NZ742061B2 - Use of volatile memory as non-volatile memory - Google Patents

Abstract

computing device may comprise a processor, a volatile memory and a non-volatile storage device. An operating system or firmware of the device may cause one or more pages of the volatile memory to be treated, by applications executing on the computing device, as non-volatile memory pages. A maximum number of pages that may be treated as non-volatile may be determined based on an amount of energy available in a battery and an amount of energy needed to transfer a page of memory to the non-volatile storage device. number of pages that may be treated as non-volatile may be determined based on an amount of energy available in a battery and an amount of energy needed to transfer a page of memory to the non-volatile storage device.

Description

USE OF VOLATILE MEMORY AS NON-VOLATILE MEMORY Technical Field This disclosure relates generally to the operation of memory s in a computing device. In particular, the disclosure relates to systems, methods, and computer m products for using le memory to provide non-volatile storage to applications executing on the computing .

Background The main memory of a computing device is typically based on c random- access (“DRAM”) memory modules. DRAM has various properties suitable for use as main memory, such as low cost and high storage density. However, DRAM memory modules typically contain capacitors or other circuits that require a continuous, or nearly continuous, supply of power to prevent data loss. DRAM memory is therefore referred to as volatile, because data stored in DRAM memory is lost in the event that its power supply is interrupted.

Other types of memory, such as Negative-AND gate (“NAND”) memory, may be ed to as non-volatile memory e a NAND memory module’s contents are not lost if the module’s power supply is interrupted. However, the main memory of a computer is not typically ucted from NAND memory modules, for various reasons such as higher cost and lower storage density compared to DRAM memory modules.

Summary [0003A] According to the present invention, there is provided a computing device sing: a volatile memory logically partitioned into a plurality of pages; a non-volatile storage device, wherein content of a page of the plurality of pages is transferable to the non-volatile storage device by a memory transfer operation; an operating system of the ing device; one or more sors that cause the computing device to at least: receive information indicative of an amount of energy in a battery, the energy available for use by the computing device; determine an amount of energy needed to perform the memory transfer operation; determine, based at least in part on the amount of energy needed to perform the memory transfer operation, a maximum number of pages of the ity of pages whose content is transferable to the non-volatile storage device using the amount of energy available for use by the computing device; and configure the operating system to treat one or more pages of the plurality of pages of the volatile memory as non-volatile memory, wherein a number of the one or more pages is based on the determined maximum number of pages. [0003B] According to the present invention, there is also provided a method of using memory of a computing device, the method comprising: obtaining information tive of an amount of energy needed to er contents of a page of volatile memory to non-volatile memory; receiving information indicative of an amount of energy available for transferring ts of the page of volatile memory to the non-volatile memory; determining, based at least in part on the amount of energy needed, that the contents of the page of volatile memory are transferable to the non-volatile memory using the amount of energy available; and configuring an operating system of the computing device to treat the page of volatile memory as a page of non-volatile .

] According to the present invention, there is also provided a system comprising: means for obtaining information indicative of an amount of energy needed to transfer ts of a volatile memory to a non-volatile memory; means for receiving information indicative of an amount of energy ble for transferring the ts of the volatile memory to the non-volatile memory; means for identifying, based at least in part on the amount of energy needed, a portion of the volatile memory that is transferable to the non-volatile memory using the amount of energy available; and means for configuring an operating system of the system to treat the portion of the volatile memory as non-volatile memory.

A computing device may comprise a volatile memory and a non-volatile storage device. While the computing device is ing on utility power, the computing device may e information indicative of how much energy would be available to the computing device if utility power were to be interrupted. The computing device may also determine how much energy would be needed to transfer a page of the volatile memory to the non-volatile storage device and, using this information, determine how many pages of memory could be ved using the energy available in the battery. Based at least in part on this information, an operating system or firmware of the computing device may identify a number of pages of the volatile memory as non-volatile, such that applications executing on the computing device may store information on the pages of volatile memory as if the pages were non-volatile.

This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key es or essential features of the claimed subject , nor is it intended to be used to limit the scope of the claimed subject matter Brief Description of the Drawings ments of the present disclosure will be described more fully hereinafter with reference to the accompanying drawings, in which: is a block m that s an example computing system with volatile memory identified by the operating system as non-volatile memory. is a block diagram that depicts adjusting the number of pages identified as latile based on available battery power.

[0009] is a block diagram depicting an example of a computing device fying volatile memory as volatile or non-volatile memory. depicts a battery shared by multiple ing devices. is a flow diagram depicting an example process for operating a computing device with volatile memory identified as non-volatile .

[0012] is a flow diagram ing an example of a process for adjusting non- volatile memory identification based on application performance parameters. is a block diagram providing an example of preserving the contents of volatile memory. is a flow diagram depicting an example process for preserving the contents of volatile memory identified as non-volatile. is a flow diagram depicting an example of lling power delivery to a processor core during memory preservation. is a flow diagram depicting an example of operating a computing device with volatile memory modules identified as non-volatile. depicts an example general purpose computing nment in which in which the techniques described herein may be embodied.

Detailed Description A computing device may se a processor, a main memory comprising volatile memory modules, and a non-volatile storage device. The volatile memory modules may be memory whose contents are lost in the event of a power loss. The volatile memory modules may, in some cases, be memory modules whose nce, with respect to retention of the contents stored in the memory, is relatively low compared to nonvolatile memories.

Power to the computing device may be ed by some combination of utility power and battery power. Utility power may refer to s such as mains power delivered over a power grid. Utility power may also refer to other power sources that may generally be considered sustainable or typically available, such as locally generated solar, wind, or l power. Utility power may sometimes include y components, such as batteries used in solar or wind power systems to store energy when there is a surplus and provide energy when there is a deficit. More generally, utility power may refer to any source of power that is typically available for an operating period of a computing device.

[0020] y power may refer to a backup source of energy, such as a device containing battery cells, capacitors, or other energy storage mechanism. When utility power is available, the battery may charge and the amount of power ble to the computing device may increase. When utility power is not ble, the computing device may operate on battery power and the amount of power ble in the battery may decrease. Other factors, such as ature, the age of the battery, and consumption of battery power by other devices may also affect the amount of available power.

The operating system or firmware of the device may represent volatile memory to applications as if the volatile memory were latile. The amount of volatile memory that may be represented, or identified, as non-volatile may be determined based on two factors. The first factor may be the amount of energy available for use by the computing device in the event that utility power is suspended. The second factor may be an estimate of an amount power required to transfer a page of volatile memory to a nonvolatile storage device. Based on these factors, the operating system or firmware may determine an estimate of how many pages of memory could be preserved to the storage device should utility power be interrupted. These pages may then be identified, by the operating system or firmware, as being non-volatile.

Should utility power be interrupted, the computing device may enter a memory preservation phase in which the contents of volatile memory identified as non-volatile are preserved. During this phase, power delivery to components of the computing device may be restricted to those components needed for memory preservation. This may permit a greater amount of volatile memory to be identified as non-volatile. In addition, the power state during memory preservation may be such that tions vis-a-vie energy ption and availability are more reliable. is a block diagram that depicts an example computing system with volatile memory identified by the operating system as non-volatile memory. A computing device 100 may comprise memories bearing instructions of an operating system 102 and firmware 104, a processor 106, DRAM memory modules 114, and a non-volatile e device 120.

The ing device 100 may typically e on a utility power source 122.

At times, such as when the utility power source 116 is interrupted, the computing device 100 may e on a battery power source 124. During a blackout or other fault d to the utility power source 122, the computing device may switch or transfer its source of power from the utility power source 122 to the battery power source 124. In some instances, the battery power source 124 may be integrated into the computing device 100.

In other instances, the battery power source may be external to the computing device 100.

The processor 106 may comprise various sub-components, including a core 108 and an uncore 110. The power consumption of the processor 102 may be controlled such that power to the core 108 and the uncore 110 may be suspended or maintained independently. For e, power delivery to the core 108 may be suspended while power to the uncore 110 may be ined. Suspending power may comprise partially or totally upting the flow of energy to the affected component. Suspending power may also refer to g the component in a low-power state. Typically, a ent whose power has been suspended does not operate while power is suspended, but may resume operation once power has been restored. Maintaining power to a component may comprise delivering sufficient power to the component, such that the component may remain operative with respect to at least some of its functions.

The core 108 may se a processing unit of the processor 106. Typically, the processor 106 may comprise of number of cores, although for simplicity in representation depicts the processor 106 as having a single core 108. As a processing unit of the sor 106, the core 108 typically executes computer-readable instructions, such as those of the ing system 102 and firmware 104, and thereby causes the computing device to perform various operations.

The uncore 110 may include portions of the processor 106 that are related to those of the core 108 but not included in it. In some cases the processor 106 may include one uncore 110 and a plurality of cores such as the depicted core 108. lly, the uncore 110 may perform functions related to L3 cache maintenance and include a memory controller 112.

The memory controller 112 may control access to data stored in the DRAM memory modules 114. This may include performing direct-memory access (“DMA”) operations. A DMA ion may involve transferring contents of memory. For example, a DMA operation may involve transferring contents of DRAM memory s 114 to a non-volatile storage device 120. A DMA operation may be ted by the core 108 executing instructions of the operating system 102 or re 104. Once the DMA operation has been initiated, the core 108 may resume other operations, or be placed in a low-power state or no-power state, while the DMA operation completes.

[0029] Interrupt signals may be transmitted to the processor 106. An interrupt signal may include signals or other transmissions from components of the computing device 100, or an external device such as the battery power source 124, of an event. Examples of interrupt signals include signals generated by network components, user interface components, and so forth. Various interrupt signals may, in some cases, be generated in response to error ions or status changes that may arise during operation of the computing device 100. Interrupt signals may be ted in relation to DMA operations.

For example, an interrupt signal might be generated in se to the completion of a DMA operation. Certain interrupts, such as those pertaining to DMA operations, may be processed by the uncore 110.

[0030] The battery power source 124 may also supply interrupt signals, or other communications, to the computing device 100. The communications may be indicative of s to the state of the battery power source 124. The state information may, for example, include information indicating whether the battery power source 124 is currently being charged from utility power source 122 and how much battery power is available to the ing device 100. In some instances, the battery power source 124 may supply power to a number of devices, so the amount of power available to the computing device 100 may be less than the total amount of power stored in the battery.

[0031] The DRAM memory modules 114 may be sub-divided into units of memory sometimes ed to as pages. The pages of memory may be associated with certain characteristics, such as memory speed and volatility. For example, the DRAM memory modules 114 may be volatile RAM, such that if power to the DRAM memory modules 114 is suspended, the contents of the DRAM memory modules 114 will lost. The characteristics of the memory may be conveyed to application programs that execute on the computing device 100. In some instances, firmware 104 may determine the teristics of the DRAM memory modules 114 at boot time and convey this information to the ing system 102. The operating system 102 may then convey these characteristics to an application program.

[0032] A page of memory, as used herein, may refer to a n of memory within a memory module. In some instances, a page of memory, sometimes referred to as a region of memory or a n of , may be grouped or logically partitioned by a characteristic of the memory device. For example, a page, , or portion of memory might correspond to a memory whose contents may be readable or writable in a single operation. In another example, a page, region, or portion of memory might share a cache line. In other instances, the boundaries of pages, regions, or portions of memory may be lly partitioned by a memory controller, firmware, or operating system.

As noted, the DRAM memory modules 114 may be volatile RAM. However, the firmware 104 and/or operating system 102 may identify pages of the volatile DRAM memory modules 114 as being non-volatile memory. The identification may comprise conveying information about characteristics of the memory to a user of the memory. For example, the firmware 104 might report to the operating system 102 that certain pages of the DRAM memory modules 114 are non-volatile memory pages. This may, for example, involve ng system description tables, such as system ption tables defined by the Advanced Configuration and Power Interface (“ACPI”). The operating system 102 might report this information to an application that is running on the computing device 100. An application running on the operating system might determine that the operating system has identified a page of memory as non-volatile by invoking an operating system application programming interface (“API”), by inspecting an ACPI system description table, and so forth. Note that in some cases, identifying memory as non-volatile may e the firmware or operating system recording that the page of memory should be treated as non-volatile, without explicitly or implicitly notifying a user of the page of memory.

The number of pages identified as volatile or non-volatile may depend on a variety of factors, including an amount of power available in the battery power source 124 and an amount of power needed to preserve the ts of a page of volatile memory that has been identified as being non-volatile. Accordingly, DRAM memory module 114 may comprise an fied non-volatile memory 116 portion and an identified volatile memory 118 portion.

Applications running on the computing device 100 may adapt their processing by, for example, writing data to memory identified as non-volatile memory without necessarily performing additional steps to ensure that the data has been committed. ations may, in some cases, achieve higher performance when greater amounts of memory are identified as being in non-volatile commit mode. An application may, for example, bypass processing related to ensuring that a write has been committed, if the write was to a region of memory that has been identified as latile. is a block diagram that depicts ing the number of pages identified as non-volatile based on available battery power. depicts a battery 200 and DRAM memory modules 208. Three states of the battery 200 are shown, corresponding to three energy levels 202, 204, 206. The example of is intended to illustrate an embodiment of a computing device, such as the computing device 100 depicted by which adjusts the amount of memory identified as non-volatile based on energy available to the device. In the example of the computing device 100 may be operating on utility power while the amount of energy available in battery 200 ates over time.

Although depicts a decreasing amount of battery power, in some instances the amount of energy might increase over time, and principles similar to those depicted in may be applied. The amount of energy might fluctuate for various reasons. For example, in some cases battery 200 might be ted to le computing devices, some of which might draw power from the y 200 while the computing device 100 remains on battery power. In another example, temperature or other ing conditions of the battery might cause the amount of energy available. In another example, the maximum capacity of the battery might degrade over time.

At energy level 202, an amount of energy in battery 200 may be sufficient to perform memory transfers on some number of the memory pages 220. In for example, the amount of energy in the battery 200 at energy level 202 may be ient to transfer the contents of seven of the ten depicted memory pages 220. The operating system or firmware may ine the number of memory pages that may be transferred based on factors such as the amount of memory in each page, the amount of energy used to perform a DMA operation, the amount of energy used by devices whose power is maintained during the memory transfer operations, and so forth. When the y 200 has a greater amount of energy available, a greater number of pages may be identified as non-volatile memory 210, and fewer pages may be identified as volatile memory 212.

At a reduced energy level 204, the amount of energy available for transferring the contents of memory pages 220 may also be reduced. There may, for e, be sufficient memory for transferring four memory pages using the available battery power. The operating system or firmware may identify four memory pages as non-volatile memory 214 and six pages identified as volatile memory 216.

Similarly, at a further reduced energy level 206, the battery 200 may not be able to supply sufficient battery power to transfer any of the pages of DRAM memory modules 208 to a non-volatile storage device. The operating system or firmware might then identify all of the pages of the DRAM memory modules 208 as volatile memory 218.

When a page of memory has been identified as non-volatile memory, its contents may be preserved prior to being subsequently identified as volatile memory. For e, when an amount of energy available in battery 200 has been d from energy level 202 to energy level 204, three pages of identified non-volatile memory 210 may then transition to being identified as volatile memory 216. The operating system or firmware may transfer the contents of the three pages of previously identified non-volatile memory 210 in response determining to transition the pages to be identified as le-memory 216. The contents of the memory may be transferred while the ing device 100 is still using utility power, and accordingly the amount of energy available in the battery 200 is not affected by the transfers.

If utility power were to fail prior to ting the transfers, there might not be sufficient energy ble to transfer all of the memory pages previously identified as non-volatile memory 210. The risk of this occurrence may be mitigated in various ways, ing but not limited to more nt adjustments to the number of pages of memory that are identified as non-volatile, and incorporating greater tolerance to battery fluctuation in the calculations used to determine the number of pages of memory to identify as non- volatile. For example, the number of pages to identify as latile may be reduced in proportion to the amount of power-level fluctuation in the battery. is a block diagram depicting an example of a computing device identifying volatile memory as volatile or non-volatile memory. A computing device 300 may se firmware 308 that receives communications from a battery 310. The battery 310 may, for example, send data containing information about the current state of y power, the amount of energy available in the y 310, and so forth. In some instances, the battery 310 may e various metrics and history information, such as the dates, times, and durations of utility power uptions.

The firmware 308 may be a basic input/output system (“BIOS”). The firmware 308 may initialize various hardware devices and components of the computing device 300.

The firmware 308 may also be involved in various aspects of operations at runtime. In some instances, the firmware 308 may provide an abstraction layer for the re components of the computing device 300, through which the operating system 304 accesses the hardware of the computing device 300.

[0044] A configuration and power interface 306 may provide the operating system 304 with access to information and updates pertaining to the memory configuration of the computing device 300. In some instances, the uration and power ace may comprise the Advanced Configuration and Power Interface (“ACPI”). The firmware 308 may provide configuration tables, through ACPI, that describe the characteristics of various memory modules installed on the computing device. In some ces, these tables might describe the characteristics of the memory modules as they are – le memory being reported as volatile memory, and non-volatile memory being reported as non-volatile. In other instances, the tables might be used to indicate that some proportion of the volatile memory modules, including potentially all of the le memory modules, are non-volatile. The tables might, in some cases, provide an indication that the nonvolatile characteristic of the memory is simulated by the firmware or operating system. In other cases, the table might not include an indication that the non-volatile characteristic is not that of the memory module itself.

The operating system 304 may enable the execution of various programs, processes, and sub-processes. These may be referred to herein as applications, such as the application 302 depicted in The application 302 may utilize memory identified as non-volatile in various ways. In an example, the operating system 304 may provide the application 302 with access to memory that is identified as latile using an application programming interface (“API”). This might comprise ng a heap creation or memory allocation API and specifying a flag ting that the returned heap or memory segment should be a non-volatile memory page. In some instances the application might be able to specify whether or not the non-volatile teristics of the supplied memory may be provided by the operating system or firmware. In other instances, pages based on volatile memory modules may be supplied transparently, such that the application may not necessarily comprise instructions that are adapted to the simulated non-volatility of the provided memory.

Data pertaining to the status of the battery 310 may be distributed to s components of the computing device 300, including firmware 308. The firmware 308 may, for example, receive or otherwise obtain data from the battery 310. The firmware 308 may distribute the information via the configuration and power interface 306 to the operating system 304. The operating system 304 may then apply the information by adjusting the amount of memory identified as non-volatile. The amount of memory identified as non- volatile may be determined based on various factors such as performance-to-risk ratio, estimated ility and length of utility power interruptions, the health of the battery, the rate at which battery power fluctuates, and so on.

The operating system 304 may respond to ses in the amount of energy available for use by the computing device 300. The response may include increasing the amount of memory fied as non-volatile. For e, the operating system 304 may select additional pages of DRAM memory modules for identification as non-volatile memory pages. This may e providing the ation 302 with access to a page of a selected DRAM module, identifying the page as being non-volatile, and causing the contents of the memory page to be transferred to a non-volatile storage device in the event of a utility power failure, or system shutdown. The memory ts may also be erred to a non-volatile storage device to make room for subsequent write operations to non-volatile memory, or for other reasons.

The operating system 304 may respond to decreases in the amount of energy available for use by the computing device 300. Aspects of the se may include selecting pages of DRAM memory modules previously identified as non-volatile and causing those pages to instead be identified as le. As noted, the identification process may include updating system configuration tables to indicate whether a page of memory is volatile or non-volatile. When the amount of energy available in the battery 310 has been reduced, the table may be d such that pages previously ted as non-volatile are indicated as volatile. Another aspect of the response may include transferring the contents of these deselected pages to a non-volatile storage device. The deselected page may have contents not yet preserved on a non-volatile storage device, and as such the contents may be preserved when the page is deselected. This may be avoided when no erved data has been written to the deselected page. As such, in some cases, pages whose contents have already been preserved, or to which no data has yet been written, may be red as s for deselection.

[0049] The battery 310 may be shared by devices in addition to computing device 300. s a battery shared by multiple computing devices. A battery 400 may provide reserve operating power to a number of computing devices 408, 410, 412. The y 400 may also provide, or allow access to, data pertaining to the state of the battery 400.

A n of battery power may be reserved for use by each of the ing devices 408, 410, 412. As depicted by each of computing devices 408, 410, 412 may have a reserved energy portion 402, 404, 406. The ed energy portions 402, 404, 406 may be reserved for use, by the respective computing devices 408, 410, 412, in the event that utility power is interrupted. In some instances, the reserved energy portions 402, 404, 406 may be reserved particularly for transferring the contents of DRAM memory modules that have been identified as non-volatile.

The amount of energy in the reserved energy portions 402, 404, 406 may be based on factors such as a target amount of memory to be identified as non-volatile. For example, computing device 408 might be ured to aggressively identify DRAM memory modules as non-volatile. As such, the reserved energy portion 402 for computing device 408 might be made larger than that of the other reserved energy portions 404, 406.

Other factors that might be incorporated into the amount of energy reserved may be risk tolerance, ted probability of losing utility power, battery health, and so forth. A model of energy needed to transfer the contents of memory may be used in conjunction with a model of the supply of energy from the battery. Each of the computing devices 408, 410, 412 may use the models to determine how much memory may be identified as nonvolatile without ering with the power ements of the other computing devices 408, 410, 412.

[0052] The energy in battery 400 may be reserved by the operating systems or firmware of the computing devices 408, 410, and 412. For example, the ing system of computing device 408 may receive or otherwise obtain information about other users of the battery 400, such as the other depicted computing devices 410, 412. The information may e factors that may affect the amount of energy reserved for each portion. In the event that utility power is interrupted, the computing device 408 may act within its ed “power budget,” e.g. by using only the amount of reserved energy 402, to transfer the contents of memory identified as non-volatile. is a flow diagram depicting an example process for operating a ing device with volatile memory identified as non-volatile memory. Although is ed as a sequence of blocks, it will be appreciated that the depicted sequence should not be construed as limiting the scope of the present disclosure to embodiments that adhere to the depicted sequence. Moreover, it will be appreciated that, in some ments of the present disclosure, certain of the operations indicated by the depicted blocks may be altered, reordered, performed in parallel, or omitted.

[0054] During the operation of a computing device, such as the computing device 100 depicted in the operating system or firmware of the computing device may periodically determine how much memory may be identified as non-volatile, based at least in part on the amount of energy needed to transfer the contents of that memory to a nonvolatile e device. Block 500 s determining a t capacity for identifying volatile memory modules as non-volatile, where the capacity may be limited by the amount of energy available to preserve the contents of memory in the event that utility power is interrupted.

Block 502 depicts the computing device operating with a number of volatile memory pages identified as non-volatile during a y power phase. The utility power phase may refer to times when utility power is available to the computing device. In some instance, the utility power phase may include periods of time in which utility power is interrupted, but for a length of time that is below a threshold length of time. The threshold may be based on an estimated probability that power will be restored before the threshold length of time has elapsed. The utility power phase may then continue if there are comparatively short periods of interruption.

The operations of block 504 may also be performed during the utility power phase. During this time, as depicted by block 504, the computing device may monitor y capacity and power state. Monitoring capacity may involve receiving or otherwise obtaining information about the amount of power stored in the battery. The computing device may, moreover, monitor the amount of power that is both stored in the battery and reserved for use by the computing device in the event of a power e. Monitoring the power state may involve receiving or otherwise obtaining information indicating whether or not the computing device and/or battery is currently being supplied with utility power, or if some other condition is causing the amount of ble energy in the battery to be reduced.

As depicted by block 506, the operations of blocks 500 to 504 may be ed while the utility power phase continues. If utility power fails or is otherwise interrupted, the operations of blocks 508 and 510 may be performed.

As depicted by and explained herein, the utility power phase may be associated with transient interruptions in utility power. However, the duration of the outage may be such that the computing device may enter a phase in which its behavior is adapted to the use of battery power. Block 508 s entering a battery power phase in which the operation of the computing device is adapted to the usage of battery power.

With respect to fying volatile memory as non-volatile memory, the operation of the computing device may be adapted in s ways. For example, the computing device might cease to identify new pages of volatile memory as non-volatile, and might opportunistically deselect le memory pages as non-volatile when the contents of those pages is transferred to a non-volatile storage . The degree to which this occurs may be based on various factors, such as tuning parameters that allow a risk versus performance tradeoff to be specified.

The operations of block 510 may be delayed until the amount of energy ble in the y has been reduced to a point that, were the battery drain to ue, there might not be enough energy available to preserve the contents of volatile memory that had been identified as non-volatile. Block 510 depicts entering a memory preservation phase for volatile memory that had been identified as non-volatile. In this phase, the computing device may enter a state in which power consumption is at ily directed to preservation of the contents of memory identified as non-volatile.

The amount of memory identified as non-volatile may have an effect on performance of a computing device. Application performance may be increased, in some instances, by identifying a greater amount of memory as non-volatile and using various techniques described herein to ensure that data written to volatile memory is preserved. is a flow diagram ing an e of a process for adjusting non-volatile memory identification based on application performance ters. gh depicted as a sequence of blocks, it will be appreciated that the ed sequence should not be construed as limiting the scope of the present disclosure to embodiments that adhere to the depicted sequence. Moreover, it will be appreciated that, in some embodiments of the present disclosure, certain of the operations indicated by the depicted blocks may be altered, reordered, med in parallel, or omitted.

Block 600 s a computing device receiving an indication of available y power. The indication may include information sufficient to determine how much of the available power would be reliably available should the computing device enter a memory preservation mode, as depicted by block 510 of Block 602 depicts determining a power budget and a power budget variance. The power budget may refer to the allocation of battery power in the event that a memory preservation mode is entered. For example, the power budget might include allocations of the available battery power to operate a core, an uncore, one or more memory modules, and the non-volatile storage device in order to preserve the contents of memory identified as non-volatile. The power budget might also include tions for other s.

The power budget variance may refer to an estimated reliability of the power budget. This may include adjustments for factors such as the amount of available power and the amount of power that items in the power budget might actually consume during a memory preservation phase. For example, a more aggressive power budget might assume that some percentage of memory pages identified as non-volatile would not actually need to be preserved during a memory preservation phase, since they may have already been preserved in the course of normal operations, or they may have never been written to and thus n g to be preserved. However, if these assumptions turn out to be inaccurate, the memory budget may be exceeded.

Block 604 depicts using tuning parameters to refine the power budget. For example, a tuning parameter might be an operating system or firmware configuration element that indicates how aggressively the computing system should identify volatile memory as non-volatile. For example, in some ations it may be acceptable to risk data being lost in the event of system failure. The power budget might then be adjusted to permit greater amounts of volatile memory to be identified as non-volatile. For other applications, data loss may be viewed as unacceptable. For these applications, the operating system or firmware configuration element might indicate that the power budget should be computing based on pessimistic projections of power usage during a memory preservation phase.

At block 606, the computing device may assign commit modes to pages of volatile memory. In other words, the ing device may determine to identify certain pages of volatile memory as having a non-volatile commit mode, while other pages may remain in a volatile commit mode. Here, the commit mode may refer to whether or not a write may be viewed as committed, i.e. persistent, when it is written to memory.

The ing system may select pages of memory for a non-volatile commit mode based on the power budget. This may include selecting up to the maximum number of pages of memory permitted by the power budget to have a non-volatile commit mode. It may also include ing pages to maximize the number of pages that may be associated with a latile commit mode, while remaining consistent with the power . In some instances, the selected pages may be grouped by memory module, so that the total number of memory modules having non-volatile commit mode pages may be reduced and conformance with the power budget may be increased. is a block diagram providing an e of preserving the contents of volatile memory. A DRAM memory module 700, having volatile memory characteristics, may comprise a number of pages of memory. Because the depicted DRAM memory module 700 is volatile, all of its constituent memory pages are volatile, i.e. their contents will be lost if power to the DRAM memory module 700 is upted. r, as noted herein, certain pages may be identified as being non-volatile, and thereby treated as non- volatile by applications executing on the computing . As depicted by the DRAM memory module may contain identified non-volatile pages 704, 706. Of these, some memory pages 704 may contain unpreserved content, while other memory pages 706 might n content that has already been preserved, or equivalently the memory page 706 might not have ever been written to, and thereby has nothing to be preserved. The DRAM memory module 700 might also n pages of memory that 708 that are not currently identified as volatile.

In the example of the contents of the identified non-volatile page 706 may have been previously preserved by transferring the contents of the page 706 to the non-volatile storage device 702. The contents of the page 706 are indicated as being stored on the non-volatile storage device 702 by the preserved page record 712.

The contents of one or more pages of memory fied as non-volatile, but not yet preserved 704 may be preserved on the non-volatile storage device by a direct memory transfer operation 714. For example, a processor and memory controller may cause a DMA operation to er the contents of a page to the non-volatile storage device 702.

Regions of the latile storage device 702 may be held in e. This is depicted in by ts entitled reserved e 710. The reserved storage 710 may include regions of space sufficient to store the contents of pages of memory identified as non-volatile 704 and 706. The amount of memory identified as non-volatile may, in some instances, be based partly on the amount of storage space available on the nonvolatile storage device 702, since a lack of available storage space might t the contents of memory identified as non-volatile from being preserved.

The memory preservation phase may begin by entering a low-power state adapted to preserving the contents of volatile memory. The low-power state may permit is a flow diagram depicting an example process for preserving the contents of volatile memory identified as non-volatile. Although depicted as a sequence of blocks, it will be appreciated that the depicted sequence should not be construed as limiting the scope of the present disclosure to embodiments that adhere to the depicted sequence.

Moreover, it will be appreciated that, in some ments of the present disclosure, certain of the operations indicated by the depicted blocks may be d, reordered, performed in parallel, or omitted.

Block 800 s the computing device suspending virtual machines or other ations, such as databases, for which a period of time to become quiescent is desired.

For example, upon a determination that the computing device has entered a memory preservation phase, or is about to, various applications such as virtual es may be notified and given a controlled period of time in which they may save as much uncommitted data as possible. Data written to memory identified as non-volatile may, in some instances, be treated by the application as if it were in a committed state, since the operating system and/or firmware will ve the contents of the memory during the memory preservation phase.

At block 802, the computing device may begin to enter the low-power state by suspending power delivery to certain devices not needed during the remainder of the memory vation phase. These devices may include graphics cards, user interface busses, networking cards, and so forth.

As depicted by block 804, the computing device may also mask processor interrupts. For example, all processor interrupts may be masked except those related to certain errors and those needed for processing memory transfer operations, such as DMA ions.

Block 806 depicts that power delivery to unused processors, including all ated cores, uncores, and other processor components, may be ded. In this t, unused may refer to those processors not needed for performing the memory preservation. For e, in some instances a single sor, or a single core of the single processor, may be sufficient to complete memory preservation. Power delivery to the remaining processors of the computing system may therefore suspended during the memory preservation phase. Note that in some instances an interrupt may wake a processor and cause power ry to be resumed. The interrupt masking depicted by block 804 may prevent this occurrence and keep the unused sors in a low-power or no-power state.

Block 808 depicts that the computing device may also suspend power delivery to volatile memory modules that have no pages identified as non-volatile. The computing device may also suspend power to any volatile memory modules whose contents have already been preserved. In addition, the computing device may also suspend power to any memory modules that are inherently non-volatile, such as negative-AND gate (“NAND”) memory modules.

In various instances, the computing device may prioritize memory transfer operations involving volatile memory modules that may be completed earliest. For example, the computing device may prioritize transferring the contents of a first memory module over the contents of a second memory , if the contents of the first memory module may be ved more quickly than the contents of the second memory module.

This approach may allow power delivery to the first memory module to be suspended sooner than power delivery to the second memory module, resulting in an overall decrease in the amount of energy used in the memory preservation phase.

Block 810 depicts that the computing device may suspend power delivery to a core of a processor while maintaining power delivery to an uncore of the processor. For multicore processors, power delivery to all of the cores may be suspended. During the memory preservation phase, power delivery to one or more of the cores may be resumed periodically. When restored, the core may be used to initiate a memory transfer operation, after which power delivery may again be suspended. This is depicted by block 812.

Meanwhile, power delivery to an uncore of the processor is ined. The uncore, ning a memory controller, may oversee the memory transfer operation and cause power delivery to the core of the processor to be resumed when the transfer is ted.

As depicted by block 814, power delivery to the non-volatile storage device, as well as any interfaces or ications busses required to write to the storage device, may be maintained during the memory vation phase so that the memory transfer operations may be completed. is a flow m depicting an example of controlling power delivery to a processor core during memory preservation. Although depicted as a sequence of blocks, it will be appreciated that the depicted sequence should not be construed as limiting the scope of the present disclosure to embodiments that adhere to the depicted sequence.

Moreover, it will be iated that, in some embodiments of the t disclosure, n of the operations indicated by the ed blocks may be altered, red, performed in parallel, or omitted.

Block 900 depicts that the core of a processor may execute instructions to initiate a direct memory transfer operation from volatile memory module to a latile storage device. The direct memory transfer operation may copy pages of the volatile memory modules that had been identified as non-volatile to the non-volatile storage device.

Block 902 depicts suspending power to the core of the processor after it has initiated a direct memory transfer operation. The direct memory transfer operation may be ongoing while power is ded. Power delivery to an uncore of the processor may be maintained during this time.

As depicted by block 904, an interrupt may be generated to indicate that the memory transfer operation is complete. The interrupt may be generated by the uncore’s memory controller. Processing of the interrupt signal may include causing the core to reawaken by at least resuming power delivery to the core. This operation is depicted by block 900.

At block 908, the awoken core may execute instructions to ine r all volatile memory previously identified as non-volatile has been preserved. This may comprise executing instructions that examine records of volatile memory pages identified as non-volatile and information indicating whether contents of the corresponding pages has already been preserved, does not need preservation because it has not been written to, or still needs to be preserved.

If all volatile memory identified as non-volatile memory has been preserved, or does not require preservation, the system may wn as depicted by block 910.

Otherwise processing may resume at block 900, where the awoken processor may initiate an additional memory transfer operation. is a flow diagram depicting an example of operating a computing device with volatile memory modules identified as non-volatile. Although ed as a sequence of blocks, it will be appreciated that the ed sequence should not be construed as limiting the scope of the present disclosure to embodiments that adhere to the depicted sequence. Moreover, it will be appreciated that, in some embodiments of the t disclosure, certain of the operations indicated by the depicted blocks may be altered, red, performed in parallel, or omitted.

[0087] Block 1000 depicts obtaining information indicative of an estimated amount of energy that would be used to transfer the contents of a page of le memory to a nonvolatile storage device. The information may be obtained from observation, experimentation, metrics recorded during operation of the ing device, and so forth.

The information may, in some cases, be supplied by configuration ters. The information may include the amount of energy used to perform a DMA transfer operation to the non-volatile storage device. The information may also include energy used by a sor core to te the DMA transfer, and the amount of energy consumed by the non-volatile storage device. Other power ption factors may also be included.

Block 1002 depicts obtaining information indicative of an amount of energy available in a battery. The information may be obtained, for example, from messages sent from the battery to the computing , or from a polling mechanism initiated by the computing . These examples are illustrative, and should not be construed as limiting.

Block 1004 s determining that the contents of the page of memory may be transferred to the non-volatile storage device using the amount of energy ble in the battery.

The determination may be based on operational modes such as a performance mode or a data-safety mode. The ional modes may be specified by parameters or configuration data, for example, the operational mode may be indicative of a level of risk which is deemed permissible when calculating the maximum number of pages of volatile memory to identify as non-volatile.

The determination may also be based on the total number of pages currently identified as non-volatile, or the number of pages that were so identified and that currently have data requiring preservation. The operating system or firmware may determine an amount of energy available per page, and compare that value to the energy required to preserve a page of memory. If the amount is more than sufficient, a greater number of pages may be identified as non-volatile. If the amount is insufficient, the operating system or firmware may take various steps, while on utility power, to re-identify a sufficient number of pages as volatile. The computing system may similarly determine a maximum number of pages of volatile memory that may be identified as non-volatile. The maximum number may be ically recalculated.

The determination may be based on a calculation of a statistical ility that, in the event of a utility power interruption, the portion of the le memory would be transferable to the non-volatile memory using the amount of energy available.

Block 1006 depicts configuring the operating system of a computing device to treat the page of volatile memory as non-volatile memory, based on the determination. ng the page of le memory as non-volatile may include ing, to an application executing on the computing device, information identifying a commit mode that is supported by the . The commit mode may, for example, indicate that data written to the memory is immediately committed, i.e. is immediately made durable with respect to the application’s involvement (since the operating or firmware will assume responsibility for preserving the ts of memory identified as non-volatile).

[0094] Configuring the operating system may comprise updating configuration interface data with information indicating that the page is latile instead of volatile. For example, the firmware may update ACPI tables so that the memory is described to its users as being non-volatile.

Block 1008 depicts receiving ation indicative of entering a memory preservation phase. The information may, for example, be sent or otherwise obtained from a y when utility power has been interrupted. In response to utility power being interrupted, the ing device may switch to using battery power and continue normal operations. r, as time passes, available power in the battery may drop to a point in which the amount of available energy, if it were to further decrease, would not be sufficient to preserve the contents of all volatile memory pages that are currently identified as non-volatile and that have contents to be preserved. At that time, the computing device may enter the memory preservation phase, which may involve shutting down power consumption not needed for memory preservation, and initiating the memory transfers ed to preserve the contents of volatile memory identified as latile.

The ts of the page may be transferred in response to receiving information tive of a reduction in the amount of energy available in the battery. The reduction might, for example, occur because of a loss of utility power, a reduction in the health of the battery, additional devices being connected to the battery, and so forth.

Prior to entering the memory preservation phase, the operating system or firmware of the computing device may begin to decommission pages of volatile memory that had been identified as non-volatile. This may involve preserving the contents of the memory and re-identifying the page as volatile. For example, while on battery power but operating normally, the operating system or firmware may monitor s such as the time since utility power was interrupted, or the amount of battery power available, and begin to decommission the simulated non-volatile pages under the assumption that the memory preservation phase is becoming sing likely, but is not yet certain. A tuning parameter or mance parameter may be used to adjust how rapidly this should occur.

Where performance rather than safety is at a premium, the ters may indicate that the system should continue to treat volatile memory as non-volatile for as long as possible.

Block 1010 s preserving the contents of the page of memory by transferring the contents of the page to the non-volatile storage device. The contents may, for example, be transferred by a DMA operation as described herein. The computing device may maintain a ed space on the non-volatile storage device in which the contents of the page may be stored.

In an embodiment, a computing device may comprise: a volatile memory logically partitioned into a plurality of pages; a non-volatile storage device, wherein content of a page of the plurality of pages is transferable to the non-volatile storage device by a memory transfer operation; an operating system of the computing device; one or more processors that cause the computing device to at least:

[0104] receive information indicative of an amount of energy in a battery, the energy available for use by the computing device; determine an amount of energy needed to perform the memory er operation; determine, based at least in part on the amount of energy needed to perform the memory transfer operation, a number of pages of the plurality of pages whose content is transferable to the non-volatile e device using the amount of energy available for use by the computing device; and configure the operating system to treat one or more pages of the plurality of pages of the volatile memory as non-volatile memory, wherein a number of the one or more pages is based on the determined number of pages.

In an embodiment, the one or more processors r cause the computing device to at least: transfer contents of the one or more pages to a reserved portion of the non-volatile storage device.

[0110] In an embodiment, the contents of the one or more pages are transferred in response to receiving information indicative of a reduction in the amount of energy in the battery.

In an embodiment, configuring the operating system comprises updating uration interface data with information indicative of one or more pages of non- volatile memory ponding to the one or more pages of volatile .

In an ment, a method of using memory of a ing device comprises: obtaining information indicative of an amount of energy needed to transfer contents of a page of volatile memory to non-volatile memory;

[0114] receiving ation indicative of an amount of energy available for transferring contents of the page of volatile memory to the non-volatile memory; determining, based at least in part on the amount of energy needed, that the contents of the page of volatile memory are transferable to the non-volatile memory using the amount of energy available; and configuring an operating system of the computing device to treat the page of le memory as a page of non-volatile memory.

In an embodiment, the method further comprises: transferring the contents of the page of volatile memory to a reserved portion of the non-volatile memory.

In an embodiment, the method further comprises:

[0120] transferring the contents of the page of volatile memory to a ed portion of the non-volatile memory in response to receiving information indicative of a ion in the amount of energy available for transferring contents of the page of volatile memory to the non-volatile memory.

In an embodiment, configuring the operating system comprises ng configuration interface data with ation indicative of one or more pages of nonvolatile memory corresponding to the one or more pages of le memory.

In an embodiment, the method further comprises: determining that the contents of the page of volatile memory are transferable to the non-volatile memory based at least in part on information indicative of at least one of a performance mode or a safety mode of the computing device.

In an embodiment, the method further comprises: receiving information indicative of the computing device having ed from utility power to battery power; configuring the operating system to treat the page of volatile memory as volatile memory based at least in part on a length of time since the computing device switched from utility power to battery power; and transferring the contents of the page of volatile memory to the non-volatile memory.

In an embodiment, the information indicative of an amount of energy needed to transfer contents of a page of volatile memory to non-volatile memory is based at least in part on energy consumed by performing a direct memory access transfer from the page of volatile memory to the non-volatile memory.

In an embodiment, the information indicative of an amount of energy needed to transfer contents of a page of volatile memory to non-volatile memory is based at least in part on energy consumed by a processor initiating an operation to transfer the contents of the page.

[0130] In an embodiment, the method further comprises: determining that the contents of the page of volatile memory are transferable based at least in part on a number of other pages of volatile memory configured to be treated by the operating system as non-volatile memory pages.

In an embodiment, the method further comprises:

[0133] determining a maximum number of pages of volatile memory that are transferable to the non-volatile memory using the energy available for transferring contents of the page of volatile memory to the non-volatile memory.

In an embodiment, a computer-readable storage medium has stored thereon er-executable instructions that, upon execution by a er, cause the computer to at least: obtain information indicative of an amount of energy needed to transfer contents of a volatile memory to a non-volatile memory; receive information indicative of an amount of energy available for erring the contents of the volatile memory to the non-volatile memory;

[0137] identify, based at least in part on the amount of energy needed, a n of the le memory that is transferable to the non-volatile memory using the amount of energy ble; and configure an ing system of the computer to treat the portion of the volatile memory as non-volatile memory.

[0139] In an embodiment, the computer-readable e medium, further ses instructions that, upon execution by the computer, cause the computer to at least: identify the portion of the volatile memory that is transferable to the non-volatile memory by at least determining an amount of memory that could be transferred to the latile memory using the amount of energy available.

[0141] In an embodiment, the computer-readable storage medium further comprises instructions that, upon execution by the computer, cause the computer to at least: calculate a statistical probability that, in an event of a utility power interruption, the portion of the volatile memory would be transferable to the non-volatile memory using the amount of energy available.

In an embodiment, the computer-readable storage medium further comprises further instructions that, upon execution by the computer, cause the computer to at least: transfer the portion of volatile memory to a ed portion of the non-volatile memory in response to receiving information indicative of a reduction in the amount of energy available for transferring the contents of the volatile memory to the nonvolatile memory.

[0145] In an ment, the operating system is configured to treat the portion of the volatile memory as non-volatile memory by at least ng configuration and interface data accessed by the ing system.

In an embodiment, the portion of the volatile memory is treated by the operating system as non-volatile memory by at least providing, to an application executing on the er, information indicative of a commit mode compatible with data stored in the portion of the volatile memory.

Aspects of the present sure may be ented on one or more computing devices or environments. depicts an example ing environment in which in which some of the ques described herein may be embodied. The computing device 1102 is only one example of a suitable computing environment and is not intended to suggest any limitation as to the scope of use or onality of the presently disclosed subject matter. Neither should the depiction of the ing environment be interpreted as implying any dependency or requirement relating to any one or combination of components illustrated in the example computing device 1102. In some embodiments the various depicted computing ts may include circuitry configured to instantiate specific aspects of the present sure. For example, the term circuitry used in the disclosure can include specialized hardware components configured to perform on(s) by firmware or switches. In other es embodiments the term circuitry can include a l purpose processing unit, memory, etc., configured by software instructions that embody logic operable to perform function(s). In example embodiments where circuitry includes a combination of hardware and software, an implementer may write source code embodying logic and the source code can be compiled into machine readable code that can be processed by the general purpose processing unit. Since one skilled in the art can appreciate that the state of the art has evolved to a point where there is little difference between hardware, software, or a combination of hardware/software, the selection of hardware versus software to effectuate ic functions is a design choice left to an implementer. More specifically, one of skill in the art can appreciate that a software process can be transformed into an equivalent hardware structure, and a re structure can itself be transformed into an equivalent software process. Thus, the selection of a hardware implementation versus a software implementation is one of design choice and left to the implementer.

Computing device 1102, which may include any of a mobile device, smart phone, tablet, laptop, desktop computer, etc., typically includes a variety of computerreadable media. Computer-readable media can be any available media that can be accessed by computing device 1102 and includes both volatile and nonvolatile media, removable and non-removable media. As used herein, media and computer readable media do not include propagating or transitory signals per se.

[0149] The system memory 1122 includes er-readable storage media in the form of memory such as read only memory (“ROM”) 1123 and random access memory (“RAM”) 1160. The RAM memory 1160 may include volatile memory modules, such as dual in-line memory modules (“DIMMs”). The RAM 1160 n of system memory 1122 may sometimes be referred to as main memory. RAM 1160 typically contains data and/or program modules that are immediately accessible to and/or presently being operated on by processor 1159. By way of example, and not limitation, illustrates operating system 1025, application ms 1126, other program modules 1127, and program data 1128.

The processor 1159 typically ns at least one primary processing unit, sometimes referred to as a core, and at least one system agent, mes referred to as an uncore. The core of the processor 1159 typically executes computer-executable ctions while the uncore performs related tasks which may include overseeing memory ers and ining a processor cache. The uncore may comprise a memory controller for interfacing between cores of the processor 1159 and system memory 1122.

[0151] A basic input/output system 1124 (“BIOS”), containing the basic routines that help to transfer information between elements within ing device 1102, such as during up, is typically stored in ROM 1123. The BIOS 1124 may be replaced, in various embodiments, by other firmware.

The computing device 1102 may also include latile storage devices. By way of e only, illustrates a hard disk drive 1138 that reads from or writes to non-removable, non-volatile magnetic media, and an optical disk drive 1114 that reads from or writes to a removable, non-volatile optical disk 1153 such as a CD ROM or other optical media. Other non-volatile storage devices that can be used in the example operating environment include, but are not limited to, flash memory, digital versatile disks, solid state disk drives, and the like. The hard disk drive 1138 is typically connected to the system bus 1121 through an movable memory interface such as interface 1134, and optical disk drive 1104 is typically connected to the system bus 1121 by a removable memory interface, such as interface 1135.

The drives and their ated computer storage media discussed above and illustrated in , provide storage of computer-readable instructions, data structures, program modules and other data for the computing device 1102. In for example, hard disk drive 1138 is illustrated as g instructions of the operating system 1158, application programs 1157, other program modules 1156, and program data 1155. Note that these components can either be the same as or different from operating system 1125, application programs 1126, other program modules 1127, and program data 1128.

Operating system 1158, application programs 1157, other program modules 1156, and program data 1155 are given different numbers here to illustrate that, at a minimum, they are different copies. A user may enter commands and information into the computing device 1102 through a user input device 1152. The user interface device 1152 may include, but is not d to, keyboards, ads, computer mice, trackballs, and so forth. Other input devices, also not shown, may include a microphone, joystick, game pad, satellite dish, scanner, or the like. These and other input devices are often connected to the processing unit 1159 through a user input interface 1136 that is coupled to the system bus, but may be connected by other interface and bus structures, such as a el port, game port or a universal serial bus (USB). A screen 1142 or other type of y device is also connected via GPU 1129, although in some instances the screen 1142 may be driven h the system bus 1121 or another interface. In on to the monitor, computers may also include other eral input/output devices such as speakers, printers, and so forth which may be connected through an input/output interface 1133. A battery 1184 may also be connected to the system by the input/output interface 1133. The battery 1184 may send and receive information via the input/output interface 1133. The information may include state ation such as the amount of energy available in the battery 1134, the state of utility power 1182, the health of the battery 1134, and so forth.

A power supply 1180 may control delivery of power to the components of computing device 1102. Power delivery may, at times, be suspended to particular components while maintained to other components. Suspension of power may involve total or partial interruption in the flow of energy to an effective component, and may therefore include causing a component to enter a low-power state.

The power supply 1180 may receive power from utility power 1182 or a battery 1184. Utility power 1182 may refer to any power source that may be ered to be generally available during an operational period of the computing device 1102. The y 1184 may include any power source ed to provide backup power in the event that utility power 1182 is interrupted.

The computing device 1102 may e in a networked environment using logical connections to one or more remote computers, such as a remote er 1146.

The remote computer 1146 may be a personal computer, a server, a router, a network PC, a peer device or other e node, and typically includes many or all of the elements described above relative to the computing device 1102. The connections depicted in include a network 1145, which may include local-area, wide-area, cellular, and mesh networks, or other types of networks.

[0157] It will also be appreciated that various items are illustrated as being stored in memory or on storage while being used, and that these items or portions thereof may be transferred between memory and other storage devices for purposes of memory management and data integrity. Alternatively, in other embodiments some or all of the software modules and/or systems may execute in memory on another device and communicate with the illustrated computing systems via inter-computer communication.

Furthermore, in some embodiments, some or all of the systems and/or s may be implemented or ed in other ways, such as at least partially in firmware and/or hardware, including, but not d to, one or more application-specific ated circuits (ASICs), standard integrated circuits, controllers (e.g., by executing riate instructions, and including microcontrollers and/or ed controllers), fieldprogrammable gate arrays (FPGAs), complex programmable logic devices ), etc.

Some or all of the modules, systems and data structures may also be stored (e.g., as software instructions or structured data) on a computer-readable medium, such as a hard disk, a memory, a network or a portable media article to be read by an appropriate drive or via an riate connection. The systems, modules and data structures may also be transmitted as generated data signals (e.g., as part of a carrier wave or other analog or digital propagated signal) on a variety of computer-readable transmission media, including wireless-based and wired/cable-based media, and may take a variety of forms (e.g., as part of a single or multiplexed analog signal, or as multiple discrete digital packets or frames).

Such computer program products may also take other forms in other embodiments.

Accordingly, the present disclosure may be practiced with other computer system urations.

[0158] Each of the processes, methods and algorithms described herein may be embodied in, and fully or partially automated by, modules comprising computer executable instructions loaded into memory and executed by one or more processors of a computing device. The processes and algorithms may also be implemented wholly or partially in application-specific circuitry. The results of the sed processes and process steps may be stored, persistently or otherwise, in any type of computer storage device such as, e.g., volatile or non-volatile e. Volatile and latile storage, as used herein, excludes propagating or transitory s per se.

The s features and processes described herein may be used independently of one another, or may be ed in various ways. All possible combinations and subcombinations are ed to fall within the scope of this disclosure. In addition, certain elements of the processes, s, and algorithms may be omitted in some implementations. The methods and processes described herein are also not limited to any particular sequence, and the depictions sing blocks or states relating thereto can be performed in other sequences that are riate. For example, described blocks or states may be performed in an order other than that specifically disclosed, or multiple blocks or states may be combined in a single block or state. The example blocks or states may be performed in serial, in parallel or in some other manner. Blocks or states may be added to or removed from the disclosed example embodiments. The example systems and components described herein may be configured differently than described. For example, elements may be added to, removed from or rearranged compared to the disclosed example embodiments.

Conditional language used herein, such as, among others, “can,” “could,” “might,” “may,” “e.g.” and the like, unless specifically stated otherwise, or otherwise understood within the context as used, is generally intended to convey that certain embodiments include, while other embodiments do not e, certain features, elements, and/or steps. Thus, such conditional language is not generally intended to imply that features, ts and/or steps are in any way required for one or more embodiments or that one or more embodiments necessarily e logic for deciding, with or without author input or prompting, whether these es, elements and/or steps are included or are to be performed in any particular embodiment. The terms “comprising,” “including,” “having” and the like are mous and are used inclusively, in an open-ended fashion, and do not exclude additional elements, features, acts, operations and so forth. Also, the term “or” is used in its inclusive sense (and not in its exclusive sense) so that when used, for example, to connect a list of elements, the term “or” means one, some or all of the elements in the list.

The embodiments presented herein are so presented by way of example, and are not intended to limit the scope of the present disclosure. Thus, nothing in the foregoing description is intended to imply that any particular feature, characteristic, step, module or block is required, necessary, or indispensable. The methods and systems described herein may be ed in a variety of forms. Various omissions, substitutions and changes in the form of the methods and s described herein may be made without departing from the spirit of what is disclosed herein. The accompanying claims and their equivalents are intended to cover such forms or modifications as would fall within the scope and spirit of certain embodiments disclosed herein.

Although the subject matter has been described in ge specific to structural features and/or acts, it is to be understood that the subject matter d in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as examples of enting the claims and other equivalent features and acts are intended to be within the scope of the claims.

Claims Defining the Invention 1. A ing device comprising: a volatile memory lly partitioned into a plurality of pages; a non-volatile storage device, wherein content of a page of the plurality of pages is transferable to the non-volatile storage device by a memory transfer operation; an operating system of the computing device; one or more processors that cause the ing device to at least: receive information indicative of an amount of energy in a battery, the energy available for use by the computing device; ine an amount of energy needed to perform the memory transfer operation; determine, based at least in part on the amount of energy needed to perform the memory er operation, a maximum number of pages of the plurality of pages whose content is transferable to the non-volatile storage device using the amount of energy available for use by the computing device; and configure the operating system to treat one or more pages of the plurality of pages of the le memory as non-volatile memory, wherein a number of the one or more pages is based on the determined maximum number of pages. 2. The computing device of claim 1, wherein the one or more processors further cause the computing device to at least: transfer contents of the one or more pages to a reserved portion of the non-volatile storage device. 3. The computing device of claim 2, wherein the contents of the one or more pages are transferred in se to receiving information tive of a reduction in the amount of energy in the battery. 4. The computing device of claim 1, wherein configuring the operating system comprises ng configuration interface data with information indicative of one or more pages of non-volatile memory corresponding to the one or more pages of volatile memory.

. A method of using memory of a computing device, the method comprising: obtaining information indicative of an amount of energy needed to er contents of a page of volatile memory to non-volatile memory; receiving information indicative of an amount of energy available for transferring contents of the page of volatile memory to the non-volatile memory; determining, based at least in part on the amount of energy needed, that the contents of the page of volatile memory are transferable to the non-volatile memory using the amount of energy available; and configuring an operating system of the computing device to treat the page of volatile memory as a page of non-volatile memory. 6. The method of claim 5, further comprising: transferring the contents of the page of le memory to a reserved portion of the non-volatile memory. 7. The method of claim 5, further comprising: transferring the contents of the page of volatile memory to a reserved portion of the non-volatile memory in response to receiving information tive of a reduction in the amount of energy available for transferring contents of the page of volatile memory to the non-volatile memory. 8. The method of claim 5, wherein configuring the operating system comprises updating uration interface data with ation indicative of one or more pages of non-volatile memory corresponding to the one or more pages of volatile . 9. The method of claim 5, further comprising: determining that the ts of the page of volatile memory are erable to the non-volatile memory based at least in part on information indicative of at least one of a performance mode or a safety mode of the computing device.

. The method of claim 5, further comprising: receiving information tive of the computing device having switched from utility power to battery power; configuring the operating system to treat the page of volatile memory as volatile memory based at least in part on a length of time since the computing device switched from utility power to battery power; and transferring the contents of the page of le memory to the non-volatile memory. 11. The method of claim 5, wherein the information indicative of an amount of energy needed to transfer ts of a page of volatile memory to non-volatile memory is based at least in part on energy consumed by ming a direct memory access transfer from the page of volatile memory to the non-volatile memory. 12. The method of claim 5, wherein the information indicative of an amount of energy needed to transfer contents of a page of volatile memory to non-volatile memory is based at least in part on energy consumed by a processor ting an operation to transfer the contents of the page. 13. The method of claim 5, further comprising: determining that the contents of the page of volatile memory are transferable based at least in part on a number of other pages of volatile memory configured to be treated by the operating system as latile memory pages. 14. The method of claim 5, further comprising: determining a maximum number of pages of volatile memory that are transferable to the non-volatile memory using the energy available for erring contents of the page of volatile memory to the non-volatile memory.

. A system comprising: means for obtaining information indicative of an amount of energy needed to transfer contents of a volatile memory to a non-volatile memory; means for receiving information indicative of an amount of energy available for transferring the ts of the volatile memory to the non-volatile memory; means for identifying, based at least in part on the amount of energy needed, a portion of the volatile memory that is transferable to the non-volatile memory using the amount of energy available; and means for configuring an ing system of the system to treat the n of the le memory as non-volatile memory. 16. The system of claim 15, comprising: means for identifying the portion of the volatile memory that is transferable to the latile memory by at least determining an amount of memory that could be transferred to the non-volatile memory using the amount of energy available. 17. The system of claim 15, comprising: means for calculating a statistical probability that, in an event of a y power interruption, the portion of the volatile memory would be transferable to the non-volatile memory using the amount of energy available. 18. The system of claim 15, comprising further: means for transferring the portion of volatile memory to a reserved portion of the non-volatile memory in response to receiving information indicative of a reduction in the amount of energy available for transferring the contents of the volatile memory to the non- le memory. 19. The system of claim 15, wherein the operating system is configured to treat the portion of the volatile memory as non-volatile memory by at least updating configuration and interface data accessed by the operating system.

. The system of claim 15, wherein the portion of the volatile memory is treated by the operating system as non-volatile memory by at least ing, to an application executing on the system, information indicative of a commit mode compatible with data stored in the n of the volatile memory. 1/11 m_:w_o>.coc o @0906 ON_\ 992E: m_:m_o> cmczcmE \CoEmE meUoE boEmE mo_>m_o \CoEmE v: fl oczsano 00? m_:m_o>.coc hag Wmm“w E<mo 52,8 charm , @838 899$ 8:259 h @969 9:9on 8:03.80 52/8 Lommmooa @838 mm? Booc: N: o: boEmE 3:2: 2/11 identified non- identified volatile volatile memory memory 210 212 (—Afiflj IIIIIIIDDB L—V—J DRAM memory modules 208 memory pages identified non-volatile ll identified volatile memory memory 214 216 *—w his? gazzuuuuluV V y DRAM memory modules 208 memory pages identified le memory 218 battery DDDDDDDDDD DRAM memory modules 208 memory pages Figure 2 3/11 I computing device I I I I I I I I I I I I application I I 302 I I I I I I I I I I I I I I I : operating system : I I I I I I I I I I I I I uration and power interface I I 306 I I I I I I I I firmware I ' 308 | I I I I battery Figure 3 WO 27709 4/11 83% >995 mESQEoo N:V 8339 mow 83% >995 v mESQEoo o:V 8539 vow charm 83% 35cm mESQEoo wow 8539 Nov /11 determine current capacity for identifying volatile memory as latile operate at current latile DRAM capacity during utility power phase monitor battery capacity and power state no end of utility power phase? enter battery power phase enter memory preservation phase for volatile memory identified as non-volatile Figure 5 6/11 receive tion of available battery power determine power budget and power budget variance use tuning parameters to refine power budget assign commit modes to pages of volatile memory Figure 6 ll 7/% WO 27709 8/11 suspend virtual machines and other applications suspend power ry to graphics cards, interfaces, and other unused devices mask processor interrupts suspend power delivery to unused processors suspend power ry to volatile memory modules without pages identified as non-volatile suspend power delivery to the cores of a processor while maintaining power delivery to an uncore of the processor periodically resume power deliver to one or more cores of the processor maintain power delivery to storage device and storage device interface Figure 8 9/11 use core to te direct memory access transfer from volatile memory identified as non-volatile to a non-volatile storage device suspend power to core of the processor while maintaining ry of power to uncore of the processor process interrupt indicating that transfer is complete wake core of processor all volatile memory identified as non-volatile preserved? shutdown complete Figure 9 /11 obtaining information indicative of an estimated amount of energy that would be used to transfer the contents of a page of le memory to a non-volatile e device 1000 obtaining information indicative of an amount of energy available in a battery 1002 determining that the contents of the page of memory may be transferred to the non-volatile storage device using the amount of energy available in the battery 1004 configuring the operating system of a ing device to treat the page of - volatile memory as non-volatile memory, based on the determination 1006 receiving information indicative of ng a memory preservation phase 1008 preserving the contents of the page of memory by transferring the contents of the page to the non-volatile storage device 1010 Figure 10 11/11 52/8 9:; r >5th vwrr 3%: H5ng vamo_>mv _ _ _ mESQEoo ow: ow: mothE 99:9 52/8 mm: {oz/EC NV: E mothE mmw Cmmgow _\ ................... __. _ ......... Fmrrwsm,tmu>m 5%: 89:9: .5223:83% mm: mo: 5w: mm: mm: 93E No: boEmE 310 m3w>oE9 m_:m_o>-coc mm: llllllllll mo_>mn_ om_o_> \CoEmE 83.95 mESQEoo Emgmoa 3:330 Lowwmooa mm: m_nm>oE9-coc m_:m_o>-coc vm: \CoEmE 89:95 5&0 38505 8: II 6©Fr§<mv mm: m0 8:8:qu mm: 36505 wESmoa mm: wcozmozaam E: mm: 550 Emgmoa

Claims (20)

Claims Defining the Invention

1. A ing device comprising: a volatile memory lly partitioned into a plurality of pages; a non-volatile storage device, wherein content of a page of the plurality of pages is transferable to the non-volatile storage device by a memory transfer operation; 5 an operating system of the computing device; one or more processors that cause the ing device to at least: receive information indicative of an amount of energy in a battery, the energy available for use by the computing device; ine an amount of energy needed to perform the memory transfer operation; determine, based at least in part on the amount of energy needed to perform the memory er operation, a maximum number of pages of the plurality of pages whose content is transferable to the non-volatile storage device using the amount of energy available for use by the computing device; and configure the operating system to treat one or more pages of the plurality of pages of the le memory as non-volatile memory, wherein a number of the one or more pages is based on the determined maximum number of pages.

2. The computing device of claim 1, wherein the one or more processors further cause the computing device to at least: transfer contents of the one or more pages to a reserved portion of the non-volatile storage device.

3. The computing device of claim 2, wherein the contents of the one or more pages are transferred in se to receiving information tive of a reduction in the amount of energy in the battery.

4. The computing device of claim 1, wherein configuring the operating system comprises ng configuration interface data with information indicative of one or more pages of non-volatile memory corresponding to the one or more pages of volatile memory.

5. A method of using memory of a computing device, the method comprising: obtaining information indicative of an amount of energy needed to er contents of a page of volatile memory to non-volatile memory; receiving information indicative of an amount of energy available for transferring contents of the page of volatile memory to the non-volatile memory; determining, based at least in part on the amount of energy needed, that the contents of the page of volatile memory are transferable to the non-volatile memory using the amount of energy available; and configuring an operating system of the computing device to treat the page of volatile memory as a page of non-volatile memory. 5

6. The method of claim 5, further comprising: transferring the contents of the page of le memory to a reserved portion of the non-volatile memory.

7. The method of claim 5, further comprising: transferring the contents of the page of volatile memory to a reserved portion of the non-volatile memory in response to receiving information tive of a reduction in the amount of energy available for transferring contents of the page of volatile memory to the non-volatile memory.

8. The method of claim 5, wherein configuring the operating system comprises updating uration interface data with ation indicative of one or more pages of non-volatile memory corresponding to the one or more pages of volatile .

9. The method of claim 5, further comprising: determining that the ts of the page of volatile memory are erable to the non-volatile memory based at least in part on information indicative of at least one of a performance mode or a safety mode of the computing device.

10. The method of claim 5, further comprising: receiving information tive of the computing device having switched from utility power to battery power; configuring the operating system to treat the page of volatile memory as volatile memory based at least in part on a length of time since the computing device switched from utility power to battery power; and transferring the contents of the page of le memory to the non-volatile memory.

11. The method of claim 5, wherein the information indicative of an amount of energy needed to transfer ts of a page of volatile memory to non-volatile memory is based at least in part on energy consumed by ming a direct memory access transfer from the page of volatile memory to the non-volatile memory.

12. The method of claim 5, wherein the information indicative of an amount of energy needed to transfer contents of a page of volatile memory to non-volatile memory is based at least in part on energy consumed by a processor ting an operation to transfer the contents of the page.

13. The method of claim 5, further comprising: determining that the contents of the page of volatile memory are transferable based at least in part on a number of other pages of volatile memory configured to be treated by the operating system as latile memory pages.

14. The method of claim 5, further comprising: determining a maximum number of pages of volatile memory that are transferable to the non-volatile memory using the energy available for erring contents of the page of volatile memory to the non-volatile memory.

15. A system comprising: means for obtaining information indicative of an amount of energy needed to transfer contents of a volatile memory to a non-volatile memory; means for receiving information indicative of an amount of energy available for transferring the ts of the volatile memory to the non-volatile memory; means for identifying, based at least in part on the amount of energy needed, a portion of the volatile memory that is transferable to the non-volatile memory using the amount of energy available; and means for configuring an ing system of the system to treat the n of the le memory as non-volatile memory.

16. The system of claim 15, comprising: means for identifying the portion of the volatile memory that is transferable to the latile memory by at least determining an amount of memory that could be transferred to the non-volatile memory using the amount of energy available.

17. The system of claim 15, comprising: means for calculating a statistical probability that, in an event of a y power interruption, the portion of the volatile memory would be transferable to the non-volatile memory using the amount of energy available.

18. The system of claim 15, comprising further: means for transferring the portion of volatile memory to a reserved portion of the non-volatile memory in response to receiving information indicative of a reduction in the amount of energy available for transferring the contents of the volatile memory to the non- 15 le memory.

19. The system of claim 15, wherein the operating system is configured to treat the portion of the volatile memory as non-volatile memory by at least updating configuration and interface data accessed by the operating system.

20. The system of claim 15, wherein the portion of the volatile memory is treated by the operating system as non-volatile memory by at least ing, to an application executing on the system, information indicative of a commit mode compatible with data stored in the n of the volatile memory.