Power management for storage controllers
A storage controller includes a plurality of pipeline stages configured to process data. A system clock signal is received that has a system frequency and at least one performance metric is determined for one or more pipeline stages of the plurality of pipeline stages. A first clock signal is generated having a first frequency for operation of a first pipeline stage of the plurality of pipeline stages. Based at least in part on the at least one determined performance metric, a second clock signal is generated having a second frequency for operation of a second pipeline stage of the plurality of pipeline stages. The second frequency is less than the system frequency and may also differ from the first frequency.
1. A storage controller, comprising:
a plurality of pipeline stages configured to process data; and
circuitry configured to:
receive a system clock signal having a system frequency;
determine at least one performance metric for one or more pipeline stages of the plurality of pipeline stages;
generate a first clock signal having a first frequency for operation of a first pipeline stage of the plurality of pipeline stages;
based at least in part on the at least one determined performance metric, generate a second clock signal having a second frequency for operation of a second pipeline stage of the plurality of pipeline stages, wherein the at least one determined performance metric indicates at least one of a throughput of commands processed by a previous pipeline stage relative to the second pipeline stage and a throughput of commands processed by a next pipeline stage relative to the second pipeline stage, and
wherein the second frequency is less than the system frequency and differs from the first frequency; and
operate the first pipeline stage at the first frequency using the first clock signal while operating the second pipeline stage at the second frequency using the second clock signal.
2. The storage controller of claim 1 , wherein the at least one performance metric further indicates at least one of an overall throughput of commands processed by the plurality of pipeline stages, a throughput of commands processed by a current pipeline stage, an overall number of pending commands to be processed by the plurality of pipeline stages, a number of pending commands to be processed by the current pipeline stage, a number of pending commands to be processed by the previous pipeline stage, and a number of pending commands to be processed by the next pipeline stage.
3. The storage controller of claim 1 , wherein the at least one performance metric further indicates at least one of a current throughput of processed commands and an average throughput of processed commands.
4. The storage controller of claim 1 , wherein the circuitry is further configured to:
determine that a performance metric for the second pipeline stage does not satisfy a threshold value; and
in response to determining that the performance metric for the second pipeline stage does not satisfy the threshold value, increase the second frequency to the system frequency.
5. The storage controller of claim 1 , wherein the circuitry is further configured to:
determine that an overall performance metric for the plurality of pipeline stages does not satisfy a threshold value; and
in response to determining that the overall performance metric does not satisfy the threshold value, set the frequency for the clock signal for each pipeline stage of the plurality of pipeline stages to the system frequency.
6. The storage controller of claim 1 , wherein the circuitry is further configured to determine the first frequency using a different number of performance metrics than the number of performance metrics used to determine the second frequency.
7. The storage controller of claim 1 , wherein the circuitry is further configured to determine the first frequency using at least one different type of performance metric than a performance metric used to determine the second frequency.
8. The storage controller of claim 1 , wherein the circuitry is further configured to:
decrease the second frequency by a predetermined frequency;
determine a subsequent performance metric after decreasing the second frequency by the predetermined frequency; and
determine whether to further decrease the second frequency by the predetermined frequency based at least in part on the subsequent performance metric.
9. The storage controller of claim 1 , wherein the circuitry is further configured to:
determine that the plurality of pipeline stages is idle; and
in response to determining that the plurality of pipeline stages is idle, reduce the frequency of the clock signal for each pipeline stage of the plurality of pipeline stages by a predetermined frequency.
10. A method for operating a storage controller, the method comprising:
receiving a system clock signal having a system frequency;
determining at least one performance metric for one or more pipeline stages of a plurality of pipeline stages of the storage controller;
generating, based at least in part on the system frequency, a first clock signal having a first frequency for operation of a first pipeline stage of the plurality of pipeline stages;
generating, based at least in part on the at least one determined performance metric, a second clock signal having a second frequency for operation of a second pipeline stage of the plurality of pipeline stages, wherein the at least one determined performance metric indicates at least one of a throughput of commands processed by a previous pipeline stage relative to the second pipeline stage and a throughput of commands processed by a next pipeline stage relative to the second pipeline stage, and
wherein the second frequency differs from the first frequency; and
operating the first pipeline stage at the first frequency using the first clock signal while operating the second pipeline stage at the second frequency using the second clock signal.
11. The method of claim 10 , wherein the at least one performance metric further indicates at least one of an overall throughput of commands processed by the plurality of pipeline stages, a throughput of commands processed by a current pipeline stage, an overall number of pending commands to be processed by the plurality of pipeline stages, a number of pending commands to be processed by the current pipeline stage, a number of pending commands to be processed by the previous pipeline stage, and a number of pending commands to be processed by the next pipeline stage.
12. The method of claim 10 , wherein the at least one performance metric further indicates at least one of a current throughput of processed commands and an average throughput of processed commands.
13. The method of claim 10 , further comprising:
determining that a performance metric for the second pipeline stage does not satisfy a threshold value; and
in response to determining that the performance metric for the second pipeline stage does not satisfy the threshold value, increasing the second frequency to the system frequency.
14. The method of claim 10 , further comprising:
determining that an overall performance metric for the plurality of pipeline stages does not satisfy a threshold value; and
in response to determining that the overall performance metric does not satisfy the threshold value, setting the frequency of the clock signal for each pipeline stage of the plurality of pipeline stages to the system frequency.
15. The method of claim 10 , further comprising determining the first frequency using a different number of performance metrics than the number of performance metrics used to determine the second frequency.
16. The method of claim 10 , further comprising determining the first frequency using at least one different type of performance metric than a performance metric used to determine the second frequency.
17. The method of claim 10 , further comprising:
decreasing the second frequency by a predetermined frequency;
determining a subsequent performance metric after decreasing the second frequency by the predetermined frequency; and
determining whether to further decrease the second frequency by the predetermined frequency based at least in part on the subsequent performance metric.
18. The method of claim 10 , further comprising:
determining that the plurality of pipeline stages is idle; and
in response to determining that the plurality of pipeline stages is idle, reducing the frequency of the clock signal for each pipeline stage of the plurality of pipeline stages by a predetermined frequency.
19. A storage controller, comprising:
a plurality of pipeline stages configured to process data; and
means for:
receiving a system clock signal having a system frequency;
determining at least one performance metric for one or more pipeline stages of the plurality of pipeline stages;
generating a first clock signal having a first frequency for operation of a first pipeline stage of the plurality of pipeline stages;
based at least in part on the at least one determined performance metric, generating a second clock signal having a second frequency for operation of a second pipeline stage of the plurality of pipeline stages, wherein the at least one determined performance metric indicates at least one of a throughput of commands processed by a previous pipeline stage relative to the second pipeline stage and a throughput of commands processed by a next pipeline stage relative to the second pipeline stage, and
wherein the second frequency is less than the system frequency; and
operating the first pipeline stage at the first frequency while operating the second pipeline stage at the second frequency.
20. The storage controller of claim 19 , wherein the second frequency differs from the first frequency.