3679 results about "Memory controller" patented technology

The invention discloses a memory fault repairing method and device, equipment, a medium and a computer program product, and relates to the technical field of computers.The memory fault repairing method includes the steps that memory error information sent by a memory controller is obtained, a fault target memory page can be positioned, and a standby memory page is obtained from a standby memory pool; the memory address mapping table is updated, the physical address of the target memory page with the fault is mapped to the physical address of the standby memory page, the repairing process does not depend on triggering of a system management interrupt mechanism, memory repairing in the system running stage is achieved, the system downtime caused by memory fault repairing is shortened, and therefore the memory repairing efficiency is improved. The problem of high delay in memory fault processing can be solved, and the technical effect of improving the memory fault repairing efficiency is achieved.

The invention discloses an accelerator for accelerating the reasoning process of an unstructured sparse large language model and an operation method of the accelerator. The accelerator comprises a processor module, a direct memory access module, a memory controller module, a DDR memory module, an interconnection bus module, a calculation core module and a router module. The interconnection bus module is connected with the direct memory access module, the memory controller module and the computing core module; the processor module is connected with the direct memory access module; the DDR memory module is connected with the memory controller module; the router module is connected with the computing core module and the adjacent router module; and the calculation core module obtains to-be-calculated data from the off-chip DDR memory module through the interconnection bus module. According to the method, the limit that data reuse is carried out only along a single row or a single column on a traditional systolic array is broken through, cross-row and cross-column data reuse is achieved, the pressure of hardware equipment is relieved, the positions of non-zero elements are coded through bit masks, and the calculation efficiency is improved.

This disclosure provides techniques for managing memory which match per-data metrics to those of other data or to memory destination. In one embodiment, wear data is tracked for at least one tier of nonvolatile memory (e.g., flash memory) and a measure of data persistence (e.g., age, write frequency, etc.) is generated or tracked for each data item. Memory wear management based on these individually-generated or tracked metrics is enhanced by storing or migrating data in a manner where persistent data is stored in relatively worn memory locations (e.g., relatively more-worn flash memory) while temporary data is stored in memory that is less worn or is less susceptible to wear. Other data placement or migration techniques are also disclosed.

Systems, methods, and apparatus related to memory devices that perform matrix vector multiplication using memory cells. In one approach, a memory cell array has memory cells used to perform matrix vector multiplication based on summing output currents from the memory cells. A context of memory cells is determined by a controller (e.g., a memory controller internal or external to a memory chip having the array). The context can include, for example, a physical location of memory cells, weight patterns being programmed, and / or neighboring cell interference, etc. Based on the determined context, the controller dynamically determines adjustments (e.g., adjusted target threshold voltages or currents) for programming the memory cells to store weights prior to performing the matrix vector multiplication.

A memory system, an operation method thereof and a readable storage medium are provided according to examples of this disclosure. The memory system including: a memory device including a plurality of memory blocks, each of which including a plurality of word lines and a plurality of memory cells coupled to the plurality of word lines; a memory controller coupled to the memory device and configured to: acquire a read retry voltage with both a first-type read retry table and a second-type read retry table; control the memory device to perform a read retry operation with the read retry voltage.

Secure communication provided with secure and non-secure root ports. One embodiment comprises: a plurality of cores; a memory controller to couple to a memory; an interconnect fabric coupled to the plurality of cores and the memory controller; and a root complex to support end-to-end encrypted channels between devices, the root complex comprising: a root port to receive non-posted requests from a requestor device, the root port to associate a first tag value with a non-posted request to indicate whether the non-posted request is received over an end-to-end encrypted channel; and a bridge device to transmit the non-posted request with the first tag value and to subsequently receive a completion message including the first tag value, wherein the root port is to determine whether the completion message is to be encrypted in accordance with the end-to-end encrypted channel based on the first tag value.

The invention provides a modularized integrated circuit unit and a distributed storage unit access method and system, and the modularized integrated circuit unit comprises at least one memory controller which sends a storage address access request to a storage unit; the at least one storage unit provides unified memory access for the storage address access request, and the storage unit comprises a local storage unit and a virtual far-end storage unit; the at least one computing unit initiates a storage address access request; and the at least one internet-on-chip is used for interleaving the storage address indicated by the storage address access request, generating interleaved storage address distribution, determining a corresponding target storage unit, and storing the target storage unit by setting a local storage unit and a virtual far-end storage unit. Uniform address access and load balancing of the distributed storage units on the modularized integrated circuit units are achieved through the address interleaving technology and the virtual storage mapping, the resource utilization rate is increased, and the performance and usability of the distributed storage system are improved.

A memory system includes a plurality of memory cells each storing a first bit and a second bit and a control circuit. The control circuit is reads out first data, first partial data, and second partial data, each corresponding to the first bit, from the plurality of memory cells, read out second data, third partial data, and fourth partial data, each corresponding to the second bit, from the plurality of memory cells, generate first compressed data based on an OR operation of the first partial data and the third partial data, generate second compressed data based on an OR operation of the second partial data and the fourth partial data, and transmit the first data, the second data, the first compressed data, and the second compressed data to an external memory controller.

Systems, apparatuses, and methods related to security management for a ferroelectric memory device are described. An example method can include receiving, at a memory controller and from a host, a command and firmware data. The memory controller can manage a non-volatile memory device, such as a ferroelectric memory device, and the host and the memory controller can communicate using a compute express link (CXL) protocol. The command can be executed to update firmware stored on the non-volatile memory device. The method can further include accessing a first public key from the non-volatile memory device. The method can further include validating the first public key with a second public key within the firmware data. The method can further include validating the firmware data. The method can further include verifying a security version of the firmware data. The method can further include updating the non-volatile memory device with the firmware data.

The invention belongs to the technical field of extensible computer architecture, and particularly relates to a two-level cache coherence protocol based on a many-core computing architecture. Comprising a many-core computing architecture, the many-core computing architecture comprises a plurality of computing clusters, each computing cluster is formed by interconnection of a plurality of core grains and an I / O and memory controller, and the I / O and memory controller is interconnected with at least two memory DDRs; each core particle is formed by interconnection of a plurality of computing cores and a shared cache LLC; each calculation core is provided with a private cache, and when the data is not in the private cache, the memory access request continues to go to the next-level shared cache LLC or the memory DDR until the latest data is obtained; the shared cache LLC can access the memory DDR only after the shared cache LLC is in butt joint with the memory controller through high-speed interconnection and I / O; and the two-level cache consistency protocol directory is composed of a first-level directory and a second-level directory. According to the invention, the delay of re-reading from the storage system is omitted, the performance is further improved, and the congestion at the storage position is reduced.

A method for performing data access management of a memory device in predetermined communications architecture with aid of unbalanced table update size and associated apparatus are provided. The method may include: utilizing a memory controller to receive a set of first commands from a host device, receive a set of first data with a first active block according to the set of first commands, and update a temporary physical-to-logical (P2L) address mapping table corresponding to the first active block; determining a selected table update size among multiple predetermined table update sizes according to at least one predetermined rule, wherein the multiple predetermined table update sizes represent multiple table entry counts, respectively; and updating at least one logical-to-physical address mapping table in the NV memory according to a set of P2L table entries corresponding to the selected table update size in the temporary P2L address mapping table, for further data accessing.

The invention relates to the technical field of SOC chips, and discloses an efficient memory management method and system based on an SOC chip, which are used for constructing a complete topological graph and calculating access delay distribution by comprehensively analyzing an internal physical structure of the chip. According to the method, the memory access behavior of each processing core is monitored in real time, dynamic memory partitioning is executed according to the access characteristics and the physical topological structure, and the optimal access area is distributed for the processing core. And when an access hot spot is detected, triggering a data migration mechanism to copy hot spot data to a relatively close memory area. According to the method, a memory address space is recombined by adopting a topology-aware address mapping algorithm, a multi-level cache collaboration mechanism is established, and the working mode of a memory controller is dynamically adjusted according to an application type. Through load balancing monitoring and periodic memory recombination, the memory access delay is effectively reduced, the bandwidth utilization rate is improved, and the overall memory access performance of the SOC chip is optimized.

A buffer chip includes a chip select signal reception circuit for receiving one or more system chip select signals transmitted from a memory controller and a chip ID reception circuit for receiving chip ID information transmitted from the memory controller. The buffer chip also includes a chip select signal generation circuit that generates memory chip select signals by using the one or more system chip select signals and the chip ID information and a chip select signal transmission circuit that transmits the memory chip select signals to a plurality of memory chips.

To provide a memory system and a memory device with high reliability.SOLUTION: A memory system of the embodiment includes a memory device and a memory controller. The memory device includes a plurality of memory cells. The memory controller is configured to determine a first voltage based on information regarding a gradient of a distribution of threshold voltages of each of the plurality of memory cells, and to perform tracking operations. The memory controller causes the memory device to perform a plurality of read operations using a plurality of read levels for the plurality of memory cells in the tracking operation. The memory controller acquires the number of ON cells in each of the plurality of read operations. The memory controller sets a first potential difference of the plurality of read levels in a voltage range lower than the first voltage and a second potential difference of the plurality of read levels in a voltage range higher than the first voltage to be different from each other.SELECTED DRAWING: Figure 12

The present disclosure relates to systems and methods implemented on a memory controller for detecting and mitigating memory attacks (e.g., row hammer attacks). For example, a memory controller may track activations of row addresses within a memory hardware (e.g., a DRAM device) and determine whether a pattern of activations is indicative of a row hammer attack. This is determined using a counting mode for corresponding memory sub-banks. Where a likely row hammer attack is detected, the memory controller may activate a sampling mode (rather than the counting mode) for a particular sub-bank to identify which of the row addresses should be refreshed on the memory hardware. The implementations described herein provide a low computational cost alternative to heavy-handed detection mechanisms that require access to significant computing resources to accurately detect and mitigate row hammer attacks.

A memory system and a method for decoding of one or more codewords. The system has a storage medium having therein a first decode, a first processor associated with the storage medium and configured to XOR results obtained from the first decoder when the one or more codewords in the storage medium are decoded, and a memory controller having therein a second decoder and a buffer, the memory controller configured to XOR results obtained from the second decoder. The second decoder decodes the one or more codewords after failure of the first decoder to decode the one or more codewords.

The invention provides a data writing method and a memory controller, which are suitable for a rewritable nonvolatile memory module with a plurality of memory units. The method comprises the following steps: acquiring original data from a host system; executing a plurality of randomization operations on the original data to obtain a plurality of write-in data; based on a plurality of directions of a three-dimensional circuit architecture of the rewritable nonvolatile memory module, executing randomization verification operation on each piece of write-in data to verify the data uniformity of each piece of write-in data in each direction so as to obtain target write-in data in the plurality of pieces of write-in data, wherein the randomized quality of the target write data is determined to be qualified; and storing the target write data to a plurality of target memory cells among the plurality of memory cells. Through multiple randomization operations and verification processes, the randomness and reliability of data storage are effectively improved, so that the overall performance and data security of a storage system are enhanced.

A memory device is provided. The memory device includes: a buffer memory; a nonvolatile backup memory; and a memory controller configured to: store map data corresponding to an external storage device in the buffer memory; provide, in response to a request from the external storage device, an address pair corresponding to the request, from among address pairs of the map data, to the external storage device; and back up, in response to a sudden power-off event, the map data to the nonvolatile backup memory.

The invention relates to the technical field of chip verification, in particular to a simulation verification system based on a DFI interface, the system comprises a memory controller, a verification component and a memory model, the memory controller is connected with the verification component through the DFI interface, the memory controller is used for transmitting an operation command and first data through the DFI interface, the verification component is connected with the memory model, and the memory model is connected with the verification component. The verification component is used for checking the operation command and the first data, analyzing the operation command to obtain address information and calling the memory model to operate the first data, and the memory model is used for operating the first data according to the received operation command, address information and first data sent by the verification component. It can be known that the verification component and the memory model can be constructed based on the verification language compatible with the verification platform, so that initialization only needs to configure the memory controller through the bus, the initialization time of the simulation verification system is saved, a more flexible backdoor data loading mode is provided, and the simulation verification efficiency is improved.

The invention discloses memory calibration with tolerance check. A memory subsystem includes a memory and a memory controller coupled to the memory. The memory controller includes a calibration circuit configured to: perform a horizontal memory calibration for some of a plurality of performance states during an initialization process; and determining information indicative of a set of differences between calibration results for pairs of performance states of the plurality of performance states and storing the information in a storage circuit. The memory controller is further configured to, after the initialization process and in response to a change from a first performance state of the plurality of performance states to a second performance state of the plurality of performance states, set an initial memory parameter for the second performance state based on the set of differences. Thereafter, operation begins in the second performance state without performing the initial level calibration.

This specification describes memory controllers with adaptive precharge scheduling. In one aspect, a memory controller includes a refresh scheduler configured to send refresh commands to dynamic random-access memory (DRAM) banks of a DRAM memory system that includes DRAM banks arranged in a set of DRAM bank groups each comprising one or more DRAM banks. The memory controller includes an adaptive precharge scheduler configured to determine a priority score for each DRAM bank group based on a set of parameters, select, based on the priority score for each DRAM bank group, a particular DRAM bank group to close so that each DRAM bank in the DRAM bank group can be refreshed by the refresh scheduler, and send the precharge command to at least one DRAM bank of the particular DRAM group.

The invention provides a memory management method and a memory controller, which are applied to a storage device configured with a memory module. The method comprises the following steps: acquiring a continuity state table, identifying a target mapping table according to a data access instruction and the continuity state table, acquiring a continuity state of the target mapping table, judging whether a continuous section mapping table is used or not according to the continuity state, and if so, loading the continuous section mapping table to a buffer memory, and if not, loading the target mapping table to the buffer memory, and obtaining the physical address according to the loaded mapping table to access the memory module. According to the method, the continuity of the entity address in the mapping table is evaluated, and the optimal mapping table processing mode is dynamically selected, so that the data access performance is effectively improved, and the use efficiency of the buffer memory is optimized.

A memory expander is disclosed. The memory expander includes a memory, a memory controller configured to control the memory, a compute express link (CXL) engine configured to acquire a CXL flit from a host device connected to the memory expander and configured to acquire a calculation request for pieces of data stored in the memory by performing conversion on the CXL flit, and a domain-specific accelerator configured to perform a calculation in response to the calculation request.

Methods that may be performed by a host controller and a memory controller of a computing device. The method synchronizes memory tables between the storage device and a host device by modifying an indicator in a first memory table on the storage device in response to a change in a memory mapping, the first memory table mapping logical addresses to physical addresses of memory on the storage device, the indicator identifying one or more address mapping changes of the first memory table, notifying the host device that the first memory table has been modified, and transmitting to the host device at least a portion of the first memory table including the one or more address mapping changes. The storage device processes memory requests from the host device based on one or more addresses affected by the one or more address mapping changes.

A memory apparatus includes a DFE receiver and a DFE reset circuit. The DFE receiver is configured to receive a data signal and a data strobe signal from a memory device. The DFE receiver includes a DFE tap that is determined according to a previous data signal, and the DFE receiver adjusts the data signal according to the DFE tap. The DFE reset circuit is configured to receive a gate enable signal and an internal enable signal from the memory controller, generate a DFE reset signal according to the gate enable signal and the internal enable signal. The DFE reset circuit outputs the DFE reset signal to the DFE receiver to reset the DFE tap of the DFE receiver between read bursts.

Apparatuses and techniques for logging a memory address associated with faulty usage-based disturbance data are described. In an example aspect, a memory device can detect, at a local-bank level, a fault associated with usage-based disturbance data. This detection enables the memory device to log a row address associated with the faulty usage-based disturbance data. To avoid increasing a complexity and / or a size of the memory device, some implementations of the memory device can perform the address logging at the multi-bank level with the assistance of an engine, such as a test engine. The memory device stores the logged address in at least one mode register to communicate the fault to a memory controller. With the logged address, the memory controller can initiate a repair procedure to fix the faulty usage-based disturbance data.

The invention provides a memory management method and a memory controller. The method comprises the following steps: acquiring block use conditions of a plurality of physical blocks of a rewritable nonvolatile memory module and a working state of storage equipment; when the working state is a busy state, whether garbage collection operation is executed or not is judged according to the block use condition and the host write-in requirement; and dynamically adjusting the write-in speed of the write-in operation of the host according to the number of the plurality of residual blocks when executing the garbage collection operation during the write-in operation of the host. According to the method, the use condition of the physical block and the write-in behavior of the host are monitored in real time, and a dynamic write-in speed adjusting mechanism is combined, so that the execution efficiency of garbage collection operation is obviously improved, and the write-in performance of the storage equipment under the high-load condition is effectively improved.

An example memory device includes a memory cell array, a page buffer including buffer units corresponding to a plurality of memory cells of a page, a control logic configured to control a first read operation such that first hard decision data based on a normal read level and first soft decision data based on an offset level with respect to a first page are stored in the page buffer. The control logic is configured to perform a control operation of outputting the first hard decision data to a memory controller after a second read operation with respect to a second page has started in response to a first command that requests read of the second page from the memory controller and outputting the first soft decision data to the memory controller while the second read operation is being performed in response to a second command from the memory controller.

Examples provide for error correction. The error correction includes: obtaining first soft data according to first hard read data and cached second hard read data after hard decision decoding for the first hard read data fails, wherein the first hard read data is read according to a first hard read voltage, and the second hard read data is read according to a second hard read voltage before reading the first hard read data; and performing first soft decision decoding according to the first soft data and the first hard read data, or according to the first soft data and the second hard read data, wherein the first hard read voltage is one of a plurality of re-read voltages corresponding to a read command; and the second hard read voltage is an initial read voltage or a re-read voltage of the plurality of re-read voltages except the first hard read voltage.