System and method for increasing capacity, performance, and flexibility of flash storage

[0001] This application is a continuation of U.S. Ser. No. 13/276,212, which was filed Oct. 18, 2011 now U.S. Pat. No. 8,370,566, which is a continuation of U.S. patent application Ser. No. 11/611,374, which was filed Dec. 15, 2006 now U.S. Pat. No. 8,055,833, which claimed the benefit of priority to U.S. Provisional Patent Application Ser. No. 60/849,631, filed on Oct. 5, 2006. Each of the prior applications is incorporated herein by reference in its entirety. However, insofar as any definitions, information used for claim interpretation, or other disclosure from the above identified application conflicts with that set forth herein, such definitions, information, etc. in the present application should apply.

[0002] The present invention relates to memory, and more particularly to enhanced capacity, performance, flexibility, and reliability in multiple flash memory circuit systems.

[0003] Flash memory devices are gaining wide popularity and are used in many products such as MP3 players, USB storage keys, digital still cameras, even flash hard drives. These applications demand higher capacity, and higher performance while the marketplaces require ever lower and lower cost. The increase in flash capacity is limited by process technology, die size and production cost. Novel solutions are required to increase capacity, performance, and flexibility of flash while still resulting in cost effective implementations.

[0015] While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims.

[0016] This description includes numerous embodiments of flash devices and flash interface circuits. Embodiments are contemplated that include any combination of one or more of the features described below, including an embodiment that comprises all features described below. Embodiments including any subset or superset of the features and other features are also contemplated.

Flash Interface Circuit

[0017] FIG. 1 shows a block diagram of several flash memory devices 104A-104N connected to a system 106 by way of a flash interface circuit 102. The system 106 may include a flash memory controller 108 configured to interface to flash memory devices. The flash interface circuit 102 is a device which exposes multiple flash memory devices attached to the flash interface circuit 102 as at least one flash memory device to the rest of the system (e.g. the flash memory controller). The flash memory device(s) exposed to the rest of the system may be referred to as virtual flash memory device(s). One or more attributes of the virtual flash memory device(s) may differ from the attributes of the flash memory devices 104A-140N. Thus, the flash memory controller 108 may interface to the flash interface circuit 102 as if the flash interface circuit 102 were the virtual flash device(s). Internally, the flash interface circuit 102 translates a request from the system 106 into requests to flash memory devices 104A-104N and responses from flash memory devices 104A-104N into a response to the system 106. During discovery of flash configuration by the system 106, the flash interface circuit 102 presents modified information to the system 106. That is, the information presented by the flash interface circuit 102 during discovery differs in one or more aspects from the information that the flash memory devices 104A-104N would present during discovery.

[0018] FIG. 1 shows a block diagram of, for example, one or more small flash memory devices 104A-104N connected to a flash interface circuit 102. Also shown are exemplary connections of data bus & control signals between flash memory devices 104A-104N and a flash interface circuit 102. Also shown are exemplary data bus & control signals between the flash interface circuit 102 and a host system 106. In general, one more signals of the interface (address, data, and control) to the flash memory devices 104A-104N may be coupled to the flash interface circuit 102 and zero or more signals of the interface to the flash memory devices 104A-104N may be coupled to the system 106. In various embodiments, the flash interface circuit 102 may be coupled to all of the interface or a subset of the signals forming the interface. In FIG. 1, the flash interface circuit 102 is coupled to L signals (where L is an integer greater than zero) and the system 106 is coupled to M signals (where M is an integer greater than or equal to zero). Similarly, the flash interface circuit 102 is coupled to S signals to the system 106 in FIG. 1 (where S is an integer greater than zero).

[0019] In one embodiment, the flash interface circuit 102 may expose a number of attached flash memory devices 104A-104N as a smaller number of flash memory devices having a larger storage capacity. For example, the flash interface circuit may expose 1, 2, 4, or 8 attached flash memory devices 104A-104N to the host system as 1, 2 or 4 flash memory devices. Embodiments are contemplated in which the same number of flash devices are attached and presented to the host system, or in which fewer flash devices are presented to the host system than are actually attached. Any number of devices may be attached and any number of devices may be presented to the host system by presentation to the system in a manner that differs in at least one respect from the presentation to the system that would occur in the absence of the flash interface circuit 102.

[0020] For example, the flash interface circuit 102 may provide vendor-specific protocol translation between attached flash memory devices and may present itself to host as a different type of flash, or a different configuration, or as a different vendor's flash device. In other embodiments, the flash interface circuit 102 may present a virtual configuration to the host system emulating one or more of the following attributes: a desired (smaller or larger) page size, a desired (wider or narrower) bus width, a desired (smaller or larger) block size, a desired redundant storage area (e.g. 16 bytes per 512 bytes), a desired plane size (e.g. 2 Gigabytes), a desired (faster) access time with slower attached devices, a desired cache size, a desired interleave configuration, auto configuration, and open NAND flash interface (ONFI).

[0021] Throughout this disclosure, the flash interface circuit may alternatively be termed a “flash interface circuit”, or a “flash interface device”. Throughout this disclosure, the flash memory chips may alternatively be termed “memory circuits”, or a “memory device”, or as “flash memory device”, or as “flash memory”.

[0022] FIG. 2 shows another embodiment with possible exemplary connections between the host system 204, the flash interface circuit 202 and the flash memory devices 206A-206D. In this example, all signals from the host system are received by the flash interface circuit before presentation to the flash memory devices. And all signals from the flash memory devices are received by the flash interface circuit before being presented to the host system 204. For example, address, control, and clock signals 208 and data signals 210 are shown in FIG. 2. The control signals may include a variety of controls in different embodiments. For example, the control signals may include chip select signals, status signals, reset signals, busy signals, etc.

[0023] For the remainder of this disclosure, the flash interface circuit will be referred to. The flash interface circuit may be, in various embodiments, the flash interface circuit 102, the flash interface circuit 202, or other flash interface circuit embodiments (e.g. embodiments shown in FIGS. 3-6). Similarly, references to the system or the host system may be, in various embodiments, the host system 106, the host system 204, or other embodiments of the host system. The flash memory devices may be, in various embodiments, the flash memory devices 104A-104N, the flash memory devices 206A-206D, or other embodiments of flash memory devices.

Relocating Bad Blocks

[0024] A flash memory is typically divided into sub-units, portions, or blocks. The flash interface circuit can be used to manage relocation of one or more bad blocks in a flash memory device transparently to the system and applications. Some systems and applications may not be designed to deal with bad blocks since the error rates in single level NAND flash memory devices were typically small. This situation has, however, changed with multi-level NAND devices where error rates are considerably increased.

[0025] In one embodiment the flash interface circuit may detect the existence of a bad block by means of monitoring the error-correction and error-detection circuits. The error-correction and error-detection circuits may signal the flash interface circuit when errors are detected or corrected. The flash interface circuit may keep a count or counts of these errors. As an example, a threshold for the number of errors detected or corrected may be set. When the threshold is exceeded the flash interface circuit may consider certain region or regions of a flash memory as a bad block. In this case the flash memory may keep a translation table that is capable of translating a logical block location or number to a physical location or number. In some embodiments the flash interface circuit may keep a temporary copy of some or all of the translation tables on the flash memories. When a block is accessed by the system, the combination of the flash interface circuit and flash memory together with the translation tables may act to ensure that the physical memory location that is accessed is not in a bad block.

[0026] The error correction and/or error detection circuitry may be located in the host system, for example in a flash memory controller or other hardware. Alternatively, the error correction and/or error detection circuitry may be located in the flash interface circuit or in the flash memory devices themselves.

Increased ECC Protection

[0027] A flash memory controller is typically capable of performing error detection and correction by means of error-detection and correction codes. A type of code suitable for this purpose is an error-correcting code (ECC). Implementations of ECC may be found in Multi-Level Cell (MLC) devices, in Single-Level Cell (SLC) devices, or in any other flash memory devices.

[0028] In one embodiment, the flash interface circuit can itself generate and check the ECC instead of or in combination with, the flash memory controller. Moving some or all of the ECC functionality into a flash interface circuit enables the use of MLC flash memory devices in applications designed for the lower error rate of a SLC flash memory devices.

Flash Driver

[0029] A flash driver is typically a piece of software that resides in host memory and acts as a device driver for flash memory. A flash driver makes the flash memory appear to the host system as a read/write memory array. The flash driver supports basic file system functions (e.g. read, write, file open, file close etc.) and directory operation (e.g. create, open, close, copy etc.). The flash driver may also support a security protocol.

[0030] In one embodiment, the flash interface circuit can perform the functions of the flash driver (or a subset of the functions) instead of, or in combination with, the flash memory controller. Moving some or all of the flash driver functionality into a flash interface circuit enables the use of standard flash devices that do not have integrated flash driver capability and/or standard flash memory controllers that do not have integrated flash driver capability. Integrating the flash driver into the flash interface circuit may thus be more cost-effective.

Garbage Collection

[0031] Garbage collection is a term used in system design to refer to the process of using and then collecting, reclaiming, and reusing those areas of host memory. Flash file blocks may be marked as garbage so that they can be reclaimed and reused. Garbage collection in flash memory is the process of erasing these garbage blocks so that they may be reused. Garbage collection may be performed, for example, when the system is idle or after a read/write operation. Garbage collection may be, and generally is, performed as a software operation.

[0032] In one embodiment, the flash interface circuit can perform garbage collection instead of, or in combination with, the flash memory controller. Moving some or all of the garbage collection functionality into a flash interface circuit enables the use of standard flash devices that do not have integrated garbage collection capability and/or standard flash memory controllers that do not have integrated garbage collection capability. Integrating the garbage collection into the flash interface circuit may thus be more cost-effective.

Wear Leveling

[0033] The term leveling, and in particular the term wear leveling, refers to the process to spread read and write operations evenly across a memory system in order to avoid using one or more areas of memory heavily and thus run the risk of wearing out these areas of memory. A NAND flash often implements wear leveling to increase the write lifetime of a flash file system. To perform wear leveling, files may be moved in the flash device in order to ensure that all flash blocks are utilized relatively evenly. Wear leveling may be performed, for example, during garbage collection. Wear leveling may be, and generally is, performed as a software operation.

[0034] In one embodiment, the flash interface circuit can perform wear leveling instead of, or in combination with, the flash memory controller. Moving some or all of the wear leveling functionality into a flash interface circuit enables the use of standard flash devices that do not have integrated wear leveling capability and/or standard flash memory controllers that do not have integrated wear leveling capability. Integrating the wear leveling into the flash interface circuit may thus be more cost-effective.

Increasing Erase and Modify Bandwidth

[0035] Typically, flash memory has a low bandwidth (e.g. for read, erase and write operations, etc.) and high latency (e.g. for read and write operations) that are limits to system performance. One limitation to performance is the time required to erase the flash memory cells. Prior to writing new data into the flash memory cells, those cells are erased. Thus, writes are often delayed by the time consumed to erase data in the flash memory cells to be written.

[0036] In a first embodiment that improves erase performance, logic circuits in the flash interface circuit may perform a pre-erase operation (e.g. advanced scheduling of erase operations, etc.). The pre-erase operation may erase unused data in one or more blocks. Thus when a future write operation is requested the block is already pre-erased and associated time delay is avoided.

[0037] In a second embodiment that improves erase performance, data need not be pre-erased. In this case performance may still be improved by accepting transactions to a portion or portion(s) of the flash memory while erase operations of the portion or portion(s) is still in progress or even not yet started. The flash interface circuit may respond to the system that an erase operation of these portion(s) has been completed, despite the fact that it has not. Writes into these portion(s) may be buffered by the flash interface circuit and written to the portion(s) once the erase is completed.

Reducing Read Latency by Prefetching

[0038] In an embodiment that reduces read latency, logic circuits in the flash interface circuit may perform a prefetching operation. The flash interface circuit may read data from the flash memory ahead of a request by the system. Various prefetch algorithms may be applied to predict or anticipate system read requests including, but not limited to, sequential, stride based prefetch, or non-sequential prefetch algorithms. The prefetch algorithms may be based on observations of actual requests from the system, for example.

[0039] The flash interface circuit may store the prefetched data read from the flash memory devices in response to the prefetch operations. If a subsequent read request from the system is received, and the read request is for the prefetched data, the prefetched data may be returned by the flash interface circuit to the system without accessing the flash memory devices. In one embodiment, if the subsequent read request is received while the prefetch operation is outstanding, the flash interface circuit may provide the read data upon completion of the prefetch operation. In either case, read latency may be decreased.

Increasing Write Bandwidth

[0040] In an embodiment that improves write bandwidth, one or more flash memory devices may be connected to a flash interface circuit. The flash interface circuit may hold (e.g. buffer etc.) write requests in internal SRAM and write them into the multiple flash memory chips in an interleaved fashion (e.g. alternating etc.) thus increasing write bandwidth. The flash interface circuit may thus present itself to system as a monolithic flash memory with increased write bandwidth performance.

Increasing Bus Bandwidth

[0041] The flash memory interface protocol typically supports either an 8-bit or 16-bit bus. For an identical bus frequency of operation, a flash memory with a 16-bit bus may deliver up to twice as much bus bandwidth as a flash memory with an 8-bit bus. In an embodiment that improves the data bus bandwidth, the flash interface circuit may be connected to one or more flash memory devices. In this embodiment, the flash interface circuit may interleave one or more data busses. For example, the flash interface circuit may interleave two 8-bit busses to create a 16-bit bus using one 8-bit bus from each of two flash memory devices. Data is alternately written or read from each 8-bit bus in a time-interleaved fashion. The interleaving allows the flash interface circuit to present the two flash memories to the system as a 16-bit flash memory with up to twice the bus bandwidth of the flash memory devices connected to the flash interface circuit. In another embodiment, the flash interface circuit may use the data buses of the flash memory devices as a parallel data bus. For example, the address and control interface to the flash memory devices may be shared, and thus the same operation is presented to each flash memory device concurrently. The flash memory device may source or sink data on its portion of the parallel data bus. In either case, the effective data bus width may be N times the width of one flash memory device, where N is a positive integer equal to the number of flash memory devices.

Cross-Vendor Compatibility

[0042] The existing flash memory devices from different vendors may use similar, but not identical, interface protocols. These different protocols may or may not be compatible with each other. The protocols may be so different that it is difficult or impossible to design a flash memory controller that is capable of controlling all possible combinations of protocols. Therefore system designers must often design a flash memory controller to support a subset of all possible protocols, and thus a subset of flash memory vendors. The designers may thus lock themselves into a subset of available flash memory vendors, reducing choice and possibly resulting in a higher price that they must pay for flash memory.

[0043] In one embodiment that provides cross-vendor compatibility, the flash interface circuit may contain logic circuits that may translate between the different protocols that are in use by various flash memory vendors. In such an embodiment, the flash interface circuit may simulate a flash memory with a first protocol using one or more flash memory chips with a second protocol. The configuration of the type (e.g. version etc.) of protocol may be selected by the vendor or user (e.g. by using a bond-out option, fuses, e-fuses, etc.). Accordingly, the flash memory controller may be designed to support a specific protocol and that protocol may be selected in the flash interface circuit, independent of the protocol(s) implemented by the flash memory devices.

Protocol Translation

[0044] NAND flash memory devices use a certain NAND-flash-specific interface protocol. NOR flash memory devices use a different, NOR-flash-specific protocol. These different NAND and NOR protocols may not and generally are not compatible with each other. The protocols may be so different that it is difficult or impossible to design a flash memory controller that is capable of controlling both NAND and NOR protocols.

[0045] In one embodiment that provides compatibility with NOR flash, the flash interface circuit may contain logic circuits that may translate between the NAND protocols that are in use by the flash memory and a NOR protocol that interfaces to a host system or CPU. Similarly, an embodiment that provides compatibility with NAND flash may include a flash interface circuit that contains logic circuits to translate between the NOR protocols used by the flash memory and a NAND protocol that interfaces to a host system or CPU.

Backward Compatibility Using Flash Memory Device Stacking

[0046] As new flash memory devices become available, it is often desirable or required to maintain pin interface compatibility with older generations of the flash memory device. For example a product may be designed to accommodate a certain capacity of flash memory that has an associated pin interface. It may then be required to produce a second generation of this product with a larger capacity of flash memory and yet keep as much of the design unchanged as possible. It may thus be desirable to present a common pin interface to a system that is compatible with multiple generations (e.g. successively larger capacity, etc.) of flash memory.

[0047] FIG. 3 shows one embodiment that provides such backward compatibility, the flash interface circuit 310 may be connected by electrical conductors 330 to multiple flash memory devices 320 in a package 300 having an array of pins 340 with a pin interface (e.g. pinout, array of pins, etc.) that is the same as an existing flash memory chip (e.g. standard pinout, JEDEC pinout, etc.). In this manner the flash interface circuit enables the replacement of flash memory devices in existing designs with a flash memory device that may have higher capacity, higher performance, lower cost, etc. The package 300 may also optionally include voltage conversion resistors or other voltage conversion circuitry to supply voltages for electrical interfaces of the flash interface circuit, if supply voltages of the flash devices differ from those of the flash interface circuit.

[0048] The pin interface implemented by pins 340, in one exemplary embodiment, may include a x8 input/output bus, a command latch enable, an address latch enable, one or more chip enables (e.g. 4), read and write enables, a write protect, one or more ready/busy outputs (e.g. 4), and power and ground connections. Other embodiments may have any other interface. The internal interface on conductors 330 may differ (e.g. a x16 interface), auto configuration controls, different numbers of chip enables and ready/busy outputs (e.g. 8), etc. Other interface signals may be similar (e.g. command and address latch enables, read and write enables, write protect, and power/ground connections).

[0049] In general, the stacked configuration shown in FIG. 3 may be used in any of the embodiments described herein.

Transparently Enabling Higher Capacity

[0050] In several of the embodiments that have been described above the flash interface circuit is used to simulate to the system the appearance of a first one (or more) flash memories from a second one (or more) flash memories that are connected to the flash interface circuit. The first one or more flash memories are said to be virtual. The second one or more flash memories are said to be physical. In such embodiments at least one aspect of the virtual flash memory may be different from the physical memory.

[0051] Typically, a flash memory controller obtains certain parameters, metrics, and other such similar information from the flash memory. Such information may include, for example, the capacity of the flash memory. Other examples of such parameters may include type of flash memory, vendor identification, model identification, modes of operation, system interface information, flash geometry information, timing parameters, voltage parameters, or other parameters that may be defined, for example, by the Common Flash Interface (CFI), available at the INTEL website, or other standard or non-standard flash interfaces. In several of the embodiments described, the flash interface circuit may translate between parameters of the virtual and physical devices. For example, the flash interface circuit may be connected to one or more physical flash memory devices of a first capacity. The flash interface circuit acts to simulate a virtual flash memory of a second capacity. The flash interface circuit may be capable of querying the attached one or more physical flash memories to obtain parameters, for example their capacities. The flash interface circuit may then compute the sum capacity of the attached flash memories and present a total capacity (which may or may not be the same as the sum capacity) in an appropriate form to the system. The flash interface circuit may contain logic circuits that translate requests from the system to requests and signals that may be directed to the one or more flash memories attached to flash interface circuit.

[0052] In another embodiment, the flash interface circuit transparently presents a higher capacity memory to the system. FIG. 3 shows a top view of a portion of one embodiment of a stacked package assembly 300. In the embodiment shown in FIG. 3, stacking the flash memory devices on top of a flash interface circuit results in a package with a very small volume. Various embodiments may be tested and burned in before assembly. The package may be manufactured using existing assembly infrastructure, tested in advance of stack assembly and require significantly less raw material, in some embodiments. Other embodiments may include a radial configuration, rather than a stack, or any other desired assembly.

[0053] In the embodiment shown in FIG. 3, the electrical connections between flash memory devices and the flash interface circuit are generally around the edge of the physical perimeter of the devices. In alternative embodiments the connections may be made through the devices, using through-wafer interconnect (TWI), for example. Other mechanisms for electrical connections are easily envisioned,

Integrated Flash Interface Circuit with One or More Flash Devices

[0054] In another embodiment, the flash interface circuit may be integrated with one or more flash devices onto a single monolithic semiconductor die. FIG. 4 shows a view of a die 400 including one or more flash memory circuits 410 and one or more flash interface circuits 420.

Flash Interface Circuit with Configuration and Translation

[0055] In the embodiment shown in FIG. 5, flash interface circuit 500 includes an electrical interface to the host system 501, an electrical interface to the flash memory device(s) 502, configuration logic 503, a configuration block 504, a read-only memory (ROM) block 505, a flash discovery block 506, discovery logic 507, an address translation unit 508, and a unit for translations other than address translations 509. The electrical interface to the flash memory devices(s) 502 is coupled to the address translation unit 508, the other translations unit 509, and the L signals to the flash memory devices (e.g. as illustrated in FIG. 1). That is, the electrical interface 502 comprises the circuitry to drive and/or receive signals to/from the flash memory devices. The electrical interface to the host system 501 is coupled to the other translations unit 509, the address translation unit 508, and the signals to the host interface (S in FIG. 5). That is, the electrical interface 501 comprises the circuitry to drive and/or receive signals to/from the host system. The discovery logic 507 is coupled to the configuration logic 503, and one or both of logic 507 and 503 is coupled to the other translations unit 509 and the address translation unit 508. The flash discovery block 506 is coupled to the discovery logic 507, and the configuration block 504 and the ROM block 505 are coupled to the configuration logic 503. Generally, the logic 503 and 507 and the translation units 508 and 509 may be implemented in any desired fashion (combinatorial logic circuitry, pipelined circuitry, processor-based software, state machines, various other circuitry, and/or any combination of the foregoing). The blocks 504, 506, and 508 may comprise any storage circuitry (e.g. register files, random access memory, etc.).

[0056] The translation units 508 and 509 may translate host flash memory access and configuration requests into requests to one or more flash memory devices, and may translate flash memory replies to host system replies if needed. That is, the translation units 508 and 509 may be configured to modify requests provided from the host system based on differences between the virtual configuration presented by the interface circuit 500 to the host system and the physical configuration of the flash memory devices, as determined by the discovery logic 507 and/or the configuration logic 503 and stored in the configuration block 504 and/or the discovery block 506. The configuration block 504, the ROM block 505, and/or the flash discovery block 506 may store data identifying the physical and virtual configurations.

[0057] There are many techniques for determining the physical configuration, and various embodiments may implement one or more of the techniques. For example, configuration using a discovery process implemented by the discovery logic 507 is one technique. In one embodiment, the discovery (or auto configuration) technique may be selected using an auto configuration signal mentioned previously (e.g. strapping the signal to an active level, either high or low). Fixed configuration information may be programmed into the ROM block 505, in another technique. The selection of this technique may be implemented by strapping the auto configuration signal to an inactive level.

[0058] In one implementation, the configuration block (CB) 504 stores the virtual configuration. The configuration may be set during the discovery process, or may be loaded from ROM block 505. Thus, the ROM block 505 may store configuration data for the flash memory devices and/or configuration data for the virtual configuration.

[0059] The flash discovery block (FB) 306 may store configuration data discovered from attached flash memory devices. In one embodiment, if some information is not discoverable from attached flash memory devices, that information may be copied from ROM block 505.

[0060] The configuration block 504, the ROM block 505, and the discovery block 506 may store configuration data in any desired format and may include any desired configuration data, in various embodiments. Exemplary configurations of the configuration block 504, the ROM block 505, and the discovery block 506 are illustrated in FIGS. 7, 8, and 9, respectively.

[0061] FIG. 7 is a table 700 illustrating one embodiment of configuration data stored in one embodiment of a configuration block 504. The configuration block 504 may comprise one or more instances of the configuration data in table 700 for various attached flash devices and for the virtual configuration. In the embodiment of FIG. 7, the configuration data comprises 8 bytes of attributes, labeled 0 to 7 in FIG. 7 and having various bit fields as shown in FIG. 7.

[0062] Byte zero includes a auto discover bit (AUTO), indicating whether or not auto discovery is used to identify the configuration data; an ONFI bit indicating if ONFI is supported; and a chips field (CHIPS) indicating how many chip selects are exposed (automatic, 1, 2, or 4 in this embodiment, although other variations are contemplated). Byte one is a code indicate the manufacturer (maker) of the device (or the maker reported to the host); and byte two is a device code identifying the particular device from that manufacturer.

[0063] Byte three includes a chip number field (CIPN) indicating the number of chips that are internal to flash memory system (e.g. stacked with the flash interface circuit or integrated on the same substrate as the interface circuit, in some embodiments). Byte three also includes a cell field (CELL) identifying the cell type, for embodiments that support multilevel cells. The simultaneously programmed field (SIMP) indicates the number of simultaneously programmed pages for the flash memory system. The interleave bit (INTRL) indicates whether or not chip interleave is supported, and the cache bit (CACHE) indicates whether or not caching is supported.

[0064] Byte four includes a page size field (PAGE), a redundancy size bit (RSIZE) indicating the amount of redundancy supported (e.g. 8 or 16 bytes of redundancy per 512 bytes, in this embodiment), bits (SMIN) indicating minimum timings for serial access, a block size field (BSIZE) indicating the block size, and an organization byte (ORG) indicating the data width organization (e.g. x8 or x16, in this embodiment, although other widths are contemplated). Byte five includes plane number and plane size fields (PLANE and PLSIZE). Some fields and bytes are reserved for future expansion.

[0065] It is noted that, while various bits are described above, multibit fields may also be used (e.g. to support additional variations for the described attribute). Similarly, a multibit field may be implemented as a single bit if fewer variations are supported for the corresponding attribute.

[0066] FIG. 8 is a table 800 of one embodiment of configuration data stored in the ROM block 505. The ROM block 505 may comprise one or more instances of the configuration data in table 800 for various attached flash devices and for the configuration presented to the host system. The configuration data, this embodiment, is a subset of the data stored in the configuration block. That is, bytes one to five are included. Byte 0 may be determined through discovery, and bytes 6 and 7 are reserved and therefore not needed in the ROM block 505 for this embodiment.

[0067] FIG. 9 is a table 900 of one embodiment of configuration data that may be stored in the discovery block 506. The discovery block 506 may comprise one or more instances of the configuration data in table 900 for various attached flash devices. The configuration data, this embodiment, is a subset of the data stored in the configuration block. That is, bytes zero to five are included (except for the AUTO bit, which is implied as a one in this case). Bytes 6 and 7 are reserved and therefore not needed in the discovery block 506 for this embodiment.

[0068] In one implementation, the discovery information is discovered using one or more read operations to the attached flash memory devices, initiated by the discovery logic 507. For example, a read cycle may be used to test if ONFI is enabled for one or more of the attached devices. The test results may be recorded in the ONFI bit of the discovery block. Another read cycle or cycles may test for the number of flash chips; and the result may be recorded in the CHIPS field. Remaining attributes may be discovered by reading the ID definition table in the attached devices. In one embodiment the attached flash chips may have the same attributes. Alternatively, multiple instances of the configuration data may be stored in the discovery block 506 and various attached flash memory devices may have differing attributes.

[0069] As mentioned above, the address translation unit 508 may translate addresses between the host and the flash memory devices. In one embodiment, the minimum page size is 1 kilobyte (KB). In another embodiment the page size is 8 KB. In yet another embodiment the page size is 2 KB. Generally, the address bits may be transmitted to the flash interface circuit over several transfers (e.g. 5 transfers, in one embodiment). In a five transfer embodiment, the first two transfers comprise the address bits for the column address, low order address bits first (e.g. 11 bits for a 1 KB page up to 14 bits for an 8 KB page). The last three transfers comprise the row address, low order bits first.

[0070] In one implementation, an internal address format for the flash interface circuit comprises a valid bit indicating whether or not a request is being transmitted; a device field identifying the addressed flash memory device; a plane field identifying a plane within the device, a block field identifying the block number within the plane; a page number identifying a page within the block; a redundant bit indicating whether or not the redundant area is being addressed, and column address field containing the column address.

[0071] In one embodiment, a host address is translated to the internal address format according the following rules (where CB_[label] corresponds to fields in FIG. 7):

[0072]

COL[7:0] = Cycle[1][7:0];
COL[12:8] = Cycle[2][4:0];
R = CB_PAGE == 0 ? Cycle[2][2]
: CB _PAGE == 1 ?	Cycle[2][3]
: CB _PAGE == 2 ?	Cycle[2][4]
:	Cycle[2][5];
// block 64,128,256,512K / page 1,2,4,8K
PW[2:0] = CB_BSIZE == 0 && CB_PAGE == 0 ?	6-6	//	0
: CB_BSIZE == 0 && CB_PAGE == 1 ?	5-6	//	−1
: CB_BSIZE == 0 && CB_PAGE == 2 ?	4-6	//	−2
: CB_BSIZE == 0 && CB_PAGE == 3 ?	3-6	//	−3
: CB_BSIZE == 1 && CB_PAGE == 0 ?	7-6	//	1
: CB_BSIZE == 1 && CB_PAGE == 1 ?	6-6	//	0
: CB_BSIZE == 1 && CB_PAGE == 2 ?	5-6	//	−1
: CB_BSIZE == 1 && CB_PAGE == 3 ?	4-6	//	−2
: CB_BSIZE == 2 && CB_PAGE == 0 ?	8-6	//	2
: CB_BSIZE == 2 && CB_PAGE == 1 ?	7-6	//	1
: CB_BSIZE == 2 && CB_PAGE == 2 ?	6-6	//	0
: CB_BSIZE == 2 && CB_PAGE == 3 ?	5-6	//	−1
: CB_BSIZE == 3 && CB_PAGE == 0 ?	9-6	//	3
: CB_BSIZE == 3 && CB_PAGE == 1 ?	8-6	//	2
: CB_BSIZE == 3 && CB_PAGE == 2 ?	7-6	//	1
:	6-6;	//	0
PW[2:0] = CB_BSIZE - CB_PAGE;		//	same
			as
			above
PAGE = PW == −3 ? {5′b0,     Cycle[3][2:0]}
: PW == −2 ?	{4′b0,	Cycle[3][3:0]}
: PW == −1 ?	{3′b0,	Cycle[3][4:0]}
: PW ==   0 ?	{2′b0,	Cycle[3][5:0]}
: PW ==   1 ?	{1′b0,	Cycle[3][6:0]}
: PW ==   2 ?	{	Cycle[3][7:0]}
:	{Cycle[4][0],	Cycle[3][7:0]};
BLOCK = PW == −3 ? {    Cycle[5], Cycle[4], Cycle[3][7:3]}
: PW == −2 ?	{1′b0, Cycle[5], Cycle[4], Cycle[3][7:4]}
: PW == −1 ?	{2′b0, Cycle[5], Cycle[4], Cycle[3][7:5]}
: PW ==   0 ?	{3′b0, Cycle[5], Cycle[4], Cycle[3][7:6]}
: PW ==   1 ?	{4′b0, Cycle[5], Cycle[4], Cycle[3][7:7]}
: PW ==   2 ?	{5′b0, Cycle[5], Cycle[4]}
:	{6′b0, Cycle[5], Cycle[4][7:1]};
// CB_PLSIZE 64Mb = 0 .. 8Gb = 7 or 8MB .. 1GB
PB[3:0] = CB_PLSIZE - CB_PAGE; // PLANE_SIZE / PAGE_SIZE
PLANE = PB == −3 ? {10′b0, BLOCK[20:11]}
: PB == −2 ?	{ 9′b0,	BLOCK[20:10]}
: PB == −1 ?	{ 8′b0,	BLOCK[20: 9]}
: PB ==   0 ?	{ 7′b0,	BLOCK[20: 8]}
: PB ==   1 ?	{ 6′b0,	BLOCK[20: 7]}
: PB ==   2 ?	{ 5′b0,	BLOCK[20: 6]}
: PB ==   3 ?	{ 4′b0,	BLOCK[20: 5]}
: PB ==   4 ?	{ 3′b0,	BLOCK[20: 4]}
: PB ==   5 ?	{ 2′b0,	BLOCK[20: 3]}
: PB ==   6 ?	{ 1′b0,	BLOCK[20: 2]}
:	{	BLOCK[20: 1]};
DEV = CE1_ == 1′b0 ? 2′d 0;
: CE2_ == 1′b0 ? 2′d 1
: CE3_ == 1′b0 ? 2′d 2
: CE4_ == 1′b0 ? 2′d 3
: 2′d 0;

[0073] Similarly, the translation from the internal address format to an address to be transmitted to the attached flash devices may be performed according to the following rules (where CB_[label] corresponds to fields in FIG. 9):

[0074]

	Cycle[1][7:0] = COL[7:0];
	Cycle[2][7:0] = FB_PAGE == 0 ? {5′b0, R, COL[ 9:8]}
	: FB_PAGE == 1 ?	{4′b0, R, COL[10:8]}
	: FB_PAGE == 2 ?	{3′b0, R, COL[11:8]}
	:	{2′b0, R, COL[12:8]};
	Cycle[3][7:0] = PAGE[7:0];
	Cycle[3][0] = PAGE[8];
	BLOCK[ ] = CB_PAGE == 0 ? Cycle [ ][ ] :
	CB_PAGE == 1 ? Cycle [ ][ ] :
	CB_PAGE == 2 ? Cycle [ ][ ] :
		Cycle [ ][ ] : ;
	PLANE = TBD
	FCE1_ = !(DEV == 0 && VALID);
	FCE2_ = !(DEV == 1 && VALID);
	FCE3_ = !(DEV == 2 && VALID);
	FCE4_ = !(DEV == 3 && VALID);
	FCE5_ = !(DEV == 4 && VALID);
	FCE6_ = !(DEV == 5 && VALID);
	FCE7_ = !(DEV == 6 && VALID);
	FCE8_ = !(DEV == 7 && VALID);

[0075] Other translations that may be performed by the other translations unit 509 may include a test to ensure that the amount of configured memory reported to the host is the same as or less than the amount of physically-attached memory. Addition, if the configured page size reported to the host is different than the discovered page size in the attached devices, a translation may be performed by the other translations unit 509. For example, if the configured page size is larger than the discovered page size, the memory request may be performed to multiple flash memory devices to form a page of the configured size. If the configured page size is larger than the discovered page size multiplied by the number of flash memory devices, the request may be performed as multiple operations to multiple pages on each device to form a page of the configured size. Similarly, if the redundant area size differs between the configured size reported to the host and the attached flash devices, the other translation unit 509 may concatenate two blocks and their redundant areas. If the organization reported to the host is narrower than the organization of the attached devices, the translation unit 509 may select a byte or bytes from the data provided by the attached devices to be output as the data for the request.

Presentation Translation

[0076] In the embodiment of FIG. 6, some or all signals of a multi-level cell (MLC) flash device 603 pass through a flash interface circuit 602 disposed between the MLC flash device and the system 601. In this embodiment, the flash interface circuit presents to the system as a single level cell (SLC)-type flash memory device. Specifically, the values representative of an SLC-type flash memory device appear coded into a configuration block that is presented to the system. In the illustrated embodiment, some MLC signals are presented to the system 601. In other embodiments, all MLC signals are received by the flash interface circuit 602 and are converted to SLC signals for interface to the system 601.

Power Supply

[0077] In some of the embodiments described above it is necessary to electrically connect one of more flash memory chips and one of more flash interface circuits to a system. These components may or may not be capable of operating from the same supply voltage. If, for example, the supply voltages of portion(s) the flash memory and portions(s) flash interface circuit are different, there are many techniques for either translating the supply voltage and/or translating the logic levels of the interconnecting signals. For example, since the supply currents required for portion(s) (e.g. core logic circuits, etc.) of the flash memory and/or portion(s) (e.g. core logic circuits, etc.) of the flash interface circuit may be relatively low (e.g. of the order of several milliamperes, etc.), a resistor (used as a voltage conversion resistor) may be used to translate between a higher voltage supply level and a lower logic supply level. Alternatively, a switching voltage regulator may be used to translate supply voltage levels. In other embodiments it may be possible to use different features of the integrated circuit process to enable or eliminate voltage and level translation. Thus for example, in one technique it may be possible to employ the I/O transistors as logic transistors, thus eliminating the need for voltage translation. In a similar fashion because the speed requirement for the flash interface circuit are relatively low (e.g. currently of the order of several tens of megaHertz, etc.) a relatively older process technology (e.g. currently 0.25 micron, 0.35 micron, etc) may be employed for the flash interface circuit compared to the technology of the flash memory (e.g. 70 nm, 110 nm, etc.). Or in another embodiment a process that provides transistors that are capable of operating at multiple supply voltages may be employed.

[0078] FIG. 10 is a flowchart illustrating one embodiment of a method of emulating one or more virtual flash memory devices using one or more physical flash memory devices having at least one differing attribute. The method may be implemented, e.g., in the flash interface circuit embodiments described herein.

[0079] After power up, the flash interface circuit may wait for the host system to attempt flash discovery (decision block 1001). When flash discovery is requested from the host (decision block 1001, “yes” leg), the flash interface circuit may perform device discovery/configuration for the physical flash memory devices coupled to the flash interface circuit (block 1002). Alternatively, the flash interface circuit may configure the physical flash memory devices before receiving the host discovery request. The flash interface circuit may determine the virtual configuration based on the discovered flash memory devices and/or other data (e.g. ROM data) (block 1003). The flash interface circuit may report the virtual configuration to the host (block 1004), thus exposing the virtual configuration to the host rather than the physical configuration.

[0080] For each host access (decision block 1005), the flash interface circuit may translate the request into one or more physical flash memory device accesses (block 1006), emulate attributes of the virtual configuration that differ from the physical flash memory devices (block 1007), and return an appropriate response to the request to the host (block 1008).

[0081] The above description, at various points, refers to a flash memory controller. The flash memory controller may be part of the host system, in one embodiment (e.g. the flash memory controller 108 shown in FIG. 1). That is, the flash interface circuit may be between the flash memory controller and the flash memory devices (although some signals may be directly coupled between the system and the flash memory devices, e.g. as shown in FIG. 1). For example, certain small processors for embedded applications may include a flash memory interface. Alternatively, larger systems may include a flash memory interface in a chipset, such as in a bus bridge or other bridge device.

[0082] In various contemplated embodiments, an interface circuit may be configured to couple to one or more flash memory devices and may be further configured to couple to a host system. The interface circuit is configured to present at least one virtual flash memory device to the host system, and the interface circuit is configured to implement the virtual flash memory device using the one or more flash memory devices to which the interface circuit is coupled. In one embodiment, the virtual flash memory device differs from the one or more flash memory devices in at least one aspect (or attribute). In one embodiment, the interface circuit is configured to translate a protocol implemented by the host system to a protocol implemented by the one or more flash memory devices, and the interface circuit may further be configured to translate the protocol implemented by the one or more flash memory devices to the protocol implemented by the host system. Either protocol may be a NAND protocol or a NOR protocol, in some embodiments. In one embodiment, the virtual flash memory device is pin-compatible with a standard pin interface and the one or more flash memories are not pin-compatible with the standard pin interface. In one embodiment, the interface circuit further comprises at least one error detection circuit configured to detect errors in data from the one or more flash memory devices. The interface circuit may still further comprise at least one error correction circuit configured to correct a detected error prior to forwarding the data to the host system. In an embodiment, the interface circuit is configured to implement wear leveling operations in the one or more flash memory devices. In an embodiment, the interface circuit comprises a prefetch circuit configured to generate one or more prefetch operations to read data from the one or more flash memory devices. In one embodiment, the virtual flash memory device comprises a data bus having a width equal to N times a width of a data bus of any one of the one or more flash devices, wherein N is an integer greater than one. In one embodiment, the interface circuit is configured to interleave data on the buses of the one or more flash memory devices to implement the data bus of the virtual flash memory device. In another embodiment, the interface circuit is configured to operate the data buses of the one or more flash memory devices in parallel to implement the data bus of the virtual flash memory device. In an embodiment, the virtual flash memory device has a bandwidth that exceeds a bandwidth of the one or more flash memory devices. In one embodiment, the virtual flash memory device has a latency that is less than the latency of the one or more flash memory devices. In an embodiment, the flash memory device is a multi-level cell (MLC) flash device, and the virtual flash memory device presented to the host system is a single-level cell (SLC) flash device.

[0083] Numerous variations and modifications will become apparent to those skilled in the art once the above disclosure is fully appreciated. It is intended that the following claims be interpreted to embrace all such variations and modifications.