Abstract: An integrated circuit input/output interface with empirically determined delay matching is disclosed. In one embodiment, the integrated circuit input/output interface uses empirical information of the signal traces to adjust the transmit/receive clock of each pin of the interface so as to compensate for delay mismatches caused by differences in signal trace lengths. The empirical information, in one embodiment, includes signal flight time of each signal trace, which can be pre-measured or pre-calculated from known signal trace lengths. The empirical information, in another embodiment, includes trace-specific phase offset values calculated from pre-calculated or pre-measured signal flight times or signal trace lengths. In yet another embodiment, a transmitting device generates a set of serially delayed write clocks, which are used to control symbol transmission over signal traces so as to reduce simultaneous switching output noise and ground bound in the transmitting device.

Abstract: A memory device including an array of memory cells, and a register circuit to store a value representative of a period of time to elapse before the memory device is ready to receive a command when recovering from a power down mode is provided in an embodiment. The command specifies an access to the array of memory cells. A delay lock loop circuit synchronizes data transfers using an external clock signal. The delay lock loop circuit reacquires synchronization with the external clock signal during the period of time.

Abstract: A dynamic random access memory device includes banks of dynamic memory cells. The device performs a refresh operation in response to receiving a self refresh command, by refreshing rows of the memory cells located in each of the banks. Further, a refresh frequency for the refresh operation is selected such that the refresh frequency is minimized to conserve power consumed by the memory device while being sufficient to refresh the rows of the memory cells.

Abstract: An integrated circuit device having a select circuit, a summing circuit and a phase mixer. The select circuit selects one of a plurality of offset values as a selected offset. The summing circuit sums the selected offset with a phase count value, the phase count value indicating a phase difference between a reference clock signal and a first plurality of clock signals. The phase mixer combines the first plurality of clock signals in accordance with the sum of the selected offset and the phase count value to generate an output clock signal.

Abstract: A memory system architecture/interconnect topology includes a configurable width buffered memory module having a configurable width buffer device with at least one bypass circuit. A buffer device, such as a configurable width buffer device, is positioned between or with at least one integrated circuit memory device positioned on a substrate surface of a memory module, such as a DIMM. The configurable width buffer device is coupled to at least one memory device (by way of an internal channel), entry pin and exit pin on the memory module. The configurable width buffer device includes a multiplexer/demultiplexer circuit coupled to the entry pin and the internal channel for accessing the memory device. A bypass circuit is coupled to the entry pin and the exit pin in order to allow information to be transferred through the memory module to another coupled memory module in the memory system by way of an external channel.

Abstract: A method and apparatus for position dependent data scheduling for communication of data for different domains along a bus is provided. Having an awareness of the relative position of different domains along a bus, one embodiment of the invention schedules bus operations to allow data from multiple bus operations to be simultaneously present on the bus while preventing interference among the data. The invention is compatible with buses having a termination on one end and those having terminations on both ends. In accordance with one embodiment of the invention, bus operations are scheduled so that first data of a first bus operation involving a first domain are not present at domains involved in a second bus operation at times that would result in interference with second data of the second bus operation.

Abstract: Circuitry for performing a memory block write is described. The memory block includes b block words, each block word having t block bytes. Each block byte has s bits of memory. Each block byte is associated with at least two associated mask value bits. A constant register has at least s.times.t bits of memory arranged as t constant bytes, each constant byte storing a constant value, each constant byte associated with one block of every block word. The block write circuitry includes control circuitry for selecting one of a normal write function and a block write function in accordance with a block write signal. When the block write function is selected, the control circuitry stores the associated constant value in every nonmasked block byte substantially simultaneously in accordance with a value of the associated mask value bits.

Abstract: A cache system which includes prefetch pointer fields for identifying lines of memory to prefetch thereby minimizing the occurrence of cache misses. This cache structure and method for implementing the same takes advantage of the previous execution history of the processor and the locality of reference exhibited by the requested addresses. In particular, each cache line contains a prefetch pointer field which contains a pointer to a line in memory to be prefetched and placed in the cache. By prefetching specified lines of data with temporal locality to the lines of data containing the prefetch pointers the number of cache misses is minimized.