Abstract: An integrated circuit comprises an interface controller to receive a message, wherein at least a portion of the message is encrypted, a primary processor coupled to the interface controller and configured to process the received message, and a secondary secure processor coupled to the primary processor and to the interface controller. The secondary secure processor is configured to decrypt the portion of the message that is encrypted on behalf of the primary processor, analyze the decrypted portion of the message to determine whether the decrypted portion comprises information pertaining to sensitive data, and responsive to determining that the decrypted portion comprises information pertaining to sensitive data, process the information pertaining to the sensitive data and provide the sensitive data to the interface controller via a secure private bus not accessible by the primary processor.

Abstract: Memory controllers, devices, modules, systems and associated methods are disclosed. In one embodiment, an integrated circuit (IC) dynamic random access memory (DRAM) device is disclosed. The IC DRAM device includes memory core circuitry organized into bank groups of storage cells, each bank group accessible via a corresponding bank group address. A command/address (C/A) interface receives C/A information defining a joint command. The joint command includes information specifying a first memory access operation, a first bank group address associated with the first memory access operation, and a second memory access operation to be automatically directed to the first bank group address.

Abstract: A sense amplifier for a memory device includes a primary latch and a holding latch that are independently controllable. The primary latch comprises a first set of transistors and the holding latch includes a second set of transistors having higher threshold voltages than the first set of transistors. In conjunction with a memory access operation, the primary latch and the holding latch sense and amplify a differential voltage of a pair of bitlines. A connectivity control circuit controls connectivity of the primary latch in different operational modes including pre-charge, offset pre-compensation, and amplification. In an active idle mode in between memory access operations while the wordline may remain active, the connectivity control circuit may turn off the primary latch while the holding latch holds the differential voltage on the bitlines to avoid leakage current through the primary latch.

Abstract: The described embodiments provide an integrated circuit (IC) memory controller to control operation of an IC memory device. The IC memory device includes a memory core and internal clock circuitry. The IC memory controller includes a command interface to transmit, to the IC memory device, a read command and a phase adjustment command. The phase adjustment command specifies a phase adjustment operation, to be performed by the IC memory device, in an intervening time between receipt of the read command by the IC memory device and the IC memory device outputting data accessed in response to receiving the read command. The internal clock circuitry is disabled after outputting the data. A data interface receives the data. A transmitter transmits a timing reference signal to the IC memory device to time outputting of the data by the IC memory device.

Abstract: A memory controller interfaces with one or more memory devices having configurable width data buses and configurable connectivity between data pins of the memory devices and data pins of the memory controller. Upon initialization of the memory devices, the memory controller automatically discovers the connectivity configuration of the one or more memory devices, including both individually selected and jointly selected devices. After discovering connectivity of the connected devices, the memory controller configures the memory devices according to the discovered connectivity and assigns unique addresses to jointly selected devices.

Abstract: Disclosed are techniques for a memory buffer to track access to paged regions of a memory system at a configurable granularity finer than the size of the paged regions to provide more detailed statistics on memory access. The memory buffer may advertise its capabilities for fine-grained cold page tracking. The memory buffer may receive from the host information to configure a granularity of sub-regions of a paged region and a size of counters used to track access to the sub-regions. The memory buffer may track access requests to the sub-regions using the counters and to provide information on sub-region tracking to the host to identify individual hot or cold sub-regions. The host may make migration decisions for the paged regions with more granular information such as compaction of sub-regions to create a cold page or to treat each sub-region as a separately compressible entity to compress a mostly cold page.

Abstract: Described is an integrated circuit with a driving amplifier that transmits a signal over a link (e.g. a wire) by raising and lowering a voltage on the link. A reference oscillator provides an error measure for the rate at which the voltage transitions between voltages, the slew rate. Slew-rate calibration circuitry adjusts the driving amplifier responsive to the error measure.

Abstract: A memory controller comprises a command interface to transmit a memory command to a plurality of memory devices associated with the memory controller. The memory controller also comprises an acknowledgement interface to receive an acknowledgment status packet from the plurality of memory devices over a shared acknowledgement link coupled between the memory controller and the plurality of memory devices, the acknowledgement status packet indicating whether the command was received by the plurality of memory devices. In addition, the memory controller comprises a memory controller core to decode the acknowledgment status packet to identify a portion of the acknowledgement status packet corresponding to each of the plurality of memory devices.

Abstract: A system includes a memory system comprising a near memory component and a far memory component, and a processing device coupled to the memory system. The processing device is configured to track, using a plurality of Bloom filters, respective access requests directed to a plurality of memory pages stored in the memory system. The processing device is further configured to identify, based on weighted results of respective queries of the plurality of Bloom filters, a subset of the plurality of memory pages having a data temperature that satisfies a threshold condition, and cause the subset of the plurality of memory pages to be stored in the near memory component. The identified subset can optionally be applied to a second stage of Bloom filters to further refine the result.

Abstract: A packaged semiconductor device includes a data pin, a first memory die, and a second memory die stacked with the first memory die. The first memory die includes a first data interface coupled to the data pin and a first memory core having a plurality of banks. The second memory die includes a second memory core having a plurality of banks. A respective bank of the first memory core and a respective bank of the second memory core perform parallel row access operations in response to a first command signal and parallel column access operations in response to a second command signal. The first data interface of the first die provides aggregated data from the parallel column access operations in the first and second die to the data pin.

Abstract: A memory controller includes a request queue and associated logic for efficiently managing the request queue based on various timing constraints of the memory device. A single request queue for the memory device stores read and write requests spanning different banks of the memory device. In each memory controller cycle, selection logic may select both a row request and a column request (relating to a different bank than the row request) for issuing to the memory device based on a set of timing status bits. Following issuance of requests, the memory controller updates the queue to maintain the queued requests in a time-ordered, compressed sequence. The memory controller furthermore updates the timing status bits that are used by the selection logic to select requests from the queue based on a history of past memory requests.

Abstract: Technologies for providing write protection to an integrated circuit memory device are described. The integrated circuit memory device has an input port to receive write data and command data and a read data output port to send read data. A device includes a write protection mechanism to prevent write data from being written to a memory core. The write protection mechanism can mask the write data or abort a write operation in response to activation. The write protection mechanism can be activated in response to an error detected by a serial data buffer (SDB) device coupled to the integrated circuit memory device.

Abstract: A dynamic random access memory (DRAM) device includes functions configured to aid with operating the DRAM device as part of data caching functions. The DRAM is configured to respond to at least two types of commands. A first type of command (cache data access command) seeks to access a cache line of data, if present in the DRAM cache. A second type of command (cache probe command) seeks to determine whether a cache line of data is present, but is not requesting the data be returned in response. In response to these types of access commands, the DRAM device is configured to receive cache tag query values and to compare stored cache tag values with the cache tag query values. A hit/miss (HM) interface/bus may indicate the result of the cache tag compare and stored cache line status bits to a controller.

Abstract: A memory module comprises an address buffer circuit, a command/address channel, and a plurality of memory components controlled by the address buffer circuit via the command/address channel. At least one memory component comprises a plurality of data ports, a memory core to store data, and a data interface. The data interface is capable of transferring data between the memory core and the data ports. The data interface supports a first data width mode in which the data interface transfers data at a first bit width and a first burst length via the data ports. The data interface also supports a second data width mode in which the data interface transfers data at a second bit width and second burst length via the data ports. The first bit width is greater than the second bit width and the first burst length is shorter than the second burst length.

Abstract: A multi-processor device is disclosed. The multi-processor device includes interface circuitry to receive requests from at least one host device. A primary processor is coupled to the interface circuitry to process the requests in the absence of a failure event associated with the primary processor. A secondary processor processes operations on behalf of the primary processor and selectively receives the requests from the interface circuitry based on detection of the failure event associated with the primary processor.

Abstract: A block of dynamic memory in a DRAM device is organized to share a common set of bitlines may be erased/destroyed/randomized by concurrently activating multiple (or all) of the wordlines of the block. The data held in the sense amplifiers and cells of an active wordline may be erased by precharging the sense amplifiers and then writing precharge voltages into the cells of the open row. Rows are selectively configured to either be refreshed or not refreshed. The rows that are not refreshed will, after a time, lose their contents thereby reducing the time interval for attack. An external signal can cause the isolation of a memory device or module and initiation of automatic erasure of the memory contents of the device or module using one of the methods disclosed herein. The trigger for the external signal may be one or more of temperature changes/conditions, loss of power, and/or external commands from a controller.

Abstract: Described are memory modules that support different error detection and correction (EDC) schemes in both single- and multiple-module memory systems. The memory modules are width configurable and support the different EDC schemes for relatively wide and narrow module data widths. Data buffers on the modules support the half-width and full-width modes, and also support time-division-multiplexing to access additional memory components on each module in support of enhanced EDC.

Abstract: A control component implements pipelined data processing operations in either of two timing domains bridged by a domain-crossing circuit according to one or more configuration signals that indicate relative clock frequencies of the two domain and/or otherwise indicate which of the two timing domains will complete the data processing operations with lowest latency.

Abstract: Space in a memory is allocated based on the highest used precision. When the maximum used precision is not being used, the bits required for that particular precision level (e.g., floating point format) are transferred between the processor and the memory while the rest are not. A given floating point number is distributed over non-contiguous addresses. Each portion of the given floating point number is located at the same offset within the access units, groups, and/or memory arrays. This allows a sequencer in the memory device to successively access a precision dependent number of access units, groups, and/or memory arrays without receiving additional requests over the memory channel.

Abstract: Multiple device stacks are interconnected in a ring topology. The inter-device stack communication may utilize a handshake protocol. This ring topology may include the host so that the host may initialize and load the device stacks with data and/or commands (e.g., software, algorithms, etc.). The inter-device stack interconnections may also be configured to include/remove the host and/or to implement varying numbers of separate ring topologies. By configuring the system with more than one ring topology, and assigning different problems to different rings, multiple, possibly unrelated, machine learning tasks may be performed in parallel by the device stack system.