History-based prediction modeling of solid-state device temperature
Aspects of a storage device are provided that apply history-based prediction modeling in advanced thermal throttling. Initially, a controller determines a temperature prediction based one or more thermal mitigation parameters using a history-based prediction model. Subsequently, the controller determines whether the temperature prediction indicates that an actual temperature of the memory is expected to meet a thermal throttling threshold of a plurality of thermal throttling thresholds. The controller then transitions into a thermal power state of a plurality of thermal power states when the temperature prediction indicates that the actual temperature of the memory is expected to meet the thermal throttling threshold. The controller applies a thermal mitigation configuration associated with the thermal power state and determines that the temperature of the memory has reached a thermal equilibrium based on the thermal mitigation configuration. Storage device performance is thus improved through history-based prediction modeling without compromising data integrity.
1. A storage device, comprising:
a memory; and
a controller coupled to the memory and configured to:
determine a temperature prediction based on one or more thermal mitigation parameters using a history-based prediction model, the one or more thermal mitigation parameters including a queue depth,
determine whether the temperature prediction indicates that an actual temperature of the memory is expected to meet a thermal throttling threshold of a plurality of thermal throttling thresholds,
transition into a thermal power state of a plurality of thermal power states in response to the temperature prediction indicating that the actual temperature of the memory is expected to meet the thermal throttling threshold,
apply a thermal mitigation configuration associated with the thermal power state,
determine whether the actual temperature of the memory reaches a thermal equilibrium in the thermal power state in response to the thermal mitigation configuration, and
refrain from further applying the thermal mitigation configuration in response to the actual temperature of the memory reaching the thermal equilibrium.
2. The storage device of claim 1 , wherein the history-based prediction model is communicatively coupled to the controller, and the history-based prediction model includes a multivariate regression model communicatively coupled to an energy prediction model and an ambient temperature prediction model.
3. The storage device of claim 2 , wherein the multivariate regression model includes one or more neural networks that are trained with historical power information and historical temperature information of the storage device.
4. The storage device of claim 2 , wherein the energy prediction model is configured to receives the one or more thermal mitigation parameters indicating a power budget parameter value, and the energy prediction model is configured to produce an energy prediction value that is fed to the multivariate regression model.
5. The storage device of claim 2 , wherein the ambient temperature prediction model is configured to receive an actual ambient temperature value along with the one or more thermal mitigation parameters indicating a queue depth parameter value, wherein the ambient temperature prediction model is configured to produce a temperature prediction value that is fed to the multivariate regression model.
6. The storage device of claim 1 , wherein the history-based prediction model is configured to send, to a host device, a host warning signal indicating that one or more of the plurality of thermal throttling thresholds have been met.
7. The storage device of claim 1 , wherein the controller is further configured to feed the temperature prediction back into the history-based prediction model and calibrate the history-based prediction model based on a difference between the temperature prediction and the actual temperature.
8. A storage device, comprising:
a memory; and
a controller coupled to the memory and configured to:
determine a temperature prediction based on one or more thermal mitigation parameters using a history-based prediction model, the one or more thermal mitigation parameters including a bus interface parameter,
determine whether the temperature prediction indicates that an actual temperature of the memory is expected to meet a thermal throttling threshold of a plurality of thermal throttling thresholds,
apply a thermal mitigation configuration associated with a thermal power state of a plurality of thermal power states in response to the temperature prediction indicating that the actual temperature of the memory is expected to meet the thermal throttling threshold,
determine whether the actual temperature of the memory reaches a thermal equilibrium in the thermal power state in response to the thermal mitigation configuration, and
refrain from further applying the thermal mitigation configuration in response to the actual temperature of the memory reaching the thermal equilibrium.
9. The storage device of claim 8 , wherein the history-based prediction model is communicatively coupled to the controller, and the history-based prediction model includes a multivariate regression model communicatively coupled to an energy prediction model and an ambient temperature prediction model.
10. The storage device of claim 9 , wherein the multivariate regression model includes one or more neural networks that are trained with historical power information and temperature information of the storage device.
11. The storage device of claim 9 , wherein the energy prediction model is configured to receive the one or more thermal mitigation parameters indicating a power budget parameter value, and the energy prediction model is configured to produce an energy prediction value that is fed to the multivariate regression model.
12. The storage device of claim 9 , wherein the ambient temperature prediction model is configured to receive an actual ambient temperature value along with the one or more thermal mitigation parameters indicating of a bus interface parameter value, wherein the ambient temperature prediction model is configured to produce a temperature prediction value that is fed to the multivariate regression model.
13. The storage device of claim 8 , wherein the history-based prediction model is configured to send, to a host device, a host warning signal indicating that one or more of the plurality of thermal throttling thresholds have been met.
14. The storage device of claim 8 , wherein the controller is further configured to feed the temperature prediction back into the history-based prediction model and calibrate the history-based prediction model based on a difference between the temperature prediction and the actual temperature.
15. A storage device, comprising:
a memory having a block of cells; and
a controller coupled to the memory and configured to:
determine a temperature prediction based on one or more thermal mitigation parameters using a history-based prediction model, the one or more thermal mitigation parameters including a bus interface parameter or a queue depth,
determine whether the temperature prediction indicates that an actual temperature of the block of cells is expected to meet a thermal throttling threshold of a plurality of thermal throttling thresholds,
apply a thermal mitigation configuration associated with a thermal power state of a plurality of thermal power states in response to the temperature prediction indicating that the actual temperature of the block of cells is expected to meet the thermal throttling threshold,
determine whether the actual temperature of the memory reaches a thermal equilibrium in the thermal power state based on the thermal mitigation configuration, and
refrain from further applying the thermal mitigation configuration in response to the actual temperature of the memory reaching the thermal equilibrium.
16. The storage device of claim 15 , wherein the history-based prediction model is communicatively coupled to the controller, and the history-based prediction model includes a multivariate regression model communicatively coupled to an energy prediction model and an ambient temperature prediction model.
17. The storage device of claim 16 , wherein the energy prediction model is configured to receive the one or more thermal mitigation parameters indicating a power budget parameter value, and the energy prediction model is configured to produce an energy prediction value that is fed to the multivariate regression model.
18. The storage device of claim 16 , wherein the ambient temperature prediction model is configured to receive an actual ambient temperature value along with the one or more thermal mitigation parameters indicating one or more of a bus interface parameter value or a queue depth parameter value, wherein the ambient temperature prediction model is configured to produce a temperature prediction value that is fed to the multivariate regression model.
19. The storage device of claim 15 , wherein the history-based prediction model is configured to send, to a host device, a host warning signal indicating that one or more of the plurality of thermal throttling thresholds have been met.
20. The storage device of claim 15 , wherein the controller is further configured to feed the temperature prediction back into the history-based prediction model and calibrate the history-based prediction model based on a difference between the temperature prediction and the actual temperature.