Dynamic cache bypass for power saving

By responding to power-down triggers, the last level cache optimizes power consumption when inefficient data access is available, and power is re-energized in beneficial situations, solving the problem of power waste in the cache inefficient situations in the prior art, and improving the Energy efficiency performance of the system.

CN119096225APending Publication Date: 2024-12-06ADVANCED MICRO DEVICES INC +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380036986.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-26
Filing Date
2023-04-20
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

In the prior art, the last level cache has waste in terms of data access latency and power consumption, especially when the data access latency is not significant or the hit rate is low, resulting in low cache effectiveness and thus waste of power.

Method used

By responding to power-down triggers, the last level caches power-down and directly serves the memory access transactions by the memory to avoid unnecessary power consumption, while providing the requester when the data exists, otherwise data is retrieved from the memory, ensuring that it is low Optimize power usage under profit situations.

Benefits of technology

Effectively reduces power consumption in inefficient data access, improves the system's performance in low-power mode, adapts to different workload characteristics, and ensures that power-on re-energies are beneficial in the case of cache.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119096225A_ABST
    Figure CN119096225A_ABST
Patent Text Reader

Abstract

A technique for operating a cache is disclosed. The technique includes powering down the cache in response to a power down trigger indicating that the cache validity is considered low.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. patent application No. 17 / 730,041, filed on April 26, 2022, which is incorporated by reference as if fully set forth herein. Background Art

[0003] Caching improves performance by storing copies of data that is considered likely to be accessed again in the future in low-latency cache memory. Improvements in cache technology are continually being made. BRIEF DESCRIPTION OF THE DRAWINGS

[0004] A more detailed understanding may be obtained from the following description, given by way of example in conjunction with the accompanying drawings, in which:

[0005] Figure 1 is a block diagram of an example computing device that may implement one or more features of the present disclosure;

[0006] Figure 2 and Figure 3 illustrating operations for powering down a last level cache in response to a power down trigger according to an example;

[0007] Figure 4A and Figure 4B illustrating operation of a computing device when a last level cache is powered down and when the last level cache is powered up according to an example;

[0008] FIG. 5A to FIG. 5D illustrating a technique for powering down a last level cache according to an example; and

[0009] Figure 6 is a method for operating a last level cache according to an example. DETAILED DESCRIPTION

[0010] A technique for operating a cache is disclosed. The technique includes powering down the cache in response to a power down trigger indicating that the cache validity is considered low.

[0011] Figure 11 is a block diagram of an example computing device 100 that may implement one or more features of the present disclosure. In various examples, computing device 100 is, for example, but not limited to, one of a computer, a gaming device, a handheld device, a set-top box, a television, a mobile phone, a tablet computer, or other computing device. Device 100 includes, but is not limited to, one or more processors 102, a memory 104, one or more auxiliary devices 106, a storage device 108, and a last level cache ("LLC") 110. An interconnect 112, which may be a bus, a combination of busses, and / or any other communication component, communicatively links one or more processors 102, memory 104, one or more auxiliary devices 106, a storage device 108, and last level cache 110.

[0012] In various alternatives, the one or more processors 102 include a central processing unit (CPU), a graphics processing unit (GPU), a CPU and a GPU on the same die, or one or more processor cores, where each processor core may be a CPU, a GPU, or a neural processor. In various alternatives, at least a portion of the memory 104 is located on the same die as one or more of the one or more processors 102, such as on the same chip or in an interposer arrangement, and / or at least a portion of the memory 104 is located separately from the one or more processors 102. The memory 104 includes volatile or non-volatile memory, such as random access memory (RAM), dynamic RAM, or cache.

[0013] The storage device 108 includes a fixed or removable storage device, such as, but not limited to, a hard disk drive, a solid state drive, an optical disk, or a flash drive. The one or more auxiliary devices 106 include, but are not limited to, one or more auxiliary processors 114 and / or one or more input / output ("IO") devices. The auxiliary processor 114 includes, but is not limited to, a processing unit capable of executing instructions, such as a central processing unit, a graphics processing unit, a parallel processing unit capable of performing compute shader operations in a single instruction multiple data form, a multimedia accelerator such as a video encoding or decoding accelerator, or any other processor. Any auxiliary processor 114 may be implemented as a programmable processor that executes instructions, a fixed function processor that processes data according to fixed hardware circuits, a combination thereof, or any other type of processor.

[0014] The one or more IO devices 116 include one or more input devices, such as a keyboard, a keypad, a touch screen, a touchpad, a detector, a microphone, an accelerometer, a gyroscope, a biometric scanner, or a network connection (e.g., a wireless LAN card for sending and / or receiving wireless IEEE 802 signals), and / or one or more output devices, such as a display, a speaker, a printer, a tactile feedback device, one or more lights, an antenna, or a network connection (e.g., a wireless LAN card for sending and / or receiving wireless IEEE 802 signals).

[0015] The last level cache 110 acts as a shared cache for various components of the device 100, such as the processor 102 and various auxiliary devices 106. In some implementations, other caches are present in the device 100. For example, in some examples, the processor 102 includes a cache hierarchy that includes different levels, such as level 1 and level 2. In some examples, each such cache level is specific to a particular logical partition of the processor 102, such as a processor core, or a processor chip, die, or package. In some examples, the hierarchy also includes other types of caches. In various examples, one or more of the auxiliary devices 106 include one or more caches.

[0016] The last level cache 110 is "last level" in the sense that such cache is the last cache that the device 100 attempts to service a memory access request before servicing the request from the memory 104 itself. For example, if the processor 102 accesses data that is not stored in any of the cache levels of the processor 102, the processor exports a memory access request to be satisfied by the last level cache 110. The last level cache 110 determines whether the requested data is stored in the last level cache 110. If the data is in the last level cache 110, the last level cache 110 services the request by providing the requested data from the last level cache 110. If the data is not in the last level cache 110, the device 100 services the request from the memory 104. It can be seen that in some specific implementations, the last level cache 110 acts as a final cache level before the memory 104, which helps to reduce the overall amount of memory access latency for accesses to the memory 104.

[0017] Although the last level cache 110 can provide certain benefits, such as improving access latency for frequently accessed data or data that is closely together, there are situations where the last level cache 110 provides no or little benefit. In such situations, the power consumed by operating the last level cache 110 can be considered wasteful.

[0018] Thus, the present disclosure provides techniques for powering down the last level cache 110 in situations where the last level cache 110 provides little or no benefit (the last level cache 110 has "low cache effectiveness"). Generally speaking, according to these techniques, in response to a power down trigger, the last level cache 110 is powered down. Memory access transactions that are serviced from data within the last level cache 110 are instead serviced directly from the memory 104. For example, when the last level cache 110 is powered up, in the event that a memory access request fails in a cache that is one level lower than the last level cache 110 (such as within the processor 102 or another component), the last level cache 110 checks whether the appropriate data is in the last level cache 110. If the data is present, the last level cache 110 provides the data from within the last level cache 110 to the requestor. When the last level cache 110 is powered off, and another unit requests access to data that is not stored in a cache one level lower than the last level cache 110, the last level cache 110 obtains the data from the memory 104 and provides the data to the requester. When the last level cache 110 is powered off, the last level cache 110 does not check whether such data is stored in the last level cache 110.

[0019] Note that the statement that the last level cache 110 is "powered off" means that some or all of the databases are powered off. However, the cache controller of the last level cache 110 may still perform operations such as transferring data and memory access requests between the last level cache 110 and the memory 104. In other examples, the last level cache 110 being powered off means that all circuitry of the last level cache 110 is powered off. In such instances, the actions stated herein as being performed by the last level cache 110 are performed by other entities such as a portion of the memory 104 (e.g., a memory controller), or by a separate unit that is not within the last level cache 110.

[0020] As described above, the last level cache 110 is powered off in response to a power-off trigger. In some examples, the power-off trigger includes: a sufficient number of entities served by the last level cache 110 are powered off. In some examples, a sufficient number is at least a threshold number of entities. In some examples, the device 100 enters a low power mode as a power-off trigger. In some examples, when a sufficient number of devices including a processor 102 and one or more auxiliary devices 106 are powered off, a power-off trigger occurs. In some examples, a sufficient number of devices include all clients of the last level cache 110 (e.g., processor 102 and all auxiliary devices 106). Therefore, in such examples, when all clients of the last level cache 110 are powered off, a power-off trigger occurs. In some examples, "power off" means being in an inactive or idle state. In some examples, an operating system, an application, or other software executes one or more instructions that instruct the device to enter a low power mode. In some such examples, entering the low power mode is a power-off trigger. In some examples, the power-off trigger includes any action that causes device 100 or a system-on-chip of device 100 to enter a low-power state. In various implementations, the power-off trigger is any combination of low-power states for any of the elements of device 100.

[0021] In other examples, a power-down trigger occurs when the access pattern in the last level cache 110 indicates that the last level cache 110 provides a low return to the device 100. In some examples, the last level cache 110 or another entity (such as the processor 102, a memory controller coupled to the last level cache 110, or some other entity) tracks misses and / or hits within the last level cache 110. In some examples, a power-down trigger occurs when the miss rate in the last level cache 110 is too high or the hit rate in the last level cache 110 is too low. In some examples, the miss rate is the percentage of misses in the last level cache 110 compared to the total number of access requests to the last level cache 110. In some examples, the hit rate is the percentage of hits in the last level cache 110 compared to the total number of access requests to the last level cache 110. A miss is the result of an access request to the last level cache 110 where the requested data is not within the last level cache 110. A hit is the data requested by the access request is within the last level cache 110. In some examples, a miss rate that is too high occurs when the miss rate is above a threshold that is considered too high. In some examples, a hit rate that is too low occurs when the hit rate is below a threshold that is considered too low. An access request is a request from another entity (e.g., processor 102 or auxiliary device 106) to access memory. In some examples, an access request is a request to read from or write to a memory address. In some examples, the request occurs because a miss occurs in all caches at one level lower than the last level cache 110 in the cache hierarchy.

[0022] Figure 2 and Figure 3 The operation for powering down the last level cache 110 in response to a power down trigger is illustrated. Figure 2 In the embodiment of the present invention, the power-off trigger is that a sufficient number of devices (e.g., processor 102 and / or one or more auxiliary devices 106) are powered off. As described above, any or all of these devices may be powered off for any reason, such as in response to a request by software (such as an operating system, application, driver, or other software executing on processor 102 or other devices). Figure 3In the example, the power-off trigger is that the access pattern of the last level cache 110 indicates that there is little or no benefit in powering up the last level cache 110. There are various possible workload types that do not benefit from the last level cache 100 at least sometimes. Some examples include workloads that do not exhibit a large degree of spatial and / or temporal memory access locality. Spatial memory access locality means that memory accesses are made to memory addresses that are close together. Such access patterns benefit from the last level cache 110 because a miss causes a cache line to be placed in the last level cache 110. Therefore, if a subsequent access request is close to the first access request that caused the miss, it is likely that a subsequent access will be made to the same cache line that is already in the cache. Temporal memory access locality means that memory accesses are close together in time. If two accesses to the same cache line are close together in time, it is likely that the subsequent access will hit the cache. The longer it takes to re-access the same cache line, the more likely the cache line is to be removed from the cache due to, for example, eviction.

[0023] In some implementations, the last level cache 110 uses Figure 2 Power-down trigger (power-down of processor 102 and / or auxiliary device 106), but not used Figure 3 In other embodiments, the last level cache 110 uses Figure 3 The power-off trigger is not used Figure 2 In other embodiments, the last level cache 110 uses Figure 2 The power-off trigger and Figure 3 In some implementations, the last level cache 110 uses Figure 2 Power-off trigger but not used Figure 3 The power failure triggers the operation in the first mode, and subsequently switches to the last level cache 110 using Figure 3 Power-off trigger but not used Figure 2 In some implementations, the last level cache 110 uses Figure 2 The power-off trigger and Figure 3 The power failure triggers the operation of both in the third mode. In some specific implementations, the last level cache 110 operates in any of the first mode, the second mode, or the third mode. In some examples, the last level cache 110 switches between the first mode, the second mode, and the third mode.

[0024] Figure 4A and Figure 4B The operation of the device 100 is illustrated when the last level cache 110 is powered off (4A) and when the last level cache 110 is powered on (4B). Figure 4A In the case where the last level cache 110 is powered off, the last level cache 110 does not provide responses to memory access requests from data within the last level cache 110. The last level cache 110 is bypassed, and transactions served by the last level cache 110 are instead directly served by the memory 104. In some examples, the last level cache 110 is completely powered off, and no components of the last level cache 110 process memory access transactions. In other examples, a portion of the last level cache 110 is powered off, but a portion of the last level cache 110 remains powered on. In an example, one or more databases of the last level cache 110 are powered off and one or more control circuits (such as a cache controller) remain powered on. In some examples, the one or more control circuits treat all received memory access requests as misses and forward such requests to the memory 104. In such examples, the one or more control circuits do not access the databases of the last level cache 110 because those databases are powered off.

[0025] Figure 4B The device 100 is illustrated in a mode in which the last level cache 110 is powered. In this mode, the last level cache 110 services transactions received by other components. Servicing some of these transactions may require communication between the last level cache 110 and the memory 104. For example, in the event of a miss in the last level cache 110, the last level cache 110 obtains the cache line in which the miss occurred from the memory 104. In another example, in the event of a dirty cache line being evicted from the last level cache 110, the last level cache 110 sends the cache line to the memory 104 to be written.

[0026] As described elsewhere herein, in response to a power down trigger, the device 100 powers down the last level cache 110 . FIG. 5A to FIG. 5D A technique for powering down the last level cache 110 according to an example is illustrated. In these figures, the last level cache 110 includes a plurality of cache lines 508, each of which includes at least data 504 and a state 506. For cache lines whose state is dirty, the last level cache 110 flushes the cache line to the memory 104. The flush copies the contents of the cache line to the memory. A cache line is dirty if it is modified compared to a copy stored in the memory 104.

[0027] It takes a lot of time to refresh the entire last level cache 110, so the last level cache 110 is likely to continue to serve memory requests while the power is off. The power-off walker 502 walks the cache lines and refreshes the cache lines to the memory 104. For cache lines that have been refreshed, such cache lines have an invalid state and are therefore not available for servicing memory access requests. However, cache lines that have not yet been refreshed can be used to serve memory access requests. For read requests that hit a cache line that has not yet been reached by the power-off walker 502 (and is therefore not invalid), such requests are served by providing the data in the cache line as a response to the request. For write requests that hit a cache line that has not yet been reached by the power-off walker 502 (and is therefore not invalid), such requests are served by modifying the data in the cache line. If the cache line is not marked as dirty before servicing the write request, the cache line is marked as dirty. In some examples, when the power-off walker 502 reaches the cache line 508, the last level cache 110 flushes the cache line to the memory 104. In other examples, the last level cache 110 flushes the cache line immediately when a write is performed, regardless of the location of the power-off walker 502, which then causes the cache line to become invalid in the last level cache 110. In the event that a miss occurs for a memory access request when the last level cache 110 is powered off, the last level cache 110 passes the memory access request to the memory 104. The last level cache 110 does not allocate cache lines into the last level cache 110 during power off. In other words, if a memory access request misses in the last level cache 110 while the last level cache 110 is being powered off, the last level cache 110 does not fetch the associated cache line from the memory 104 into the last level cache 110. Instead, the last level cache 110 passes the request to the memory 104.

[0028] FIG. 5A to FIG. 5D An example sequence of operations for powering down the last level cache 110 is illustrated. Figure 5A In , power-down detector 502 is flushing cache line 508(1). The lines (508(2), 508(3), and 508(4)) that pass cache line 508(1) are valid. Any such line can service an incoming request. Figure 5B , power-down walker 502 has invalidated cache line 508(1) and is flushing cache line 508(2). Cache line 508(3) and cache line 508(4) can service incoming requests, but cache line 508(1) cannot service incoming requests. Figure 5C, power-down checker 502 has advanced to cache line 508(3). Power-down checker 502 flushes the contents of cache line 508(3) to memory 104 and invalidates cache line 508(3). Figure 5D In FIG. 5 , power-down walker 502 walks cache line 508 (4), but because the cache line is not dirty, power-down walker 502 does not flush the cache line. However, power-down walker 502 does invalidate the cache line.

[0029] When the last level cache 110 is operating in a power-down mode and the device 100 detects a power-on trigger, the device causes the last level cache 110 to be powered back on. In some examples, such as when the power-down trigger that causes the last level cache 110 to be powered off is a sufficient number of entities being powered off, the power-on trigger is that at least some of those entities are powered on. In some examples, the power-on trigger is that all of those entities are powered on. In some examples, when the power-down trigger is that the access pattern of the last level cache 110 indicates that the last level cache 110 is not providing sufficient benefits (such as when the miss rate of the last level cache 110 is too high or the hit rate of the last level cache 110 is too low), the power-on trigger detects that the device 100 is executing a workload characteristic that is substantially different from when the last level cache 110 was initially powered off. In other words, in some examples, in a situation where the device 100 powers down the last level cache 110 because the last level cache 110 does not provide sufficient benefit, the device 100 powers up the last level cache 110 in response to detecting that the workload characteristics have changed. The change in workload characteristics indicates that the last level cache 110 is likely to provide substantial benefit to the operation of the device 100 again. For example, if the last level cache 110 is powered down because it does not provide sufficient benefit to a first workload, when the workload characteristics change, the last level cache 110 is likely to provide sufficient benefit to the operation of the device resulting in the new workload characteristics. More simply, even if the last level cache 110 does not provide benefit to the first workload, the last level cache 110 is likely to provide benefit to the second workload. The change in workload characteristics prompts that a new workload can be executed on the device 100, and prompts that the last level cache 110 should be powered up again to provide benefit to the new workload. In the event that the last level cache 110 does not provide sufficient benefit for the new workload, the last level cache 110 may be powered down again.

[0030] There are various ways that the device 100 can detect a change in workload as a power-on trigger for the last level cache 110. In one example, a change in resource utilization is a power-on trigger. In various examples, the change in resource utilization is a change in the amount of memory used or a change in the load on the processor 102. In another example, the power-on trigger is a change in the application with user focus. When the user is actively interacting with the application, the application has user focus (for example, when the user is browsing a web page and providing mouse clicks and keyboard input to the web browser, the web browser has user focus). The change in the application with user focus indicates that the workload is likely to have changed. Another power-on trigger is that a specific auxiliary device 106 changes from low utilization to high utilization or from high utilization to low utilization. In some examples, changing from low utilization to high utilization means that the utilization of the auxiliary device 106 exceeds a threshold. In some examples, changing from high utilization to low utilization means that the utilization of the auxiliary device 106 is below a threshold. In some examples, the utilization of the auxiliary device means the amount of work performed on the auxiliary device compared to the total amount of work that can be performed on the auxiliary device. In some examples, the auxiliary device 106 is a video decoder that can handle the following: decoding four different video streams in parallel at very high resolution with high image settings, and an example of low utilization would be decoding one stream at low resolution with low image settings. In another example, low utilization means that the video decoder is not active at all. Changes in the utilization of the auxiliary device suggest that the device 100 is performing a different workload, and therefore the last level cache 110 may be useful for such new workload. Any other changes in the operating characteristics of the device 100 may be a power-up trigger.

[0031] Once the power-on trigger is detected, the device 100 powers up the last level cache 110. In some examples, an initialization sequence is required after the cache is powered up and before the cache can begin to be used. In some examples, powering up the last level cache 110 includes powering up a powered-off bank and beginning to operate normally as a cache (e.g., servicing requests from other entities in the device 100, including fetching cache lines from the memory 104 in response to misses and placing those cache lines into the last level cache 110).

[0032] Figure 6 is a method 600 for operating the last level cache 110 according to an example. Figures 1 to 5D Although the system is described as being configured to perform the steps of method 600 in any technically feasible order, those skilled in the art will recognize that any system configured to perform the steps of method 600 in any technically feasible order is within the scope of the present disclosure.

[0033] In step 602, the device 100 detects a power down trigger. The power down trigger is any of the power down triggers described herein, such as a sufficient number of components of the device 100 being powered down, or such as detecting that the access pattern for the last level cache 110 indicates that the operation of the last level cache 110 is not providing significant benefits.

[0034] In step 604, in response to the power-off trigger, the device 100 powers off the last level cache. In some examples, the power-off is described elsewhere herein (including with respect to FIG. 5A to FIG. 5D ) occurs as described.

[0035] In step 606, the device 100 detects a power-on trigger. In various examples, the power-on trigger is one of the power-on triggers described elsewhere herein. In step 608, the device 100 powers up the last level cache 110 in response to the power-on trigger. In various examples, powering up the last level cache 110 includes bringing components of the last level cache 110 into an operational state, thereby allowing the last level cache 110 to service memory access requests again.

[0036] Elements in the figures are embodied as software executing on a processor, a fixed function processor, a programmable processor, or a combination thereof, where appropriate. Processor 102, last level cache 110, interconnect 112, memory 104, storage 108, and various auxiliary devices 106 include at least some hardware circuitry and, in some implementations, software executing on the processor, within that component or within another component. Certain elements of last level cache 110 are FIG. 5A to FIG. 5D Cache line 508 represents the portion of the memory bank of last level cache 110 that is sized to store a cache line. Data 504 and state 506 are those portions of cache line 508 that are designed to store data or state.

[0037] It should be understood that many variations are possible based on the disclosure herein. Although the above features and elements are described in particular combinations, each feature or element may be used alone without the other features and elements, or in various combinations with or without the other features or elements.

[0038] The provided methods may be implemented in a general purpose computer, a processor, or a processor core. Suitable processors include, by way of example, a general purpose processor, a special purpose processor, a conventional processor, a digital signal processor (DSP), a plurality of microprocessors, one or more microprocessors associated with a DSP core, a controller, a microcontroller, an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) circuit, any other type of integrated circuit (IC), and / or a state machine. Such processors may be manufactured by configuring a manufacturing process using the results of processed hardware description language (HDL) instructions and other intermediate data including a netlist (such instructions capable of being stored on a computer readable medium). The results of such processing may be masks, which are then used in a semiconductor manufacturing process to manufacture processors that implement the features of the present disclosure.

[0039] The methods or flow charts provided herein may be implemented in a computer program, software, or firmware incorporated into a non-transitory computer-readable storage medium for execution by a general-purpose computer or processor. Examples of non-transitory computer-readable storage media include read-only memory (ROM), random access memory (RAM), registers, cache memory, semiconductor memory devices, magnetic media (such as internal hard disks and removable disks), magneto-optical media, and optical media (such as CD-ROM disks), and digital versatile disks (DVDs).

Claims

1. A method for operating a cache, the method comprising: In response to a power down trigger indicating that cache validity is considered low, the cache is powered down.

2. The method according to claim 1, further comprising: In response to a power-on trigger, the cache is powered on again.

3. The method according to claim 1, wherein the power-off trigger comprises: One or more components of a device including the cache are powered off.

4. The method according to claim 1, wherein the power-off trigger comprises: An access pattern to the cache indicates a low profitability of the cache.

5. The method according to claim 4, wherein the access pattern comprises: A miss rate in the cache is above a threshold.

6. The method of claim 4, wherein the access pattern comprises: A hit rate in the cache is below a threshold.

7. The method according to claim 2, wherein the power-on trigger comprises: One or more components of a device are powered on, wherein the device includes the cache.

8. The method according to claim 2, wherein the power-on trigger comprises: A device including such a cache has substantial changes to operating characteristics.

9. The method of claim 1 , wherein powering down the cache comprises: Lines of the cache that are indicated as dirty are flushed.

10. A system, comprising: Cache; and a processor configured to execute instructions to store data in the cache, Wherein the cache is configured to be powered down in response to a power down trigger indicating that cache validity is considered to be low.

11. The system of claim 10, wherein the cache is configured to power up in response to a power-up trigger.

12. The system of claim 10, wherein the power-off trigger comprises: One or more components of a device including the cache are powered off.

13. The system of claim 10, wherein the power-off trigger comprises: An access pattern to the cache indicates a low profitability of the cache.

14. The system of claim 13, wherein the access pattern comprises: A miss rate in the cache is above a threshold.

15. The system of claim 13, wherein the access pattern comprises: A hit rate in the cache is below a threshold.

16. The system of claim 11, wherein the power-on trigger comprises: One or more components of a device are powered on, wherein the device includes the cache.

17. The system of claim 11, wherein the power-on trigger comprises: A device including such a cache has substantial changes to operating characteristics.

18. The system of claim 10, wherein powering off the cache comprises: Lines of the cache that are indicated as dirty are flushed.

19. A system, comprising: processor; Multiple auxiliary devices; and A cache, wherein the processor and the plurality of auxiliary devices are clients of the cache, wherein the cache is configured to be powered down in response to a power down trigger indicating that cache validity is considered low.

20. The system of claim 19, wherein the cache is further configured to: Restarting power in response to a power-on trigger.