(19)
(11)EP 2 808 783 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
27.11.2019 Bulletin 2019/48

(21)Application number: 12867234.2

(22)Date of filing:  29.06.2012
(51)Int. Cl.: 
G06F 12/0893  (2016.01)
G06F 12/0831  (2016.01)
G06F 12/0862  (2016.01)
G06F 12/0864  (2016.01)
(86)International application number:
PCT/CN2012/077953
(87)International publication number:
WO 2013/113206 (08.08.2013 Gazette  2013/32)

(54)

SMART CACHE AND SMART TERMINAL

INTELLIGENTER PUFFER-SPEICHER UND INTELLIGENTES ENDGERÄT

ANTÉMÉMOIRE INTELLIGENTE ET TERMINAL INTELLIGENT


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 01.02.2012 CN 201210022513

(43)Date of publication of application:
03.12.2014 Bulletin 2014/49

(73)Proprietor: ZTE Corporation
Shenzhen, Guangdong 518057 (CN)

(72)Inventors:
  • YUAN, Cissy
    Shenzhen, Guangdong 518057 (CN)
  • MAO, Erkun
    Shenzhen Guangdong 518057 (CN)
  • CHEN, Qian
    Shenzhen Guangdong 518057 (CN)
  • WANG, Jian
    Shenzhen Guangdong 518057 (CN)
  • TIAN, Xuehong
    Shenzhen Guangdong 518057 (CN)
  • ZENG, Daibing
    Shenzhen Guangdong 518057 (CN)
  • TIAN, Wanting
    Shenzhen Guangdong 518057 (CN)

(74)Representative: Herzog IP Patentanwalts GmbH 
Immermannstraße 40
40210 Düsseldorf
40210 Düsseldorf (DE)


(56)References cited: : 
EP-A1- 0 549 508
WO-A2-03/038626
CN-A- 101 788 927
CN-A- 102 289 390
US-A1- 2005 138 264
US-A1- 2008 010 408
US-B1- 6 334 177
US-B1- 6 678 790
EP-A2- 1 045 307
CN-A- 1 746 865
CN-A- 101 814 039
US-A1- 2004 215 893
US-A1- 2006 075 192
US-A2- 2011 145 626
US-B1- 6 347 346
  
      
    Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


    Description

    TECHNICAL FIELD



    [0001] The disclosure relates to the technology of intelligence cache, in particular to a flexible and configurable intelligence cache defined by software by the method of closely combining software and hardware, and an intelligence terminal having the intelligence cache.

    BACKGROUND



    [0002] By using a time or space locality of a processor executing programs, a traditional cache memory temporarily stores the latest and/or frequently executed instructions and data in the cache memory close to a processor unit. When it is required to access a certain instruction or data, the cache memory will be accessed first, and then a next-level memory of low speed and large memory space will be accessed if the cache memory is missing.

    [0003] WO 03/038626 A2 discloses a reconfigurable cache memory including a programmable memory unit, a functional unit in communication with the programmable memory unit, and reconfiguration module. WO 03/038626 A2 further discloses a method for reconfiguring cache memory including determining an optimal configuration of memory for a particular application executed by a functional unit using a programmable memory unit and programming the programmable memory unit to the optimal configuration.

    [0004] US 2008/010408 A1 discloses a method for reconfiguring a cache memory. The method includes analysing one or more characteristics of an execution entity accessing a cache memory and reconfiguring the cache based on the one or more characteristics analysed.

    [0005] EP 1045307 A2 discloses a configurable cache memory for improving the performance of a CPU. The CPU provides configuration data based on the program that being executed by the CPU. A portion of the cache memory is configured as one of a plurality of cache memory types wherein each of the plurality of cache memory types is associated with a corresponding index.

    [0006] US 6334177 B1 discloses a method for supporting software partition and dynamic reconfiguration within a non-uniform memory access computer system.

    [0007] EP 0549508 A1 discloses cache memory which is reconfigurable to accommodate differing operational requirements by effectively merging a predetermined number of least significant bits of the tag field of the main memory address with the cache memory address.

    [0008] CN101788927 A discloses a method for realizing dynamic allocation of internal resource by reconfiguring the units in the idle condition into a corresponding hardware circuit.

    [0009] Fig. 1 shows a typical cache structure. As shown in Fig. 1, since instructions and data loaded into a cache are updated in real time according to dynamic executions of programs, the processor core will first search a tag matrix in the cache to confirm whether the required instructions or data are in the cache. Once cache miss, the tag search and data comparisons are invalid, and then the next-level memory will be accessed, which results in a waste of a plurality of execution cycles of the processor and a waste of the power consumption of the cache. To increase the hit rate of the cache, a multilevel cache structure for example of set association, complex replacement operation, perfecting, predicting read and hierarchy are used. Of course, such performance improvements are dependent entirely on increasing hardware complexity and chip area overhead. Since the cache structure shown in Fig. 1 is a well known cache, the functions and principles of respective parts will not explained in detail here.

    [0010] Another disadvantage of cache is that access delays for hit and miss are completely different, it is difficult to predict the delays for accessing the cache, and Tightly Coupled Memory (TCM) is introduced in many occasions. TCM is a Static Random Access Memory (SRAM) close to the core of a processor, and characterized in its high speed and fixed delay. The contents of TCM cannot be replaced in real time, and the TCM has a fixed and small capacity. Refreshing the TCM is dependent completely on software scheduling. Before refreshing the TCM, software will find out when to perform the refreshing and performing corresponding configuration operations, during which the TCM is inaccessible. Those factors limit the application of TCM.

    [0011] Content Addressable Memory (CAM) is a dedicated memory. As a general module, it is incapable to make full use of its performance for some specific application scenarios. Further, comparing all the memory entries with input entries in parallel will lead to highly complicated and expensive hardware.

    [0012] Therefore, it is difficult to improve performance by relying entirely on hardware complexity, power consumption or software intervention. Further, the fine granularity of processor execution and memory access (as per instructions), resources are fixedly categorized and divided, which is inefficient and wastes the memory resources of a system. With the close combination of software and hardware, it is possible to perform flexible and intelligence processing according to the program execution and features of data structure, the performance can be improved greatly and the performance, power consumption, cost and the like are more balanced.

    SUMMARY



    [0013] In view of this, the present disclosure provides an intelligence cache and an intelligence terminal, capable of being flexibly defined, configured and reconstructed by software according to specific applications, and solving the disadvantages such as high complexity, high overhead, high energy consumption and unpredictable delay of conventional caches as well as the problems of low efficiency of updating TCM data and low flexibility and narrow application area of memory units.

    [0014] To achieve the objective above, the technical solution of the disclosure is implemented as follows:

    An intelligence cache, including: a general interface, a software define and reconfiguration unit, a control unit, a memory unit and an intelligence processing unit, wherein,

    the general interface is configured to receive configuration information and/or control information and/or data information from a core or a bus and return target data;

    the software define and reconfiguration unit is configured to define a memory as a required cache memory according to configuration information;

    the control unit is configured to control writing and reading of a cache memory, monitor an instruction and a data stream in real time, and control the memory unit to load the required data in advance according to system information, features of tasks to be executed and features of data structures used;

    the memory unit is composed of a number of memory modules and configured to cache data; wherein the required cache memory is formed by the memory modules according to the definition of the software define and reconfiguration unit;

    the intelligence processing unit is configured to process input and output data and perform transfer, convert and operate on data among multiple structures defined in the control unit;

    characterized in that the software define and reconfiguration unit is further configured to dynamically reconstruct idle memory modules in the operating process and to define cache memories with different properties and same structure that are simultaneously existent with types of memory structures including Tightly Coupled Memory, TCM, and Tightly Coupled Access Memory, TCAM, wherein the cache memories with different properties and same structure include at least one of the following: full associative cache, 16-way associative cache, 4-way associative cache, 2-way associative cache and direct-mapped cache.



    [0015] Preferably, the required cache memory can be configured to include at least one of the following memories: TCM, CAM and Cache.

    [0016] Preferably, the general interface further includes a coherent interface in the multi-core environment.

    [0017] Preferably, the intelligence processing unit transferring, converting and operating on data among multiple structures defined in the control unit includes:
    matrix operation, bit level operation, data search, data sorting, data comparison, logic operation, setting/resetting, read-modify-write operations, and operations of increment, decrement, addition and subtraction.

    [0018] Preferably, the intelligence processing unit is further configured to fill and update data and transfer data to a next-level memory.

    [0019] Preferably, the control unit loads data according to the size of data blocks defined by the software define and reconfiguration unit or loads data automatically; and a dedicated memory area is defined in the memory unit to load abnormal or disordered control programs.

    [0020] An intelligence terminal includes the intelligence cache above.

    [0021] The intelligence terminal includes a computer, a notebook, a cell phone, a personal digital assistant or a game machine, for example.

    [0022] The intelligence cache of the disclosure can process complex operations and disordered controls by a core only and submit a large amount of frequently used and simple data to the intelligence processing unit of the intelligence cache for processing. Furthermore, the data processing by the intelligence processing unit is not only related to individual data, but also to a whole specific data structure, so as to process data in proximity to memories as much as possible, thereby reducing the dependency on bus and the burden on core and achieving the balance among performance, power consumption, cost and the like. Due to close operation with software, even though there is no prefetching, predicting read and complex replacement algorithm, an efficient memory system can be realized according to the operating status of software, the features of tasks to be executed and the features of data structures through the flexible organization and management by the control unit and the close cooperation of the intelligence processing unit.

    BRIEF DESCRIPTION OF THE DRAWINGS



    [0023] 

    Fig. 1 shows a typical cache structure;

    Fig. 2 shows a structural diagram of an intelligence cache according to one embodiment of the disclosure;

    Fig. 3 shows a structural diagram of an intelligence cache according to first embodiment of the disclosure;

    Fig. 4 shows a structural diagram of an intelligence cache according to second embodiment of the disclosure;

    Fig. 5 shows a structural diagram of an intelligence cache according to third embodiment of the disclosure;

    Fig. 6 shows a structural diagram of the composition of an intelligence cache according to fourth embodiment of the disclosure;

    Fig. 7 shows a structural diagram of an intelligence cache according to fifth embodiment of the disclosure;

    Fig. 8 shows a structural diagram of an intelligence cache according to sixth embodiment of the disclosure;

    Fig. 9 shows a structural diagram of an intelligence cache according to seventh embodiment of the disclosure;

    Fig. 10 shows a structural diagram of an intelligence cache according to eighth embodiment of the disclosure; and

    Fig. 11 shows a structural diagram of an intelligence cache according to ninth embodiment of the disclosure.


    DETAILED DESCRIPTION



    [0024] The disclosure provides an intelligence cache, including: a general interface, a software define and reconfiguration unit, a control unit, a memory unit and an intelligence processing unit, wherein the general interface is configured to receive configuration information and/or control information and/or data information from a core or a bus and return target data; the software define and reconfiguration unit is configured to define a memory as a required cache memory according to configuration information; the control unit is configured to control writing and reading of the cache memory, monitor an instruction and a data stream in real time, and control the memory unit to load the required data in advance according to system information, features of tasks to be executed and features of data structures used; the required memory system structure is responsible for the function of associative arrays such as Cache TAG of caching data and formed by a data cache memory unit according to the definition of the software define and reconfiguration unit; and the intelligence processing unit is configured to process input and output data and transfer, convert and operating on data among multiple structures defined in the control unit.

    [0025] To highlight technical solutions and advantages of the disclosure, the disclosure is further explained below with reference to the embodiments and the accompanying drawings in detail.

    [0026] Fig. 2 shows a structural diagram of an intelligence cache according to one embodiment of the disclosure. It should be noted that the intelligence cache in the embodiment of the disclosure is also referred to as Software Define Cache Memory (SDCM); and in this application, SDCM represents the intelligence cache in the embodiment. As shown in Fig. 2, the intelligence cache of the disclosure mainly includes five processing units: a General Interface (GI), a Software Define and Reconfiguration Unit (SDRU), a Control Unit (CU), a Memory Unit (MU), and an Intelligence Processing Unit (IPU), wherein the GI is configured to receive configuration information and/or control information and/or data information from a core or a bus and return target data to the core or bus, and includes a data direct loading interface (for example, Direct Memory Access (DMA) interface) and a coherent interface in a multi-core environment. The SDRU is configured to define, according to configuration information, a memory of the SDCM as a required memory system structure, TCM, TCAM or Cache, for example, and define multiple cache memories of different properties and same structure, such as full associative cache, 4-way associative cache, 16-way associative cache, 2-way associative cache and direct-mapped cache, which can be simultaneously existent with other different types of memory structures such as TCM and TCAM. Idle memories can also be dynamically reconfigured in the operating process to make full use system memory resources; in addition to controlling reading and writing of a memory, the CU is further configured to monitor instructions and data streams in real time, and load the required data in advance according to system information, features of tasks to be executed and features of data structures used through close cooperation with the IPU to increase the hit rate in maximum; the MU is composed of a number of memory modules, and the functions of these modules can entirely be defined as required, and these modules can be configured to store indexes, tags, identifiers, data or other information, and combined freely to realize complex memory structures such as TCM, TCAM or Cache mentioned above. The IPU can process input and output data from SDCM, or transfer, convert and operate on data among several structures defined in the MU, such as matrix operation, bit level operation, data search, data sorting, data comparison, logic operation, setting/resetting, operations of reading, modification and writing, and operations of increment, decrement, addition and subtraction. The IPU can further cooperate with the CU to realize the data filling and update and data transfer to the next-level memory.

    [0027] Based on the SDCM and requirements on software, the whole memory system can be defined as Cache, TCM, CAM or other memory structures, and the properties of these memory structures are configurable such as cache size, degree of association, row size, allocation policy and write-back approach; and the cache interface can be configured as a coherent interface to apply to a multi-core architecture; even, a Translation Lookaside Buffer (TLB) can be defined for the cache to realize the conversion of virtual and real addresses. The sizes of TCM and CAM can also be configured, and TCM and CAM can even be configured as a memory system structure with multiple cache structures and TCMs.

    [0028] In the embodiments below, since the structure is relatively simple, the connections between respective functional units will not be explained, and those skilled in the art should understand that the respective processing units above can be connected by buses or dedicated interfaces.

    [0029] Fig. 3 shows a structural diagram of an intelligence cache according to first embodiment of the disclosure. As shown in Fig. 3, the SDCM defines full associative cache, 4-way associative cache, CAM, TCM and other memory structures. In the disclosure, all the data or indexes input or output by the memory structures defined by the SDCM can directly be processed in the MU by the IPU, such as conversion, bit insertion, bit lifting, set reset, displacement, bit reversal, and operations of increment, decrement, addition and subtraction, without submitting all the data processing tasks to the core, and data can flow among the memory structures in one MU to save the bandwidth of buses and reduce the burden of processors. The core is only responsible for complex operations and controls to improve the processing performance.

    [0030] The technical solution of the disclosure can realize various required memory system structures, that is, the memory system structures can be defined freely as required, and several common implementations are shown below. In Figs. 4 to 8, IF_CU represents all the units except for the memory and IPU of SDCM in the embodiments of the disclosure.

    [0031] Fig. 4 shows a structural diagram of an intelligence cache according to a second embodiment of the disclosure. As shown in Fig. 4, the SDCM is defined as the cache structure shown in Fig. 4 by simple definition commands, and the defined cache can work as a normal cache under the combined control of CU. Of course, to reduce the processing complexity and closely cooperate with software, the defined cache will not use a complex replacement algorithm and methods such as prefetching and predicting read that improve performance. Instead, according to the information provided by software and features of tasks and data, the CU completes the organization and management of data, and finishes loading data from the next-level memory system and updating the next-level memory system. The granularity for loading data and update to the next-level are processed based on tasks, and the required data will not be replaced when tasks are in process. Those skilled in the art should understand that according to the intelligence cache structure provided by the disclosure, the cache configuration shown in Fig. 4 is easy to implement.

    [0032] Fig. 5 shows a structural diagram of an intelligence cache according to a third embodiment of the disclosure. As shown in Fig. 5, when the SDCM is defined as TCAM, more memory blocks are required to read data records in parallel under the control of IF_CU, and the parallel data is compared with key words in the IPU which shows whether data is found successfully by outputting results. If so, IF_CU reads and outputs the corresponding index data in the memory blocks of the data records.

    [0033] Fig. 6 shows a structural diagram of an intelligence cache according to a fourth embodiment of the disclosure. As shown in Fig. 6, when the SDCM is defined as TCM, the CU is only required to finish functions similar to simple reading and writing ram operations. The TCM needs only 1 memory block in minimum, and the minimum reading and writing cycle is 1 clock cycle. Of course, TCMs of different sizes can be obtained by combining different number of memory blocks.

    [0034] In the disclosure, the SDCM can further be defined as a memory system with the capability of simple data processing, for example but not limited to bit operation, data search and matrix operation.

    [0035] Under the control of IPU, it is more convenient to perform the operations such as preamble "0" or "1", and bit reversal (the bit switch of most significant bit and least significant bit, second most significant bit and second least significant bit and so on). It should be noted that the operations of data bits above are normal processing, so the implementation detail will not be explained here.

    [0036] Fig. 7 shows a structural diagram of an intelligence cache according to a fifth embodiment of the disclosure, which is an example of the intelligence cache structure defined by the SDCM and capable of performing the bit operation. As shown in Fig. 7, the data to be processed is stored in the cache, and the SDCM can return the result to the core or still store the result in the cache as long as the core sends a command for reading the preamble "0" or "1" of data and bit operation. The way how the SDCM performs computation is transparent for the core.

    [0037] The structure shown in Fig. 7 can also be used for realizing data search. As long as the core notifies the SDCM of the array to be searched for and data of interest, the IF_CU is responsible for reading data in the array of the cache and submitting the data to the IPU by a simple instruction of search task, and the IPU compares and determines whether the data is target data, and if so, returns to the address where the data is located through an interface. Such structure in the disclosure reduces the burden on processors and saves the bus bandwidth of processors, so that the processors can focus on the processing of other tasks.

    [0038] The SDCM can further perform data conversion on a large amount of data, such as matrix transposition. Fig. 8 shows a structural diagram of an intelligence cache according to a sixth embodiment of the disclosure, which is an example of the intelligence cache structure defined by the SDCM and capable of performing the matrix operation. As shown in Fig. 8, the SDCM can be defined as the structure shown in Fig. 8. During the matrix transposition, as long as the core gives the storage position information of the original matrix and the size of the matrix data block, and gives a command of starting computation, the SDCM can finish the transposition. The SDCM first reads column vectors of the original matrix from the cache into the TCM and writes the column vectors back to the cache as row vectors by IF_CU. Here, TCM is composed of multiple small blocks and written in a single block, while, under the control of IPU, it is able to realize the simultaneous reading of multiple blocks and perform bit extraction from the read data to realize the transposition. In addition, the offset computation of all addresses is finished by the IPU, and a transposed matrix of the original matrix is formed in the cache after all data in the column vectors is transferred. The processing time of such transposition is related to the matrix size, and software needs to know the delay of finishing the transposition.

    [0039] Fig. 9 shows a structural diagram of an intelligence cache according to a seventh embodiment of the disclosure, which is an example of the SDCM used as a shared memory. The SDCM is connected to an operating system though the GI, and the connection way can be standard bus or network on chip. The position of the SDCM in the system can be the shared memory system as shown in Fig. 9. In addition to the slave interface, the SDCM has the function of master interface, so it can launch the data transfer anytime and anywhere.

    [0040] Indeed, the SDCM can also be a private memory of the core, or form a hierarchical memory structure. Figs. 10 shows a structural diagram of an intelligence cache according to an eighth embodiment of the disclosure, and Fig. 11 shows a structural diagram of an intelligence cache according to a ninth embodiments of the disclosure. As shown in Figs. 10 and 11, SDCMs are used as private memories in the core or a hierarchical memory structure. The SDCMs can be connected as a Symmetrical Multiple Processor (SMP) structure as shown in Fig. 10 or an Asymmetric Multiple Processor (AMP) structure as shown in Fig. 11. Whether SMP or AMP, the coherence among SDCMs can be realized.

    [0041] Those skilled in the art should understand that according to the requirements on application the SDCMs can be connected as many structures, even it is possible to add one or more SDCMs in the other current memory structures.

    [0042] The disclosure further records an intelligence terminal including the intelligence cache above. The intelligence terminal includes a computer, a notebook, a cell phone, a personal digital assistant, a console or others having a CPU control unit, for example.

    [0043] What explained above are only the preferred embodiments of the disclosure and not intended to limit the scope of protection of the disclosure.

    INDUSTRIAL APPLICABILITY



    [0044] The intelligence cache of the disclosure can process complex operations and disordered controls by a core only and submit a large amount of frequently used and simple data to the intelligence processing unit of the intelligence cache for processing. Furthermore, the data processing by the intelligence processing unit is not only related to individual data, but also to a whole specific data structure, so as to process data in proximity to memories as much as possible, thereby reducing the dependency on bus and the burden on core and achieving the balance among performance, power consumption, cost and the like.


    Claims

    1. An intelligence cache, comprising: a general interface, a software define and reconfiguration unit, a control unit, a memory unit and an intelligence processing unit, wherein
    the general interface is configured to receive configuration information and/or control information, and/or data information from a core or a bus, and return target data;
    the software define and reconfiguration unit is configured to define a memory as a required cache memory according to the configuration information;
    the control unit is configured to control writing and reading of a cache memory and monitor instructions and data streams in real time, and control the memory unit to load the required data in advance according to system information, features of tasks to be executed and features of data structures used;
    the memory unit is composed of a number of memory modules and configured to cache data; wherein the required cache memory is formed by the memory modules according to the definition of the software define and reconfiguration unit; and
    the intelligence processing unit is configured to process input and output data, and transfer, convert and operate on data among multiple structures defined in the control unit;
    characterized in that the software define and reconfiguration unit is further configured to dynamically reconstruct idle memory modules in the operating process and to define cache memories with different properties and same structure that are simultaneously existent with types of memory structures including Tightly Coupled Memory, TCM, and Tightly Coupled Access Memory, TCAM, the cache memories with different properties and same structure including at least one of: full associative cache, 16-way associative cache, 4-way associative cache, 2-way associative cache and direct-mapped cache.
     
    2. The intelligence cache according to claim 1, wherein the required cache memory can be configured as at least one of the following:
    Tightly Coupled Memory, TCM, Content Addressable Memory, CAM, and Cache.
     
    3. The intelligence cache according to claim 1, wherein the general interface further comprises a coherent interface in a multi-core environment.
     
    4. The intelligence cache according to claim 1, wherein the intelligence processing unit transferring, converting and operating on data among multiple structures defined in the control unit comprises:
    matrix operation, bit level operation, data search, data sorting, data comparison, logic operation, setting/resetting, read-modify-write operations, and operations of increment, decrement, addition and subtraction.
     
    5. The intelligence cache according to claim 1, wherein the intelligence processing unit is further configured to fill and update data and transfer data to a next-level memory.
     
    6. The intelligence cache according to claim 1, wherein the control unit loads data according to the size of data blocks defined by the software define and reconfiguration unit or loads data automatically; and a dedicated memory area is defined in the memory unit to load abnormal or disordered control programs.
     
    7. An intelligence terminal, comprising the intelligence cache according to any one of claims 1 to 6.
     
    8. The intelligence terminal according to claim 7, comprising: a computer, a notebook, a cell phone, a personal digital assistant or a game machine.
     


    Ansprüche

    1. Intelligenzcache, der Folgendes umfasst: eine allgemeine Schnittstelle, eine Softwaredefinitions- und -rekonfigurationseinheit, eine Steuereinheit, eine Speichereinheit und eine Intelligenzverarbeitungseinheit, wobei
    die allgemeine Schnittstelle dazu ausgelegt ist, Konfigurationsinformationen und/oder Steuerinformationen und/oder Dateninformationen von einem Kern oder einem Bus zu empfangen und Zieldaten zurückzugeben;
    die Softwaredefinitions- und -rekonfigurationseinheit dazu ausgelegt ist, gemäß den Auslegungsinformationen einen Speicher als einen erforderlichen Cachespeicher zu definieren;
    die Steuereinheit dazu ausgelegt ist, das Schreiben und Lesen eines Cachespeichers zu steuern und Anweisungen und Datenströme in Echtzeit zu überwachen und die Speichereinheit zu steuern, um die erforderlichen Daten gemäß Systeminformationen, Merkmalen von auszuführenden Aufgaben und Merkmalen von verwendeten Datenstrukturen vorab zu laden;
    die Speichereinheit aus einer Reihe von Speichermodulen zusammengesetzt und dazu ausgelegt ist, Daten zwischenzuspeichern; wobei der erforderliche Cachespeicher gemäß der Definition der Softwaredefinitions- und -rekonfigurationseinheit von den Speichermodulen gebildet wird; und
    die Intelligenzverarbeitungseinheit dazu ausgelegt ist, Eingangs- und Ausgangsdaten zu verarbeiten und Daten unter mehreren Strukturen, die in der Steuereinheit definiert sind, zu übertragen, umzuwandeln und zu verarbeiten;
    dadurch gekennzeichnet, dass die Softwaredefinitions- und -rekonfigurationseinheit ferner dazu ausgelegt ist, ruhende Speichermodule im Betriebsprozess dynamisch neu zu konstruieren und Cachespeicher mit verschiedenen Eigenschaften und selben Strukturen, die mit Typen von Speicherstrukturen, die eng gekoppelten Speicher, TCM, und eng gekoppelten Zugriffsspeicher, TCAM, beinhalten, gleichzeitig existieren, zu definieren, wobei die Cachespeicher mit verschiedenen Eigenschaften und selben Strukturen mindestens eines von Folgendem beinhalten: vollassoziativen Cache, 16-fach-assoziativen Cache, 4-fach-assoziativen Cache, 2-fach-assoziativen Cache und Direktzuordnungscache.
     
    2. Intelligenzcache nach Anspruch 1, wobei der erforderliche Cachespeicher als mindestens eines von Folgendem ausgelegt werden kann:
    eng gekoppelter Speicher, TCM, inhaltsadressierbarer Speicher, CAM, und Cache.
     
    3. Intelligenzcache nach Anspruch 1, wobei die allgemeine Schnittstelle ferner eine kohärente Schnittstelle in einer Mehrkernumgebung umfasst.
     
    4. Intelligenzcache nach Anspruch 1, wobei die Intelligenzverarbeitungseinheit, die Daten unter mehreren Strukturen, die in der Steuereinheit definiert sind, überträgt, umwandelt und vearbeitet, Folgendes umfasst:
    Matrixbetrieb, Bitebenenbetrieb, Datensuche, Datensortierung, Datenvergleich, Logikbetrieb, Einstellung/Zurücksetzung, Lese-/Modifizierungs-/Schreibvorgänge und Vorgänge der Inkrementierung, Dekrementierung, Addition und Subtraktion.
     
    5. Intelligenzcache nach Anspruch 1, wobei die Intelligenzverarbeitungseinheit ferner dazu ausgelegt ist, Daten aufzufüllen und zu aktualisieren und Daten zu einem Speicher einer nächsten Ebene zu übertragen.
     
    6. Intelligenzcache nach Anspruch 1, wobei die Steuereinheit Daten gemäß der Größe von Datenblöcken, die von der Softwaredefinitions- und -rekonfigurationseinheit definiert wird, lädt oder Daten automatisch lädt und ein dedizierter Speicherbereich in der Speichereinheit definiert ist, um unnormale oder ungeordnete Steuerprogramme zu laden.
     
    7. Intelligentes Endgerät, das den Intelligenzcache nach einem der Ansprüche 1 bis 6 umfasst.
     
    8. Intelligentes Endgerät nach Anspruch 7, das Folgendes umfasst: einen Computer, ein Notebook, ein Mobiltelefon, einen persönlichen digitalen Assistenten oder eine Spielemaschine.
     


    Revendications

    1. Cache intelligent, comprenant : une interface générale, une unité de définition et de reconfiguration de logiciel, une unité de contrôle, une unité de mémoire et une unité de traitement intelligente,
    l'interface générale étant configurée pour recevoir une information de configuration et/ou une information de contrôle et/ou une information de données en provenance d'un cœur ou d'un bus, et renvoyer des données cibles ;
    l'unité de définition et de reconfiguration de logiciel étant configurée pour définir une mémoire en tant que mémoire cache requise selon l'information de configuration ;
    l'unité de contrôle étant configurée pour contrôler l'accessibilité en écriture et en lecture d'une mémoire cache et surveiller des instructions et des flux de données en temps réel, et contrôler l'unité de mémoire pour charger les données requises à l'avance selon une information de système, des attributs de tâches à exécuter et des attributs de structures de données utilisées ;
    l'unité de mémoire étant composée d'un certain nombre de modules de mémoire et configurée pour mettre en cache des données ; la mémoire cache requise étant formée par les modules de mémoire selon la définition de l'unité de définition et de reconfiguration de logiciel ; et
    l'unité de traitement intelligente étant configurée pour traiter des données d'entrée et de sortie, et transférer, convertir et exploiter des données parmi de multiples structures définies dans l'unité de contrôle ;
    le cache intelligent étant caractérisé en ce que l'unité de définition et de reconfiguration de logiciel est en outre configurée pour reconstruire dynamiquement des modules de mémoire inactifs dans le processus d'exploitation et pour définir des mémoires caches avec différentes propriétés et une même structure qui existent simultanément avec des types de structures de mémoire incluant une mémoire étroitement couplée (TCM) et une mémoire d'accès étroitement couplée (TCAM), les mémoires caches avec différentes propriétés et une même structure incluant au moins un cache parmi : un cache pleinement associatif, un cache associatif à 16 voies, un cache associatif à 4 voies, un cache associatif à 2 voies et un cache à correspondance préétablie.
     
    2. Cache intelligent selon la revendication 1, dans lequel la mémoire cache requise peut être configurée comme au moins un des éléments suivants :
    une mémoire étroitement couplée (TCM), une mémoire adressable par le contenu (CAM) et un cache.
     
    3. Cache intelligent selon la revendication 1, dans lequel l'interface générale comprend en outre une interface inhérente dans un environnement multicœur.
     
    4. Cache intelligent selon la revendication 1, dans lequel le transfert, la conversion et l'exploitation, par l'unité de traitement intelligente, de données parmi de multiples structures définies dans l'unité de contrôle, comprennent :
    une opération matricielle, une opération au niveau binaire, une recherche de données, un tri de données, une comparaison de données, une opération logique, une initialisation/réinitialisation, des opérations de lecture-modification-écriture et des opérations d'incrémentation, de décrémentation, d'addition et de soustraction.
     
    5. Cache intelligent selon la revendication 1, dans lequel l'unité de traitement intelligente est en outre configurée pour remplir et mettre à jour des données et transférer des données à une mémoire d'un niveau suivant.
     
    6. Cache intelligent selon la revendication 1, dans lequel l'unité de contrôle charge des données selon la taille de blocs de données définies par l'unité de définition et de reconfiguration de logiciel ou charge des données automatiquement ; et une zone de mémoire dédiée est définie dans l'unité de mémoire pour charger des programmes de contrôle anormaux ou désordonnés.
     
    7. Terminal intelligent, comprenant le cache intelligent selon l'une quelconque des revendications 1 à 6.
     
    8. Terminal intelligent selon la revendication 7, comprenant : un ordinateur, un ordinateur portable, un téléphone cellulaire, un assistant numérique personnel ou une machine de jeu.
     




    Drawing
























    REFERENCES CITED IN THE DESCRIPTION



    This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

    Patent documents cited in the description