics of the memory cells read at first test voltages;
determining, by the device, that an optimized read voltage is outside of a range of the first test voltages;
computing, by the device based on the first signal and noise characteristics, an estimate of the optimized read voltage based on determining a minimum value of a curve derived from the first data and extended from within the range of the first test voltages to outside of the range of the first test voltages;
instructing, by the device in response to the determining that the optimized read voltage is outside of the range of the first test voltages, the calibration circuit to measure, according to the estimate, second signal and noise characteristics of the memory cells;
generate, by utilizing a machine learning model and based on the second signal and noise characteristics, a prediction as to whether the decoding of hard bit data retrieved from the memory cells using the optimized read voltage requires use of soft bit data; and
selectively transmitting soft bit data from the memory cells based on the prediction.
|