Abstract: A plurality of processing elements (PEs) include memory local to at least one of the processing elements in a data packet-switched network interconnecting the processing elements and the memory to enable any of the PEs to access the memory. The network consists of nodes arranged linearly or in a grid to connect the PEs and their local memories to a common controller. The processor performs memory accesses on data stored in the memory in response to control signals sent by the controller to the memory. The local memories share the same memory map or space. The packet-switched network supports multiple concurrent transfers between PEs and memory. Memory accesses include block and/or broadcast read and write operations, in which data can be replicated within the nodes and, according to the operation, written into the shared memory or into the local PE memory.

Abstract: Memory modules, systems, memory controllers and associated methods are disclosed. In one embodiment, a memory module includes a module substrate having first and second memory devices. Buffer circuitry disposed on the substrate couples to the first and second memory devices via respective first and second secondary interfaces. The buffer circuitry includes a primary signaling interface for coupling to a group of signaling links associated with a memory controller. The primary signaling interface operates at a primary signaling rate and the first and second secondary data interfaces operate at a secondary signaling rate. During a first mode of operation, the primary interface signaling rate is at least twice the secondary signaling rate. A first time interval associated with a transfer of first column data via the first secondary interface temporally overlaps a second time interval involving second column data transferred via the second secondary interface.

Abstract: A receiver serial data streams generates a local timing reference clock from an approximate frequency reference clock by phase-aligning the local clock to transitions in the data stream. This process is commonly known as clock and data recovery (CDR). Certain transitions of the data signals are selected for use in phase-aligning the local clock, and certain transitions are ignored. Phase-error signals from multiple receivers receiving the multiple serial data streams are combined and used to make common phase adjustments to the frequency reference clock. These common adjustments track jitter that is common to the received data streams. Local adjustments that better align each respective local clock to the transitions of its respective serial data stream are made using a local phase-error signal. These local adjustments track jitter that is more unique to each of the respective serial data streams.

Abstract: A clock signal is transmitted to first and second integrated circuit (IC) components via a clock signal line, the clock signal having a first arrival time at the first IC component and a second, later arrival time at the second IC component. A write command is transmitted to the first and second IC components to be sampled by those components at respective times corresponding to transitions of the clock signal, and write data is transmitted to the first and second IC components in association with the write command. First and second strobe signals are transmitted to the first and second IC components, respectively, to time reception of the first and second write data in those components. The first and second strobe signals are selected from a plurality of phase-offset timing signals to compensate for respective timing skews between the clock signal and the first and second strobe signals.

Abstract: The timing of the synchronous interface is controlled by a dock signal driven by a controller. The clock is toggled in order to send a command to a memory device via the interface. If there are no additional commands to be sent via the interface, the controller suspends the clock signal. When the memory device is ready, the memory device drives a signal back to the controller. The timing of this signal is not dependent upon the clock signal. Receipt of this signal by the controller indicates that the memory device is ready and the clock signal should be resumed so that a status of the command can be returned via the interface, or another command issued via the interface.

Abstract: The embodiments described herein describe technologies for memory systems. One implementation of a memory system includes a motherboard substrate with multiple module sockets, at least one of which is populated with a memory module. A first set of data lines is disposed on the motherboard substrate and coupled to the module sockets. The first set of data lines includes a first subset of point-to-point data lines coupled between a memory controller and a first socket and a second subset of point-to-point data lines coupled between the memory controller and a second socket. A second set of data lines is disposed on the motherboard substrate and coupled between the first socket and the second socket. The first and second sets of data lines can make up a memory channel.

Abstract: A first timing reference signal and a second timing reference signal are sent to a memory device. The second timing reference signal has approximately a quadrature phase relationship with respect to the first timing reference signal. A plurality of serial data patterns are received from the memory device. The transitions of the first timing reference and the second timing reference determining when transitions occur between the bits of the plurality of data patterns. Timing indicators associated with when received transitions occur between the bits of the plurality of data patterns are received from the memory device. The timing indicators are each measured using a single sampler. Based on the timing indicators, a first duty cycle adjustment for the first timing reference signal, a second duty cycle adjustment for the second timing reference signal, and a quadrature phase adjustment are determined and applied.

Abstract: A memory component internally generates and stores the check bits of error detect and correct code (EDC). In a first mode, during a read transaction, the check bits are sent to the memory controller along with the data on the data mask (DM) signal lines. In a second mode, an unmasked write transaction is defined where the check bits are sent to the memory component on the data mask signal lines. In a third mode, a masked write transaction is defined where at least a portion of the check bits are sent from the memory controller on the data signal lines coincident with an asserted data mask signal line. By sending the check bits along with the data, the EDC code can be used to detect and correct errors that occur between the memory component and the memory controller.

Abstract: In response to a first memory access transaction having a first base address, data fields and a repair fields are retrieved from a first DRAM channel. The data fields include a first data field. The repair fields include a first repair field storing repair data. The repair data is to replace any data in the first data field. In response to a second memory access transaction having a second base address, repair tag fields are retrieved from a second DRAM channel. The repair tag fields include a repair tag field that indicates the repair data is be replace the data stored in the first data field.

Abstract: Described are dynamic memory systems that perform overlapping refresh and data-access (read or write) transactions that minimize the impact of the refresh transaction on memory performance. The memory systems support independent and simultaneous activate and precharge operations directed to different banks. Two sets of address registers enable the system to simultaneously specify different banks for refresh and data-access transactions.

Abstract: Described are memory systems in which a memory controller issues commands and addresses to multiple memory modules that collectively support each read and write transactions. A common set of control signal lines from the controller communicates the same command and address signals to the modules. For write commands, the controller sends subsets of write data to each module over a respective subset of data lines. For read commands, each module responds with a subset of the requested data over the respective subset of data lines. The memory modules can be width configurable so that a single full-width module can connect to both subsets of data lines to convey full-width data, or two half-width modules can connect one each to the subsets of data lines.

Abstract: A memory device includes an array of resistive memory cells wherein each pair of resistive memory cells includes a first switching element electrically coupled in series to a first resistive memory element and a second switching element electrically coupled in series to a second resistive memory element. A source of the first switching element and a source of the second switching element receive a common source line signal.

Abstract: An image sensor generates first digital samples and second digital samples during respective first and second sampling intervals, the first digital samples including at least one digital sample of each pixel of a first plurality of pixels, and the second digital samples including at least one digital sample of each pixel of a second plurality of pixels. A sum of the first digital samples is accumulated within a first counter as the first sampling interval transpires, and a sum of the second digital samples is accumulated within the first counter as the second sampling interval transpires.

Abstract: In a multirank memory system in which the clock distribution trees of each rank are permitted to drift over a wide range (e.g., low power memory systems), the fine-interleaving of commands between ranks is facilitated through the use of techniques that cause each addressed rank to properly sample commands intended for that rank, notwithstanding the drift. The ability to perform such “microthreading” provides for substantially enhanced memory capacity without sacrificing the performance of single rank systems. This disclosure provides methods, memory controllers, memory devices and system designs adapted to these ends.

Abstract: The gate of the access transistor of a 1 transistor 1 resistor (1T1R) type RRAM cell is biased relative to the source of the access transistor using a current mirror. Under the influence of a voltage applied across the 1T1R cell (e.g., via the bit line), the RRAM memory element switches from a higher resistance to a lower resistance. As the RRAM memory element switches from the higher resistance to the lower resistance, the current through the RRAM cell switches from being substantially determined by the higher resistance of the RRAM device (while the access transistor is operating in the linear region) to being substantially determined by the saturation region operating point of the access transistor.

Abstract: A memory is disclosed that includes a logic die having first and second memory interface circuits. A first memory die is stacked with the logic die, and includes first and second memory arrays. The first memory array couples to the first memory interface circuit. The second memory array couples to the second interface circuit. A second memory die is stacked with the logic die and the first memory die. The second memory die includes third and fourth memory arrays. The third memory array couples to the first memory interface circuit. The fourth memory array couples to the second memory interface circuit. Accesses to the first and third memory arrays are carried out independently from accesses to the second and fourth memory arrays.

Abstract: Memory system enabling memory mirroring in single write operations for the primary and backup data storage. The memory system utilizes a memory channel including one or more latency groups, with each latency group encompassing a number of memory modules that have the same signal timing to the controller. A primary copy and a backup copy of a data element can be written to two memory modules in the same latency group of the channel and in a single write operation. The buses of the channel may have the same trace length to each of the memory modules within a latency group.

Abstract: A signaling circuit having a selectable-tap equalizer. The signaling circuit includes a buffer, a select circuit and an equalizing circuit. The buffer is used to store a plurality of data values that correspond to data signals transmitted on a signaling path during a first time interval. The select circuit is coupled to the buffer to select a subset of data values from the plurality of data values according to a select value. The equalizing circuit is coupled to receive the subset of data values from the select circuit and is adapted to adjust, according to the subset of data values, a signal level that corresponds to a data signal transmitted on the signaling path during a second time interval.

Abstract: Memory system enabling memory mirroring in single write operations. The memory system includes a memory channel which can store duplicate copies of a data element into multiple locations in the memory channel. The multiple locations are disposed in different memory modules and have different propagation times with respect to a data signal transmitted from the memory controller. In a write operation, the relative timings of the chip select, command and address signals among the multiple locations are adjusted according to the data propagation delay. As a result, a data element can be written into the multiple locations responsive to a data signal transmitted from the memory controller in a single transmission event.

Abstract: A memory module can be programmed to deliver relatively wide, low-latency data in a first access mode, or to sacrifice some latency in return for a narrower data width, a narrower command width, or both, in a second access mode. The narrow, higher-latency mode requires fewer connections and traces. A controller can therefore support more modules, and thus increased system capacity. Programmable modules thus allow computer manufacturers to strike a desired balance between memory latency, capacity, and cost.