Selective refresh of memory devices
Through selective refresh technology, the active data is placed into a specific memory segment and the segment mask is set, which solves the problem of performance degradation and increased delay of memory devices during frequent reads or writes, and achieves more efficient memory management and data reserve.
Patent Information
- Application Number
- CN202380034538.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-05-03
- Filing Date
- 2023-04-05
- Publication Date
- 2024-11-26
AI Technical Summary
Existing memory devices cause performance degradation and increased latency when memory segments are frequently read or written, because units are not available to process services during the refresh process, and conventional refresh commands are redundant and may interrupt the use of algorithms.
Using selective refresh technology, the active data is placed into specific memory segments through the memory controller and a segment mask is set to prevent frequent read or written memory segments from being actively refreshed, and these are indirectly refreshed using read or write commands. segment, reduce redundant refresh operations.
Improves the performance and availability of memory devices, reduces redundant refresh commands, ensures data integrity in frequently used data segments, and provides a security network mechanism in case of interruptions to ensure data refresh.
Smart Images

Figure CN119032349A_ABST
Abstract
Description
[0001] Priority application
[0002] This patent application claims priority to U.S. patent application serial number 17 / 661,810, filed on May 3, 2022, entitled “SELECTIVE REFRESH FOR MEMORY DEVICES,” which is incorporated herein by reference in its entirety. background I. Technical field
[0003] The technology of the present disclosure generally relates to memory systems requiring memory cell refresh. II. Background Technology
[0004] Computing devices, more specifically mobile communication devices, are ubiquitous in modern society and have become increasingly common. The popularity of these mobile communication devices is due in part to the many features now enabled on such devices. The increased processing power in such devices means that mobile communication devices have evolved from pure communication tools to complex mobile entertainment centers, enabling an enhanced user experience. Access to such functionality often depends on interoperating memory systems with control systems to store instructions and data. One popular format for memory is the low power double data rate (LPDDR) synchronous dynamic random access memory (SDRAM) standard. JEDEC is the standard setting body for LPDDR and has released various versions of the standard, including the LPDDR5 update in June 2021. The existence of such standards provides opportunities for improvement and innovation, and such innovations can be used in existing standards or anticipated standards or other specific implementations. Summary of the invention
[0005] Aspects disclosed in the specific embodiments include selective refresh techniques for memory devices. Specifically, a memory device used with an application having frequent repeated read or write commands to certain memory segments may be able to set a flag or similar indication that exempts these certain memory segments from being actively refreshed. By exempting these memory segments from being actively refreshed, these memory segments can be used continuously, thereby improving performance. Similarly, because these memory segments are frequently the subject of read or write commands, these memory segments are indirectly refreshed by executing the read or write commands.
[0006] In this regard, in one aspect, an integrated circuit (IC) is disclosed. The IC includes a memory bus interface configured to be coupled to a memory device via a memory bus. The IC also includes a memory controller coupled to the memory bus interface. The memory controller is configured to instruct the memory device to place active data into a memory segment. The memory controller is also configured to set a segment mask for the memory segment.
[0007] In another aspect, an IC is disclosed. The IC includes a memory bus interface configured to be coupled to a memory controller via a memory bus. The IC also includes a memory bank coupled to the memory bus interface. The memory bank is configured to receive a command sequence to place active data into a memory segment. The memory bank is also configured to set a segment mask for the memory segment.
[0008] In another aspect, a method for managing memory is disclosed. The method includes instructing a memory device to place active data into a memory segment. The method also includes setting a segment mask for the memory segment. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Figure 1 is a block diagram of an exemplary mobile computing device that may include a memory element that operates in accordance with the JEDEC memory standard;
[0010] Figure 2A is a block diagram of a memory device configured to comply with the Low Power Double Data Rate (LPDDR) version 5 standard (LPDDR5);
[0011] Figure 2B is a block diagram of an LPDDR5 channel configuration showing how memory banks are arranged within a channel;
[0012] Figure 3 is a block diagram of a system on a chip (SoC) operating with a memory device through a memory bus according to an exemplary aspect of the present disclosure;
[0013] Figure 4 is a flow chart illustrating an exemplary process for processing a selective refresh within a memory device; and
[0014] Figure 5 is a flow chart illustrating an alternative exemplary process for handling a selective refresh within a memory device. DETAILED DESCRIPTION
[0015] Referring now to the accompanying drawings, several exemplary aspects of the present disclosure are described. The word "exemplary" is used herein to mean "serving as an example, instance, or illustration." Any aspect described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other aspects.
[0016] Aspects disclosed in the specific embodiments include selective refresh techniques for memory devices. Specifically, a memory device used with an application having frequent repeated read or write commands to certain memory segments may be able to set a flag or similar indication that exempts these certain memory segments from being actively refreshed. By exempting these memory segments from being actively refreshed, these memory segments can be used continuously, thereby improving performance. Similarly, because these memory segments are frequently the subject of read or write commands, these memory segments are indirectly refreshed by executing the read or write commands.
[0017] In yet another exemplary aspect, the memory device may be a low power double data rate (LPDDR) memory device that relies on dynamic random access memory (DRAM) that stores data by holding an electric charge within a cell formed by a plurality of transistors. The charge decays over time and must be refreshed or the data indicated by the charge will be lost or corrupted.
[0018] Before describing certain aspects of the present disclosure, Figure 1 An overview of computing devices is provided, showing where memory devices are used in such computing devices. Figures 2A to 3 Additional details about memory devices and memory controllers are provided. Figure 4 A discussion of exemplary aspects of the disclosure begins.
[0019] As a starting point for nomenclature, it should be understood that double data rate (DDR) is generally a technical term within the JEDEC specification and memory world. As used herein, DDR is defined as a signaling technique that uses both the falling edge and the rising edge of the clock signal. The use of these two edges is independent of the frequency, and a change in frequency (e.g., doubling) does not fall within DDR unless both edges are used. Furthermore, DDR is contrasted with single data rate (SDR), which can transmit data on either the rising edge or the falling edge, but not on both at the same time.
[0020] Figure 1 is a system-level block diagram of an exemplary mobile terminal 100, such as a smart phone, a mobile computing device tablet, etc. Although mobile terminals having an LPDDR bus are particularly contemplated as being able to benefit from the exemplary aspects of the present disclosure, it should be understood that the present disclosure is not limited thereto and may be used in any system having a comparable memory bus.
[0021] Continue to refer Figure 1, the mobile terminal 100 includes an application processor 104 (sometimes referred to as a host or system-on-chip SoC) that communicates with a mass storage element 106 via a universal flash memory (UFS) bus 108. In addition, the application processor 104 can communicate with a memory device 106A via an LPDDR bus 108A. Although the exemplary aspects of the present disclosure are specifically contemplated to be applicable to LPDDR version 4 or 5 (i.e., LPDDR4 or LPDDR5) and other emerging memory standards, the present disclosure is not limited thereto and may be applicable to other memory buses. It should be understood that the application processor 104 includes a bus interface configured to interoperate with a memory bus of the present disclosure (such as LPDDR bus 108A). In addition, there may be a memory controller circuit (not shown) within the application processor 104 that implements various aspects of the present disclosure. Figure 1 Likewise, the memory device 106A may have a bus interface and some form of control circuitry that processes commands received from the LPDDR bus 108A and accesses memory cells within the memory device 106A.
[0022] The application processor 104 may further be coupled to the display 110 via a display serial interface (DSI) bus 112, and to the camera 114 via a camera serial interface (CSI) bus 116. Various audio elements, such as a microphone 118, a speaker 120, and an audio codec 122, may be coupled to the application processor 104 via a serial low-power chip-to-chip multimedia bus (SLIMbus) 124. Additionally, the audio elements may communicate with each other via a SOUNDWIRE bus 126. A modem 128 may also be coupled to the SLIMbus 124 and / or the SOUNDWIRE bus 126. The modem 128 may further be coupled to the application processor 104 via a peripheral component interconnect (PCI) or PCI Express (PCIe) bus 130 and / or a system power management interface (SPMI) bus 132.
[0023] Continue to refer Figure 1 , the SPMI bus 132 may also be coupled to a local area network (LAN or WLAN) integrated circuit (IC) (LAN IC or WLAN IC) 134, a power management integrated circuit (PMIC) 136, a companion IC (sometimes referred to as a bridge chip) 138, and a radio frequency IC (RFIC) 140. It should be understood that separate PCI buses 142 and 144 may also couple the application processor 104 to the companion IC 138 and the WLAN IC 134. The application processor 104 may further be connected to a sensor 146 via a sensor bus 148. The modem 128 and the RFIC 140 may communicate using a bus 150.
[0024] Continue to refer Figure 1, RFIC 140 may be coupled to one or more RFFE elements, such as antenna tuner 152, switch 154, and power amplifier 156, via a radio frequency front end (RFFE) bus 158. Additionally, RFIC 140 may be coupled to envelope tracking power supply (ETPS) 160 via bus 162, and ETPS 160 may communicate with power amplifier 156. Together, these RFFE elements (including RFIC 140) may be considered an RFFE system 164. It should be understood that RFFE bus 158 may be formed of clock lines and data lines (not illustrated).
[0025] The LPDDR5 standard envisions Figure 2A The memory die or memory device 200 illustrated in FIG. 1 includes a memory block 202 having four memory bank groups (BGs), each memory bank group having four memory banks, for a total of sixteen memory banks, the memory block forming a Figure 2B Memory device 200 may include an interface 208 ( Figure 2A ), the interface having: a first group 210 of first 8 DQ conductors (DQ[7:0]), a first data mask inversion conductor (DMI0), two conductors forming a first differential write clock (WCK), and a pair of conductors forming a first read data strobe (RDQS0); and a second group 212 of second 8 DQ conductors (DQ[15:8]), a second data mask inversion conductor (DMI1), two conductors forming a second differential write clock (WCK), and two conductors forming a second read data strobe (RDQS1). The groups 210, 212 share a differential clock (CK), command and address conductors (CA[6:0]), a chip select conductor (CS), and a reset conductor (all shown in the middle group 214). The memory device 200 also includes one or more registers, at least one of which is a mode register 216. A typical LPDDR5 memory device 200 may have: a maximum bandwidth of 12.8 gigabytes per second (GB / s), an input / output speed of 6400 megabits per second (Mbps), a maximum CK frequency of 800 megahertz (MHz), a maximum WCK frequency of 3200 MHz, a CA speed of 1600 megatransfers per second (MT / s), and the memory device operates in accordance with a non-return-to-zero (NRZ) signaling scheme. Although banks and bank groups are specifically contemplated, they are not a core part of the present disclosure and may be omitted in a memory device using the present disclosure.
[0026] In practice, the memory block 202 is formed of various units (e.g., one transistor-one capacitor (1T-1C or sometimes 1T1C)). Each unit is expected to have a retention time of sixty-four milliseconds (64ms), and therefore each unit must be refreshed within a 64ms window. This refresh task is divided into refresh operations of equal size, each of which takes about three hundred nanoseconds (300ns). When a unit is being refreshed, it is not available for processing business (e.g., read or write commands). If a read or write command to a given unit occurs while the given unit is being refreshed, the business to the unit may be stalled until the refresh is completed, resulting in increased latency. Such latency is generally undesirable.
[0027] Figure 3 Some additional details of the memory system 300 separated from the mobile terminal 100 are provided. Specifically, the application processor 104 may include a memory controller 302. The memory controller 302 may interoperate with one or more processors 304 (one is shown). The processor 304 may be, for example, a central processing unit (CPU), a video encoder, a graphics processing unit (GPU), a neural signal processor (NSP), a neural processing unit (NPU), a digital signal processor (DSP), etc. The processor 304 may include a timer (not explicitly shown). According to exemplary aspects of the present disclosure, the memory controller 302 and / or the processor 304 may work with a cache or local memory 306 to store certain information. In addition, the application processor 104 may include a memory bus interface 308 configured to be coupled to the memory bus 108A.
[0028] The memory device 106A / 200 is also configured to be coupled to the memory bus 108A via a memory bus interface, such as the memory bus interface 208. The memory device 106A / 200 may be a DRAM module and include a memory block 202. The memory block 202 may have memory segments 310A, 310B that store active data. As used herein, active data is data that is frequently read or written and may be used multiple times within a refresh interval. Additional details about this active data are discussed below. Additional memory segments 312 may be used for normal data.
[0029] Exemplary aspects of the present disclosure take advantage of certain types of memory usage scenarios where active data exists. Because the data is active and may be used multiple times during each refresh interval, it is not necessary to refresh the cells containing the data to maintain the data. That is, each time a cell is read, a voltage is applied that refreshes the charge on the cell being read. Similarly, each time a cell is written, a voltage is applied that places a charge on the cell. Therefore, exemplary aspects of the present disclosure place this active data in specific memory segments in a timely manner, and do not issue refresh commands to those memory segments, but rely on frequent use of the active data to refresh the data in these memory segments.
[0030] Optionally, a safety net mechanism can be used to allow data that is normally active but temporarily underutilized to be refreshed via normal refresh techniques. Figure 4 and Figure 5 More details are provided. However, before going further, an example use case is first explored so that the example characteristics of activity data and examples of outages that might occur can be examined.
[0031] One exemplary use case is in inference workloads, such as machine learning. In machine learning, there are typically two types of active data. The first type of active data is sometimes described as "weights" that are repeatedly read within a single refresh interval. These weights are used in the algorithm to weight the variables used in the algorithm of the machine learning process. The second type of active data is sometimes called "activations," which are repeatedly overwritten within a single refresh interval. That is, the above algorithm may generate multiple intermediate values that are stored, used in another calculation, and overwritten multiple times. In both cases, the read command and the write command effectively continuously refresh the data. Using conventional refresh commands for such data is redundant and may interrupt the algorithm's use of the data because the unit is unavailable during the refresh period. It should be understood that there are other use cases where active data may be present, including streaming applications, such as a camera for display or a graphics processing unit (GPU) for display.
[0032] Machine learning can be based on a client-server model, such as over the Internet or through the cloud to utilize computers with greater processing power than is locally available to the entity initiating the machine learning algorithm. Due to the distributed nature of this client-server model, there may be situations where the client loses the communication link to the server, or there are delays in the network that cause a steady stream of packets to be received intermittently. In such instances, there may be gaps in the use of activity data because no commands are received to the server processor. Exemplary aspects of the present disclosure provide a safety net mechanism that can refresh activity data in such situations even if the initial aspects of the present disclosure have suspended the refresh operation of activity data.
[0033] Figure 4 A flow chart of a process 400 is shown that eliminates redundant refresh commands for active data and provides an optional safety net mechanism that preserves the active data even if there are certain interruptions in the use of the active data. In this regard, the process 400 begins when the processor determines that certain data is active data while running an inference workload (such as machine learning) or other similar workloads with active data. The processor 104 uses the memory controller 302 to place the active data in a specified memory segment (e.g., memory segments 310A, 310B) within the memory device 200 (box 402). In addition, the memory controller 302 may fill the mode register 216 with a segment mask (box 404). In the LPDDR standard, this segment mask may take the form of a partial array automatic refresh (PAAR) flag or bit that indicates to the DRAM module that the segment displaying the PAAR bit will not be refreshed. In effect, the operations of boxes 402 and 404 place the active data in specific memory segments and then mark these memory segments in a manner such that normal refresh commands will ignore these memory segments. It should be understood that this type of segment mask was originally intended for memory segments that were not used for actual data. That is, there is no reason to refresh a memory segment that has no data in it, so a segment mask is used to reduce the number of refresh commands required, thereby reducing power consumption. Exemplary aspects of the present disclosure reverse this concept and explicitly apply segment masks to memory segments that have actual data.
[0034] Continue to refer Figure 4 , process 400 continues, where the memory device 200 and the application processor 104 operate normally using standard refresh commands for the unmasked memory segments (box 406). The processor 304 may set a timer in response to the event (box 408). In an exemplary aspect, the event is the start of an inference workload of a machine learning algorithm. The processor 304 determines whether the event occurs again (box 410). If the determination result is yes, the process 400 returns to normal operation until the event occurs again. That is, the reoccurrence of the event resets the timer. However, if the determination result of box 410 is no, the event will not occur again, and the processor 304 determines whether the timer expires (box 412). Note that the timer can be a countdown timer that expires at zero or an up-counting timer that expires when a threshold is exceeded. If the determination result of box 412 is no, the timer counts (box 414), and the processor 304 checks for the reoccurrence of the event (box 410). However, if the answer to the decision at block 412 is yes, the processor 304 removes the segment mask from the register and begins refreshing the previously masked segments normally (block 416).
[0035] The use of a timer in process 400 acts as a safety net mechanism to ensure that masked segments are refreshed in the event of an interruption in a normally frequent event. For example, if the refresh interval is 64ms, the timer can be 32ms, but other values can also be selected. The value (refresh interval-timer) should be greater than the time required to refresh the segment mask. So, for example, if it takes 10 microseconds (μs) to refresh each segment and there are ten masked segments, the time required should be 100μs, and the value of the timer can be selected to satisfy (64ms-timer)>100μs. These values are exemplary only.
[0036] Instead of using a timer as described in process 400, exemplary aspects of the present disclosure may place the burden of tracking refreshes on the application that is using the memory device 200. Figure 5 500 is illustrated in FIG. 501 . Specifically, the operating system (OS) of the processor 304 communicates with the application to identify the masking capability of the memory device and the ability of the application to control the refresh (block 502). The processor 304 places the data in the specified memory segment (block 504) and fills the mode register with the segment mask (block 506).
[0037] The application may query the OS for the refresh interval (which may vary depending on temperature or for other reasons) (block 508). The application then reads and writes all cells of the specified memory segment within each refresh interval (block 510), but there is a risk that data may be lost if an interruption occurs.
[0038] The selective refresh techniques for memory devices according to the various aspects disclosed herein can be provided in or integrated into any processor-based device. Examples, not limited thereto, include: a set-top box, an entertainment unit, a navigation device, a communication device, a fixed location data unit, a mobile location data unit, a global positioning system (GPS) device, a mobile phone, a cellular phone, a smart phone, a session initiation protocol (SIP) phone, a tablet device, a tablet phone, a server, a computer, a portable computer, a mobile computing device, a wearable computing device (e.g., a smart watch, a health or fitness tracker, glasses, etc.), a desktop computer, a personal digital assistant (PDA), a monitor, a computer monitor, a television, a tuner, a radio, a satellite radio, a music player, a digital music player, a portable music player, a digital video player, a video player, a digital video disc (DVD) player, a portable digital video player, an automobile, a vehicle assembly, an avionics system, an unmanned aerial vehicle, and a multirotor aircraft.
[0039] It will be further understood by those skilled in the art that various exemplary logic blocks, modules, circuits and algorithms described in conjunction with the various aspects disclosed herein may be implemented as electronic hardware, instructions stored in a memory or in another computer-readable medium, wherein any such instructions are executed by a processor or other processing device or a combination of the two. As an example, the master device and slave device described herein may be adopted in any circuit, hardware component, IC, or IC chip. The memory disclosed herein may be a memory of any type and size, and may be configured to store any type of information desired. In order to clearly illustrate this interchangeability, the functionality of various exemplary components, blocks, modules, circuits and steps has been generally described above. How such functionality is implemented depends on specific applications, design choices, and / or design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such specific implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
[0040] The various illustrative logical blocks, modules, and circuits described in conjunction with the various aspects disclosed herein may be implemented or executed with a processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic components, discrete hardware components, or any combination thereof designed to perform the functions described herein. The processor may be a microprocessor, but in an alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices (e.g., a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors cooperating with a DSP core, or any other such configuration).
[0041] The various aspects disclosed herein may be embodied in hardware and instructions stored in hardware, and may reside in, for example, random access memory (RAM), flash memory, read-only memory (ROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), registers, hard disks, removable disks, CD-ROMs, or any other form of computer-readable medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from and write information to the storage medium. In an alternative, the storage medium may be integral to the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a remote station. In an alternative, the processor and storage medium may reside in a remote station, a base station, or a server as discrete components.
[0042] It is also noted that the operational steps described in any exemplary aspect of this article are described in order to provide examples and discussions. The described operations can be performed in many different orders other than the illustrated order. In addition, the operations described in a single operational step can actually be performed in multiple different steps. In addition, one or more operational steps discussed in the exemplary aspects can be combined. It will be understood that, as will be apparent to those skilled in the art, many different modifications can be made to the operational steps illustrated in the flow chart. Those skilled in the art will also understand that any of a variety of different technologies and techniques can be used to represent information and signals. For example, data, instructions, commands, information, signals, bits, symbols, and code chips that may be mentioned throughout the above description can be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or optical particles, or any combination thereof.
[0043] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations. Therefore, the disclosure is not intended to be limited to the examples and designs described herein, but to conform to the widest scope consistent with the principles and novel features disclosed herein.
[0044] Specific implementation examples are described in the following numbered clauses:
[0045] 1. An integrated circuit (IC), comprising:
[0046] a memory bus interface configured to couple to a memory device via a memory bus; and
[0047] a memory controller coupled to the memory bus interface and configured to:
[0048] Instructing the memory device to place active data into a memory segment; and setting a segment mask for the memory segment.
[0049] 2. The IC of clause 1, further comprising a processor coupled to the memory controller.
[0050] 3. The IC of clause 2, wherein the processor comprises a neural signal processor (NSP).
[0051] 4. The IC of clause 2, wherein the processor comprises a graphics processing unit (GPU).
[0052] 5. The IC of clause 2, wherein the processor comprises a video encoder.
[0053] 6. An IC as recited in any one of clauses 2 to 5, wherein the memory controller instructs in response to a command from the processor.
[0054] 7. An IC as described in any of clauses 2 to 6, wherein the processor comprises a neural processing unit (NPU).
[0055] 8. An IC according to any preceding clause, wherein the memory bus interface comprises a low power double data rate (LPDDR) bus interface.
[0056] 9. An IC according to any preceding clause, wherein the memory controller is further configured to instruct the memory device to refresh unmasked segments.
[0057] 10. An IC according to any preceding clause, wherein the segment mask includes at least one partial array auto-refresh (PAAR) bit.
[0058] 11. The IC of clause 10, wherein the segment mask comprises a plurality of PAAR bits.
[0059] 12. The IC of any one of clauses 2 to 11, further comprising a timer associated with the processor.
[0060] 13. The IC of clause 12, wherein the processor is configured to start the timer in response to an event.
[0061] 14. The IC of clause 13, wherein the processor is configured to restart the timer when the event occurs again.
[0062] 15. The IC of any of clause 13 or clause 14, wherein the processor is configured to instruct the memory controller to refresh masked segments when the timer expires and the event does not reoccur.
[0063] 16. The IC of any one of clauses 2 to 15, wherein the processor is further configured to communicate with an application to provide a refresh interval to the application.
[0064] 17. The IC of clause 16, wherein the refresh interval comprises a temperature compensated refresh interval.
[0065] 18. An IC as recited in any one of clause 16 or clause 17, wherein the application reads or writes the memory segment in each refresh interval.
[0066] 19. An IC according to any preceding clause, wherein the IC comprises a system on a chip (SoC).
[0067] 20. An IC according to any preceding clause, wherein the activity data includes weights.
[0068] 21. An IC as described in any of clauses 1 to 19, wherein the activity data includes activations.
[0069] 22. An integrated circuit (IC), comprising:
[0070] a memory bus interface configured to be coupled to a memory controller via a memory bus; and
[0071] a memory bank coupled to the memory bus interface and configured to:
[0072] receiving a command sequence to place active data into a memory segment; and
[0073] Sets a segment mask for the memory segment.
[0074] 23. A method for managing a memory, the method comprising:
[0075] instructing the memory device to place active data into the memory segment; and
[0076] Sets a segment mask for the memory segment.
[0077] 24. A method for managing a memory, the method comprising:
[0078] At a memory bank, receiving a command sequence to place active data into a memory segment; and
[0079] Sets a segment mask for the memory segment.
Claims
1. An integrated circuit (IC), comprising: a memory bus interface configured to couple to a memory device via a memory bus; and a memory controller coupled to the memory bus interface and configured to: instructing the memory device to place active data into the memory segment; as well as Sets a segment mask for the memory segment.
2. The IC of claim 1 further comprising a processor coupled to the memory controller.
3. The IC of claim 2, wherein the processor comprises a neural signal processor (NSP).
4. The IC of claim 2, wherein the processor comprises a graphics processing unit (GPU).
5. The IC of claim 2, wherein the processor comprises a video encoder.
6. The IC of claim 2, wherein the memory controller instructs in response to a command from the processor.
7. The IC of claim 2, wherein the processor comprises a neural processing unit (NPU).
8. The IC of claim 1, wherein the memory bus interface comprises a low power double data rate (LPDDR) bus interface.
9. The IC of claim 1, wherein the memory controller is further configured to instruct the memory device to refresh unmasked segments.
10. The IC of claim 1, wherein the segment mask includes at least one partial array auto-refresh (PAAR) bit. The IC of claim 10 , wherein the segment mask comprises a plurality of PAAR bits.
12. The IC of claim 2, further comprising a timer associated with the processor.
13. The IC of claim 12, wherein the processor is configured to start the timer in response to an event.
14. The IC of claim 13, wherein the processor is configured to restart the timer when the event occurs again.
15. The IC of claim 13, wherein the processor is configured to instruct the memory controller to refresh masked segments when the timer expires and the event does not reoccur.
16. The IC of claim 2, wherein the processor is further configured to communicate with an application to provide a refresh interval to the application.
17. The IC of claim 16, wherein the refresh interval comprises a temperature compensated refresh interval.
18. The IC of claim 16, wherein the application reads or writes to the memory segment in each refresh interval.
19. The IC of claim 1, wherein the IC comprises a system on a chip (SoC).
20. The IC of claim 1, wherein the activity data comprises a weight.
21. The IC of claim 1, wherein the activity data comprises activation.
22. An integrated circuit (IC), comprising: a memory bus interface configured to be coupled to a memory controller via a memory bus; and a memory bank coupled to the memory bus interface and configured to: receiving a command sequence to place active data into a memory segment; as well as Sets a segment mask for the memory segment.
23. A method for managing a memory, the method comprising: instructing the memory device to place active data into the memory segment; as well as Sets a segment mask for the memory segment.
24. A method for managing a memory, the method comprising: At a memory bank, receiving a command sequence to place active data into a memory segment; as well as Sets a segment mask for the memory segment.
