Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0866—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches for peripheral storage systems, e.g. disk cache
  • G06F12/0868—Data transfer between cache memory and other subsystems, e.g. storage devices or host systems
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0888—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches using selective caching, e.g. bypass
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F13/00—Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
  • G06F13/14—Handling requests for interconnection or transfer
  • G06F13/16—Handling requests for interconnection or transfer for access to memory bus
  • G06F13/1668—Details of memory controller
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
  • G06F2212/22—Employing cache memory using specific memory technology
  • G06F2212/222—Non-volatile memory
• • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
  • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
  • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
  • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

• the invention relates to cache utilization. More particularly, some embodiments of the invention relate to an apparatus and method for utilizing a nonvolatile cache in an electronic system such as a processor-based system.
• U.S. Patent No. 7,360,015 describes some utilizations of a cache memory which may include determining whether requested information is part of a streaming access, and directly transferring the requested information between a storage device and a memory if the requested information is part of the streaming access. Alternately, if the requested information is not part of the streaming access, it may be transferred between the storage device and a cache.
• the cache may be a non-volatile disk cache.
• a white paper published at ftp://download.intel.com/design/flash/NAND/turbomemory/whitepaper.pdf, a white paper describes Intel® Turbo Memory as consisting of an Intel Turbo Memory controller ASIC (Application Specific Integrated Circuit) chip and two Intel NAND flash non-volatile memory components that enable faster resume to productivity after hibernate, providing additional power savings by limiting hard disk drive accesses and increasing application responsiveness for a richer user experience.
• ASIC Application Specific Integrated Circuit
• FIG. 1 is a block diagram of an electronic system in accordance with some embodiments of the invention.
• FIG. 2 is a block diagram of a processor-based system in accordance with some embodiments of the invention.
• Fig. 3 is a flow diagram in accordance with some embodiments of the invention.
• Fig. 4 is another flow diagram in accordance with some embodiments of the invention.
• Fig. 5 is another flow diagram in accordance with some embodiments of the invention.
• Fig. 6 is another flow diagram in accordance with some embodiments of the invention.
• Fig. 7 is another flow diagram in accordance with some embodiments of the invention.
• Fig. 8 is a graph of number unique sectors versus number of accesses for comparison with some embodiments of the invention.
• Fig. 9 is a graph of number unique sectors versus number of accesses in accordance with some embodiments of the invention.
• Fig. 10 is another flow diagram in accordance with some embodiments of the invention.
• an electronic system 10 may include a cache 12 located between a mass storage 14 and a system memory 16.
• the system 10 may include code stored on the electronic system 10 to prevent storage of stream data in the cache 12 and to send the stream data directly between the system memory 16 and the mass storage 14 based on a comparison of first metadata of a first request for first information and pre-boot stream information stored in a previous boot context of the electronic system 10.
• the code may be stored on the mass storage 14, the system memory 16, or another memory or storage device coupled to the electronic system 10.
• the code may further cause the electronic system 10 to cache the first information if the first information is not determined to be part of a streaming access based on the comparison of first metadata associated with the first request and the pre-boot stream information.
• the code may further cause the electronic system 10 to receive a second request for mass storage access, the second request to request second information, and, if the first information has not previously been determined to be part of the streaming access, determine whether the second information is part of the streaming access based on a comparison of first metadata associated with the first request and second metadata associated with the second request, and if the second information is determined to be part of the streaming access, store stream information corresponding to the streaming access which is persistent across different power states of the electronic system including a hard boot of the electronic system 10, wherein the stored stream information can be later used as the pre-boot stream information for subsequent boots.
• the code may further cause the system 10 to determine if the first information is part of a streaming access based on the comparison of first metadata associated with the first request and the pre-boot stream information, compare frequency information for the streaming access with a frequency threshold, perform the first request for mass storage access directly with the mass storage 14 if the first information is determined to be part of the streaming access and the frequency information for the streaming access does not exceed the frequency threshold, and cache the first information if the first information is determined to be part of the streaming access and the frequency information for the streaming access exceeds the frequency threshold.
• the code may further update the frequency information for the streaming access.
• a processor-based system 20 may include a processor 21 , a system memory 22 coupled to the processor 21 , a mass storage device 23, and a non-volatile cache memory 24 located between the system memory 22 and the mass storage device 23.
• the processor 21 may be a central processing unit (CPU).
• the system memory 22 may be a dynamic random access memory (DRAM).
• the system memory 22 may be coupled to the processor 21 via a memory controller hub (MCH) 25.
• the cache 24 may be a non-volatile memory (NVM) cache.
• the mass storage device 23 may be a rotating media such as a hard disk drive or an optical disk drive.
• the mass storage device 23 may be a non-rotating media such as a solid-state drive.
• both the cache 24 and the mass storage device 23 may be coupled to the MCH via an input / output controller hub (ICH) 26.
• ICH input / output controller hub
• Code stored on the processor-based system 20 may cause the processor- based system to receive a first request for mass storage access, the first request to request first information, retrieve pre-boot stream information stored during a previous boot context of the processor-based system, the previous boot context being other than a current boot context.
• the code may cause the processor-based system 20 to determine whether the first information is part of a streaming access based on a comparison of first metadata associated with the first request and the pre-boot stream information, and cache the first information if the first information is not determined be part of the streaming access based on the comparison of first metadata associated with the first request and the pre-boot stream information.
• the code may cause the processor-based system 20 to compare frequency information for the streaming access with a frequency threshold, perform the first request for mass storage access directly with the mass storage device 23 if the first information is determined to be part of the streaming access and the frequency information for the streaming access does not exceed the frequency threshold, and cache the first information if the first information is determined to be part of the streaming access and the frequency information for the streaming access exceeds the frequency threshold.
• the code may be stored on the mass storage device 23, the system memory 22, or another memory or storage device coupled to the processor-based system 20.
• the code may be stored as part of a basic input / output system (BIOS) 27 coupled to the ICH 26.
• BIOS basic input / output system
• the code may further cause the processor-based system 20 to receive a second request for mass storage access, the second request to request second information. If the first information has not previously been determined to be part of the streaming access, the code may cause the processor-based system 20 to determine whether the second information is part of the streaming access based on a comparison of the first metadata associated with the first request and second metadata associated with the second request. If the second information is determined to be part of the streaming access, the code may cause the processor-based system 20 to store stream information corresponding to the streaming access which is persistent across different power states including a hard boot of the processor-based system 20, wherein the stream information can be later used as the pre-boot stream information for subsequent boots.
• the code may further update the frequency information for the streaming access.
• the code may further compare length information for the streaming access with a stream length threshold, and prevent the storage of stream data in the cache 24 and send the stream data directly between the system memory 22 and the mass storage device 23 if the length information for the streaming access exceeds the length threshold.
• utilizing a cache in an electronic system may include locating a cache between a mass storage device and a system memory (e.g. at block 30), receiving a first request for mass storage access, the first request requesting first information (e.g.
• some embodiments of the invention may further include caching the first information (e.g. at block 35) if the first information is not determined to be part of the streaming access based on the comparison of the first metadata associated with the first request and the pre-boot stream information.
• some embodiments of the invention may further include receiving a second request for mass storage access, the second request requesting second information (e.g. at block 36), and, if the first information has not previously been determined to be part of the streaming access, determining whether the second information is part of the streaming access based on a comparison of the first metadata associated with the first request and second metadata associated with the second request (e.g.
• utilizing a cache in an electronic system may include locating a cache between a mass storage device and a system memory (e.g. at block 50), receiving a first request for mass storage access, the first request requesting first information (e.g.
• determining whether the first information is part of a streaming access based on a comparison of first metadata associated with the first request and other information associated with the streaming access e.g. at block 52
• comparing frequency information for the streaming access with a frequency threshold e.g. at block 53
• performing the first request for mass storage access directly with the mass storage device e.g. at block 54
• caching the first information e.g. at block 55
• the other stream information associated with the streaming access may be pre-boot stream information, stream information saved in the current boot context, or other information associated with the streaming access which can be used to identify the first information as part of the streaming access.
• some embodiments may further include updating the frequency information for the streaming access (e.g. at block 56).
• some embodiments may further include comparing length information for the streaming access with a stream length threshold (e.g. at block 57), and preventing the caching of the first information if the length information for the streaming access exceeds the length threshold (e.g. at block 58).
• utilizing a cache in a processor-based system may include locating a cache between a mass storage device and a system memory (e.g. at block 70), receiving a first request for mass storage access, the first request requesting first information (e.g. at block 71 ), retrieving pre-boot stream information stored during a previous boot context, the previous boot context being other than a current boot context (e.g. at block 72), determining whether the first information is part of a streaming access based on a comparison of first metadata associated with the first request and the pre-boot stream information (e.g. at block 73), and caching the first information (e.g.
• some embodiments may further include comparing frequency information for the streaming access with a frequency threshold (e.g. at block 75), performing the first request for mass storage access directly with the mass storage device (e.g. at block 76) if the first information is determined to be part of the streaming access and the frequency information for the streaming access does not exceed the frequency threshold, and caching the first information (e.g. at block 77) if the first information is determined to be part of the streaming access and the frequency information for the streaming access exceeds the frequency threshold.
• a frequency threshold e.g. at block 75
• performing the first request for mass storage access directly with the mass storage device e.g. at block 76
• caching the first information e.g. at block 77
• Some embodiments of the invention may further include updating the frequency information for the streaming access (e.g. at block 78). For example, some embodiments may further utilize the cache following entry point A (e.g. see Fig. 3). For example, some embodiments may further utilize the cache following entry point B (e.g. see Fig. 5).
• some embodiments of the invention may provide improved techniques for handling stream accesses between rotating media and a cache.
• Intel Turbo Memory provides a cache strategy which may speed hard disk accesses.
• the hard disk does well, and at the same time, may be detrimental to the cache.
• media file playback demands can readily be satisfied by the hard disk at the rates required for HD video display.
• temporal locality because such files are large and rarely reused, they violate a fundamental premise of caching, namely temporal locality. If these files were cached, the impact would be that locations with good temporal locality may be evicted from the cache to be replaced with those with bad temporal locality.
• some embodiments of the invention may improve the timing and quality of the decision of when and whether to cache information.
• the decision process may be complicated by the location of the cache driver software in the file system software stack.
• the cache driver software may not directly have access to file name information.
• the cache driver software may have access to limited information such as the following metadata:
• Command - Command to the mass storage device, e.g. a read or write;
• Length The number of sectors for the command.
• some embodiments of the invention may provide improved techniques to assemble and utilize this limited information available to the cache driver software to identify and manage streams.
• a stream may be a set of commands that represent the same command, and all or almost all of the sectors.
• a minimum size may be a useful parameter to detect and identify streams. The minimum size parameter may have the effect of making the cache seem larger, because unlikely candidates may be removed.
• Fig. 8 shows an example simulated cache utilization without the minimum size parameter
• Fig. 9 shows an example simulated cache utilization with a four megabyte (MB) minimum size parameter.
• MB megabyte
• information regarding the stream may be stored in metadata. Such information may be used to later identify the same stream, if later requested for access. Such metadata may be used to directly access the data for this later request from, for example, a disk drive and forego storage in the cache.
• a streaming access may be detected and identified by recording information regarding prior accesses into an array of potential stream information. Such an array may be stored in various locations, such as system memory or in the cache itself, for example. Each element in this array may include certain metadata, such as a starting location (e.g., starting disk address), a length, a count of the number of requests in the stream (or potential stream) and / or a timestamp indicating the arrival of the latest request in the stream (or potential stream). For example, after the stream is completed, the final length of the stream may be denoted in metadata.
• some embodiments of the invention may include identification of a stream (e.g. block 101 ). For example, this identification includes sequential accesses, but is not limited to only sequential accesses. For example, if the information is close enough to sequential, the access may also be detected as a stream. Both read accesses and write accesses may be identified as streams.
• some embodiments of the invention may identify two additional cases: a) when the stream is not accessed sequentially; and b) when the stream is not arranged sequentially on the mass storage device (e.g. a rotating media) due to, for example, fragmentation or compression (but not limited to these examples).
• Some embodiments of the invention may further include directing all remaining accesses of the stream directly to the mass storage device (e.g. hard drive) without insertion into the cache (e.g. block 102), marking any cached contents of a stream as "early eviction" candidates (e.g. block 103), prefetching and buffering of the stream in system memory (e.g. DRAM) (e.g. block 104), and closure of a stream when it is no longer valid (e.g. block 105), for example, due to reuse or time based aging.
• the mass storage device e.g. hard drive
• marking any cached contents of a stream as "early eviction" candidates e.g. block 103
• prefetching and buffering of the stream in system memory e.g. DRAM
• closure of a stream when it is no longer valid e.g. block 105
• Some embodiments of the invention may further include saving stream information across system power states (e.g. block 106).
• stream information that is detected in the current boot context may be retained over large periods of time (e.g. not limited to a particular length of time in minutes, hours or days) to be used as pre-boot stream information in subsequent boots.
• storing the pre-boot stream information may help detect streams on access of the first element of the stream itself thereby identifying the stream earlier (e.g. before the second access request) and making better utilization of the cache (e.g. by avoiding caching information from the first access request).
• the pre-boot stream information may be stored in a persistent storage device such as a hard disk, a non-volatile memory, or a battery powered memory.
• the system power states may include a reduced power state and a shutdown power state.
• Some processor-based systems utilize a variety of power states including, for example, a ready state, a stand-by state, a suspended state, a hibernation state, and an off state.
• a hard boot may correspond to a restart of the system from an off state.
• Some embodiments of the invention may include saving frequency of stream access information (e.g. block 107) and utilizing the stream frequency information to determine whether to perform the stream access directly to disk or save the stream access in the cache (e.g. block 108).
• a cache may be utilized in accordance with the following pseudo code:
• N the number of stream detection slots.
• IsAStream the number of sectors of streaming data before the start of filtering.
• M the number of known stream slots.
• storing stream information across system power states and / or saving stream frequency information may improve the timing and quality of deciding when and whether to store information in a cache.
• making better decisions regarding the utilization of cache memory may make small caches more effective, thereby reducing system size and / or cost.
• another advantage is that accesses which are unlikely to be beneficial to the cache may be removed, thereby improving responsiveness of the cache by removing queue traffic, while preserving good media performance from the mass storage device.
• caching frequently used streams of certain sizes may provide a better user experience.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
• Memory System Of A Hierarchy Structure (AREA)
• Information Transfer Between Computers (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Publications (2)

Family

ID=41445201

Family Applications (1)

Country Status (6)

Cited By (2)

Families Citing this family (12)

Family Cites Families (17)

• 2008
  • 2008-06-25 US US12/215,093 patent/US8433854B2/en not_active Expired - Fee Related
• 2009
  • 2009-06-09 JP JP2010547880A patent/JP5340315B2/ja not_active Expired - Fee Related
  • 2009-06-09 CN CN200980112390.9A patent/CN101981551B/zh not_active Expired - Fee Related
  • 2009-06-09 GB GB1015976.2A patent/GB2473149B/en not_active Expired - Fee Related
  • 2009-06-09 DE DE112009000418T patent/DE112009000418T5/de not_active Ceased
  • 2009-06-09 WO PCT/US2009/046731 patent/WO2009158183A2/en not_active Ceased
• 2013
  • 2013-05-21 JP JP2013106926A patent/JP5717790B2/ja not_active Expired - Fee Related

Also Published As

Similar Documents

Legal Events