Automatic performance tuning for memory arrangements
View Patent ↗An arrangement is disclosed comprising a memory arrangement configured to store and retrieve data; an interface to allow data to be received and transmitted by the arrangement from a host and a processor configured to dynamically conduct automatic performance tuning for the memory arrangement.
1. An arrangement, comprising:
a non-stacked memory arrangement for a solid state drive configured to store and retrieve data, wherein the solid state drive has a workload performance profile mode page;
an interface to allow data to be received and transmitted by the arrangement from a host; and
a processor configured to dynamically conduct automatic performance tuning for the memory arrangement, wherein the automatic performance tuning uses an automatic performance tuning configuration mode parameter and the workload performance profile mode page and a deep learning mode bit configured to reset a machine learning initial starting point and search algorithm adjustments to enable characterization.
2. The arrangement according to claim 1 , wherein the processor is configured to dynamically conduct the automatic performance tuning through an algorithm.
3. The arrangement according to claim 1 , further comprising:
a workload detector configured to monitor a command reception from a host to characterize a workload, the workload detector configured to use the workload performance profile mode page.
4. The arrangement according to claim 1 , further comprising:
a command performance statistics monitor configured to capture command statistics to measure a performance of the arrangement.
5. The arrangement according to claim 1 , wherein the memory arrangement is a portion of a hybrid disk dive.
6. The arrangement according to claim 1 , wherein the host is one of a computer and a server.
7. A method for altering a performance of a solid state drive, comprising:
selecting at least one performance profile target;
detecting a workload to be accomplished by the solid state drive by using a workload performance profile mode page;
loading initial starting values for an automatic performance tuning algorithm using an automatic performance tuning configuration mode parameter;
configuring a command performance statistics monitor to measure workload input and output performance;
measuring performance of the solid state drive with the command performance statistics monitor;
comparing the performance measurement of the solid state drive with the at least one performance profile target; and
adjusting performance of the solid state drive through the automatic performance tuning algorithm when the comparing of the performance measurement of the solid state drive with the at least one performance profile target indicates that the performance of the solid state drive is less than the at least one performance profile target.
8. The method according to claim 7 , wherein the selecting the at least one performance profile target is through a host.
9. The method according to claim 8 , wherein the selecting the host is one of a computer and a server.
10. The method according to claim 7 , wherein the detecting of the workload is performed by a workload detector.
11. The method according to claim 7 , wherein the selecting the at least one performance profile target is an average input and output performance operations.
12. The method according to claim 11 , wherein the measuring performance of the solid state drive with the command performance statistics monitor determines an average input and output performance operations.
13. The method according to claim 12 , further comprising:
determining if the measured performance of the solid state drive input and output performance operations is within a desired threshold of the average input and output performance operations profile target.
14. The method according to claim 13 , further comprising:
adjusting parameters of the automatic performance tuning algorithm when the measured performance of the solid state drive input and output performance operations is not within the desired threshold of the average input and output performance operations profile target.
15. The method according to claim 13 , further comprising:
fine tuning the automatic performance tuning when the measured performance of the solid state drive input and output performance operations is within the desired threshold of the average input and output performance operations profile target.
16. The method according to claim 15 , further comprising:
disabling the automatic performance tuning algorithm.
17. The method according to claim 13 , further comprising:
evaluating automatic performance tuning performance for at least one parameter compared to a threshold.
18. An arrangement, comprising:
means for selecting at least one performance profile target;
means for detecting a workload to be accomplished by the solid state drive;
means for loading initial starting values for an automatic performance tuning algorithm;
means for configuring a command performance statistics monitor to measure workload input and output performance;
means for measuring performance of the solid state drive with the command performance statistics monitor;
means for comparing the performance measurement of the solid state drive with the at least one performance profile target; and
means for adjusting performance of the solid state drive through the automatic performance tuning algorithm when the comparing of the performance measurement of the solid state drive with the at least one performance profile target indicates that the performance of the solid state drive is less than the at least one performance profile target wherein the automatic performance tuning uses an automatic performance tuning configuration mode parameter and the workload performance profile mode page and a deep learning mode bit.
19. The arrangement according to claim 18 , further comprising:
means for adjusting parameters of the automatic performance tuning algorithm when the measured performance of the solid state drive input and output performance operations is not within the desired threshold of the average input and output performance operations profile target.
20. A method, comprising:
selecting at least one performance profile target;
detecting a workload to be accomplished by the solid state drive;
loading initial starting values for an automatic performance tuning algorithm, the initial starting values obtained from an automatic performance tuning log;
configuring a command performance statistics monitor to measure workload input and output performance;
measuring performance of the solid state drive with the command performance statistics monitor;
comparing the performance measurement of the solid state drive with the at least one performance profile target; and
adjusting performance of the solid state drive through the automatic performance tuning algorithm when the comparing of the performance measurement of the solid state drive with the at least one performance profile target indicates that the performance of the solid state drive is less than the at least one performance profile target wherein the automatic performance tuning uses an automatic performance tuning configuration mode parameter and the workload performance profile mode page and a deep learning mode bit.
21. The method according to claim 20 , further comprising:
generating a second automatic performance tuning log.
22. The method according to claim 21 , wherein the loading of the initial starting values for the automatic performance tuning algorithm obtained from the automatic performance tuning log is from the second automatic performance tuning log.