Abstract: Techniques are disclosed relating to enabling virtual machines to access data on a physical recording medium. In one embodiment, a computing system provides a logical address space for a storage device to an allocation agent that is executable to allocate the logical address space to a plurality of virtual machines having access to the storage device. In such an embodiment, the logical address space is larger than a physical address space of the storage device. The computing system may then process a storage request from one of the plurality of virtual machines. In some embodiments, the allocation agent is a hypervisor executing on the computing system. In some embodiments, the computing system tracks utilizations of the storage device by the plurality of virtual machines, and based on the utilizations, enforces a quality of service level associated with one or more of the plurality of virtual machines.

Abstract: A variable compensation pass bias based on a state being sensed in non-volatile memory based is provided. Shifts in the apparent charge stored by a memory cell can occur because of coupling based on charge stored by adjacent cells. To account for the shift, compensations can be applied to an adjacent word line when reading based on the different possible conditions of an adjacent cell. The effects of coupling may be more pronounced for memory cells in lower states corresponding to lower threshold voltages. A compensation pass bias can be reduced as the state being sensed at a selected word line increases to account for the different effects. A compensation pass bias for an adjacent word line may be reduced with the application of larger read reference voltages to a selected word line. Other variations to a compensation pass bias are provided.

Abstract: A controller of a non-volatile memory system may be configured to identify bits of data to be stored in memory elements of non-volatile memory that are identified as unreliable. The controller may be configured to bias at least some of these bits to a predetermined logic value at which the bits are likely to be read from the unreliable memory elements. The controller may do so using a biasing key that the controller generates based on identification of the bits. Subsequently, when the data is read, the controller may assign log likelihood ratio values for the bits to correspond to a percent likelihood of the bits being biased to the predetermined logic value. The bits may also be unbiased using the biasing key.

Abstract: Sense circuits in a memory device can be pre-charged to different levels in a sensing process to reduce the amount of time used for sensing. During sensing of first and second memory cells, a control circuit pre-charges first and second sense circuits to first and second voltages, respectively. The first and second sense circuits are associated with the first and second memory cells, respectively. Also, during the sensing, a control gate voltage is applied to the first and second memory cells. The control circuit allows the first and second sense node voltages to discharge in a common discharge period and the cells are sensed using a common trip condition. The first and second memory cells are therefore subject to different concurrent verify tests.

Abstract: A storage system and method for adaptive thermal throttling are disclosed. In one embodiment, a method for adaptive thermal throttling is provided that is performed in a storage system having a memory. This method comprises determining if a temperature of the storage system is above a threshold temperature; and in response to determining that the temperature of the storage system is above the threshold temperature: reducing performance of the storage system in an iterative manner until the temperature of the storage system is within a temperature envelope around the threshold temperature; and storing, in the memory, a value indicating what the performance of the storage system was reduced to in order to get the temperature within the temperature envelope. Other embodiments are possible, and each of the embodiments can be used alone or together in combination.

Abstract: Methods and systems are provided where a memory controller for non-volatile memory transfers data to and from random access memory over a second double data rate bus, and a host system access the random access memory over a first double data rate bus. The memory controller may transfer the data in response to a command received by the memory controller from a host system via a block storage protocol bus. Alternatively or in addition, the memory controller may transfer the data as part of caching data internal to the non-volatile memory.

Abstract: Systems and methods for immediate physical erasure of data in a memory system in response to a user command are disclosed. In one implementation, a memory system includes a non-volatile memory and a controller in communication with the non-volatile memory. The controller comprises a processor that is configured to receive from a host in communication with the memory system, a destruct command that indicates a user request to make the memory system inoperable. The processor is further configured to perform one or more operations to render the memory system inoperable in response to the destruct command received from the host.

Abstract: Techniques are provided for improving the accuracy of read operations of memory cells, where the threshold voltage (Vth) of a memory cell can shift depending on when the read operation occurs. Countermeasures are provided for a first read situation in which a memory is read after a power on event or after a long delay since a last read. Read voltages of lower or higher programmed data states are set according to a positive or negative temperature coefficient (Tco), respectively. Read voltages for error recovery can be set similarly. In another aspect, a wait period between a dummy voltage and a read voltage is a function of temperature. In another aspect, word line voltages of unselected blocks are set according to a negative Tco. In another aspect, pass voltages are set based on a Tco for each programmed data state.

Abstract: Techniques for forming 3D memory arrays are disclosed. Memory openings are filled with a sacrificial material, such as silicon or nitride. Afterwards, a replacement technique is used to remove nitride from an ONON stack and replace it with a conductive material such as tungsten. Afterwards, memory cell films are formed in the memory openings. The conductive material serves as control gates of the memory cells. The control gate will not suffer from corner rounding. ONON shrinkage is avoided, which will prevent control gate shrinkage. Block oxide between the charge storage region and control gate may be deposited after control gate replacement, so the uniformity is good. Block oxide may be deposited after control gate replacement, so TiN adjacent to control gates can be thicker to prevent fluorine attacking the insulator between adjacent control gates. Therefore, control gate to control gate shorting is prevented.

Abstract: A non-volatile memory system includes a plurality of groups of connected non-volatile memory cells (e.g., charge trapping memory cells), a select line, and a plurality of select gates connected to the select line. Each select gate is connected at an end (e.g. source end or drain side) of one of the groups of memory cells. The system includes one or more control circuits that are configured to determine whether the select gates are abnormal. If a select gate is determined to be abnormal, then one of the memory cells connected to the select gate is converted to operate as a select gate. The system will then perform memory operations by operating the converted memory cell as a select gate.

Abstract: Apparatuses, systems, and methods are disclosed for managing multiple sets of metadata. A method includes maintaining a first set of metadata on a volatile recording medium and a second set of metadata on a non-volatile recording medium. The first and second sets of metadata are associated with one or more logical addresses for data stored on the non-volatile recording medium. The first and second sets of metadata relate to a state of the data. A method includes updating the second set of metadata in response to a first operation performed on the data. The second set may be updated based on the first operation. A method includes updating the first set of metadata in response to a subsequent operation performed on the data. The first set may be updated based on the first operation.

Abstract: A memory system or flash card may include an algorithm for identifying and accounting for the rewrite frequency of data to be written to the card. The file system partition or file type of data may be used for monitoring rewrite frequency and predicting future rewrites. A learning algorithm that monitors rewrites may be implemented in firmware for accurate and dynamic identification of file types/partitions with the most likely rewrites. The identification of rewrites may be used to sort the data into groups (e.g. hot data=likely rewritten, and cold data=not likely to be rewritten). The hot data may stay in single level cell (SLC) update blocks longer, while the cold data can be moved to MLC blocks sooner.

Abstract: In an integrated-circuit memory, performance is increased by reducing an electrical contact resistance between a metal layer and an upper poly layer (a control gate poly). The electrical contact resistance is reduced by increasing the thickness of a silicide layer between the metal layer and the upper poly layer. The memory has a memory cell region and a non-memory cell region. The thickness of the silicide layer is typically restricted by consideration of integrated-circuit fabrication geometry for each memory cell not to exceed a predetermined aspect ratio. The present implementation allows independent optimization of the thickness of silicide layer in the memory cells region and the non-memory cell region. In particular, in the non-memory cell region, a thicker silicide layer significantly improves the contact resistance of a slit contact for components having the upper poly layer in contact with a lower poly layer (a floating gate poly).

Abstract: A method of forming a shallow trench isolation trench in a semiconductor substrate is described. The method includes forming a trench in a region of the substrate, forming a liner in the trench, wherein the liner includes a first dielectric material, adhering a halogen element to the liner, forming a second dielectric material in the trench, annealing the first dielectric material and the second dielectric material, exposing a portion of a surface of the second dielectric material, and isotropically etching the exposed portion of the surface of the second dielectric material to form an air gap in the shallow trench isolation trench.

Abstract: Three-dimensional (3D) non-volatile memory arrays having a vertically-oriented thin film transistor (TFT) select device and method of fabricating such a memory are described. The vertically-oriented TFT may be used as a vertical bit line selection device to couple a global bit line to a vertical bit line. A select device pillar includes a body and upper and lower source/drain regions. At least one gate is separated horizontally from the select device pillar by a gate dielectric. Beneath each gate, a single gap fill dielectric layer extends vertically from a lower surface of the gate, at least partially separating the gate from the underlying global bit line. Between horizontally adjacent pillars, this same dielectric layer extends from its same lower level beneath the gates vertically to a level of the upper source/drain region.

Abstract: Techniques are provided for improving the accuracy of read operations of memory cells, where the threshold voltage of a memory cell can shift depending on when the read operation occurs. In one aspect, read voltages are set and optimized based on a time period since a last sensing operation. A timing device such as an n-bit digital counter may be provided for each block of memory cells to track the time. The counter is set to all 1's when the device is powered on. When a sensing operation occurs, the counter is periodically incremented based on a clock. When a next read operation occurs, the value of the counter is cross-referenced to an optimal set of read voltage shifts. Each block of cells may have its own counter, where the counters are incremented using a local or global clock.

Abstract: A dynamic bus inversion (DBI) circuit disposed between a transmitter and a receiver for generating an inversion control signal that is communicated to the receiver and used to perform inversion control on data communicated along a data path between the transmitter and the receiver includes a delay data setup circuit to receive the data from the transmitter. A majority vote function circuit is used to perform majority voting for consecutive bits of data output by the delay data setup circuit to generate majority data output. An inversion control circuit receives the majority data output, retrieves feedback data from a preceding inversion control output and interprets the two data to generate inversion control signal, which is used to perform inversion control on data along the data path before being communicated to the receiver. The inversion control signal is used by the receiver to interpret the data received from the data path.

Abstract: A content addressable memory cell is provided that includes plurality of transistors having a minimum feature size F, and a plurality of memory elements coupled to the plurality of transistors. The content addressable memory cell occupies an area of between 18F2 and 36F2.

Abstract: To prevent data loss due to latent defects, a non-volatile memory system will use a leakage detection circuit to test for small amounts of leakage that indicate that the memory is susceptible to failure.

Abstract: Body bias can be applied to optimize performance in a non-volatile storage system. Body bias can be set in an adaptive manner to reduce an error count of an error correcting and/or detecting code when reading data from non-volatile storage elements. Also, a body bias level can be increased or decreased as a number of programming cycles increases. Also, body bias levels can be set and applied separately for a chip, plane, block and/or page. A body bias can be applied to a first set of NAND strings for which operations are being performed by controlling a first voltage provided to a source side of the first set of NAND strings and a second voltage provided to a p-well. A source side of a second set of NAND strings for which operations are not being performed is floated or receives a fixed voltage.