Abstract: A data-shifting scheme is implemented where a group of arrays may be selected from a larger set of arrays. The arrays are connected to output-buffers and input-buffers such that data from the selected arrays may be read or written without changing addresses. The arrays are selected by programming the control signals controlling the output-buffers and input-buffers. The control signals may be programmed by several methods, for example, by blowing fuses or storing data in registers. The fuses do not have to be on pitch with the arrays. DRAMs, SRAMs, register arrays, and PLAs are examples of arrays that may be used with this invention. This invention is particularly useful for adding redundancy to an integrated circuit.

Abstract: A system and method which implement a memory component of an integrated circuit as multiple, relatively small sub-arrays of memory to enable great flexibility in organizing memory within the integrated circuit are provided. In a preferred embodiment, the memory component of an integrated circuit is implemented as multiple, relatively small sub-arrays of memory, which enable a designer great flexibility in arranging such sub-arrays within an integrated circuit. Also, in a preferred embodiment, the memory component of an integrated circuit is implemented as multiple memory sub-arrays that are each independent. For example, in a preferred embodiment, each memory sub-array comprises its own decode circuitry for decoding memory addresses that are being requested to be accessed by an instruction, and each memory sub-array comprises its own I/O circuitry.

Abstract: A static random access memory (SRAM) cell is provided that optimizes the density of memory cells in an array with the maximum speed possible in addressing the memory cells for reading and writing operations. The SRAM cell is divided into groups of SRAM arrays of cells with a centrally located distributed global decoder to address any individual memory cell in the SRAM array. The global decoder accepts an addressing input and outputs a signal for selecting an individual column of memory cells in the SRAM array. The global decoder also outputs a signal selecting an individual row of memory cells contained in the SRAM array. The global decoder may include logic to decode addressing bits to produce a group select signal. Thus, the global decoder is able to select any single memory cell in the SRAM cell for reading or writing specific logical states.

Abstract: A single-ended read, differential write CMOS SRAM cell has two inverters connected in a regenerative feedback circuit. Each inverter includes two complementary FETs. FETs of the same type in each inverter have differing gate widths and/or drive currents. The cell includes pass gate FETs having gate regions of approximately the same widths but differing lengths.

Abstract: A single-ended read, differential write CMOS SRAM cell has two inverters connected in a regenerative feedback circuit. Each inverter includes two complementary FETs. FETs of the same type in each inverter have differing gate widths and/or drive currents.

Abstract: A circuit for signalling if any like ordered bits Ak and Bk in first and second binary words differ comprises a comparator for each pair of like ordered bits and a common terminal. Each comparator includes first and second FETs arranged so: (a) the first and second levels of Ak are coupled to the common terminal via the first FET in response to Bk having the first value, (b) the first and second levels of Bk are coupled to the common terminal via the second FET in response to Ak having the first value, (c) the first FET decouples Ak from the common terminal and tends to cause the common terminal to be at the second level in response to Bk having the second value, (d) the second FET decouples Bk from the common terminal and tends to cause the common terminal to be at the second level in response to Ak having the second value, and (e) the common terminal is at the second level only in response to Ak Bk.

Abstract: A system and method are provided which enable a data carrier, such as a BIT line, to be held to a desired value while performing a memory access (e.g., a read or write operation) of SRAM in an efficient manner. In a preferred embodiment, cross-coupled PFETs are implemented to hold the BIT line to a desired value during a memory access of SRAM. As a result, a preferred embodiment enables a BIT line to transition from a high voltage value to a low voltage value free from conflict. That is, in a preferred embodiment, a holder PFET is not attempting to hold the BIT line high, while the SRAM or outside source (e.g., a “writing source”) is attempting to drive the BIT line to a low voltage value. Also, in a preferred embodiment, the BIT and NBIT lines (i.e., a complementary data carrier) can be driven to “true” low and “true” high voltage values.

Abstract: A static random access memory (SRAM) cell is provided that optimizes the density of memory cells in an array with the maximum speed possible in addressing the memory cells for reading and writing operations. The SRAM cell is divided into groups of SRAM arrays of cells with a centrally located distributed global decoder to address any individual memory cell in the SRAM array. The global decoder accepts an addressing input and outputs a signal for selecting an individual column of memory cells in the SRAM array. The global decoder also outputs a signal selecting an individual row of memory cells contained in the SRAM array. The global decoder may include logic to decode addressing bits to produce a group select signal. Thus, the global decoder is able to select any single memory cell in the SRAM cell for reading or writing specific logical states.

Abstract: A system and method are disclosed which provide the capability of repairing an optimum number of defective memory segments, such as RAM segments, in order to minimize the amount of unused repairing circuitry, such as fuses used for repairing defects within the memory. A preferred embodiment of the present invention provides a RAM block implemented such that the number of fuses required for repairing defects therein is proportional to the optimum number of defective segments capable of being repaired. A preferred embodiment allows for repairing an optimum number of defective segments, while being capable of repairing any of the segments (up to the optimum number) by mapping repair data to an appropriate defective segment.

Abstract: A static random access memory (SRAM) cell is provided that optimizes the density of memory cells in an array with the maximum speed possible in addressing the memory cells for reading and writing operations. The SRAM cell is divided into groups of SRAM arrays of cells with a centrally located distributed global decoder to address any individual memory cell in the SRAM array. The global decoder includes a first logic block that accepts addressing input and outputs a signal for selecting an individual column of memory cells in the SRAM array. The global decoder includes a second logic block that accepts addressing input and outputs a signal selecting an individual row of memory cells contained in the SRAM array. The global decoder may include a third logic block to decode addressing bits to produce a group select signal. Thus, the global decoder is able to select any signal memory cell in the SRAM cell for reading or writing specific logical states.

Abstract: A single-ended read, differential write CMOS SRAM cell has two inverters connected in a regenerative feedback circuit. Each inverter includes two complementary FETs. FETs of the same type in each inverter have differing gate widths and/or drive currents. The cell includes pass gate FETs having gate regions of approximately the same widths but differing lengths.

Abstract: A system and method are disclosed which provide a register structure enabling a dual-ended write thereto with a minimum amount of high-level metal tracks and components, thereby minimizing the amount of surface area required for such register structure. A data carrier (e.g., a BIT line) is utilized to carry a data value desired to be written from a port to a memory cell of a register structure. Such a data carrier may be implemented as a high-level metal track that spans multiple register structures to enable a port the capability of writing to such multiple register structures. Also, a line for triggering a write operation for a port (e.g., a WORD line) is implemented, and such a triggering line may be implemented as a high-level metal track. A preferred embodiment provides a register structure that includes a dual-ended write mechanism. In a preferred embodiment, a complementary data carrier for a port is generated locally within a register structure.

Abstract: A system and method are disclosed which provide a pulse write mechanism to enable a port to write to a register structure without requiring a large amount of circuitry. One or more ports may be coupled to a register structure in a manner that enables the ports to write data to the register structure without requiring a large amount of circuitry. The ports may be coupled to the register structure in a manner that enables them the capability of reading data from the register structure without requiring additional circuitry beyond that required for a write operation. A preferred embodiment implements a single-ended write structure, wherein a data carrier (e.g., BIT line) is utilized to carry a data value desired to be written for a port. A preferred embodiment comprises a write pulse mechanism, such as a NFET, capable of setting the memory cell to an initial value before performing a write thereto.

Abstract: The present invention integrates a WWTM circuit with the write driver circuitry, which is an inherent part of any conventional SRAM design. Thus, a circuit for writing data into and weak write testing a memory cell is provided. In one embodiment, the circuit comprises a write driver that has an output for applying a write or a weak write output signal at the memory cell. The write driver has first and second selectable operating modes. In the first mode, the write driver is set to apply a weak write output signal from the output for performing a weak write test on the cell. In the second mode, the write driver is set to apply a normal write output signal that is sufficiently strong for writing a data value into the cell when it is healthy.

Abstract: A single-ended SRAM cell design reduces SRAM size and provides high storage cell noise margin. A virtual ground line is coupled to the source of the driver NFET of each I/O port inverter of each storage cell in a common bitline column. An isolation mechanism couples the virtual ground line to a low reference voltage during reads and during a write of a "0" to a storage cell, and isolates the virtual ground line from the low reference voltage during a write of a "1" to a storage cell. A clamping device is coupled to the virtual ground line to prevent the potential on the virtual ground line from exceeding the threshold voltage of the isolation mechanism and flipping the stored value in any of the other commonly coupled storage cells when a "1" is being written to another of the commonly coupled storage cells.

Abstract: An improved random access memory (RAM) system enhances the speed and reduces power dissipation and logic complexity associated with a RAM. The RAM system includes first and second pluralities of RAM cell columns. Each of the columns includes (1) at least one RAM cell, each RAM cell configured to read and write a respective logic state and (2) bit and nbit connections (differential and complimentary) connected to each of the RAM cells. A first multiplexer is designed to multiplex the bit and nbit connections of the first plurality of RAM cell columns. A second multiplexer is configured to multiplex the bit and nbit connections of the second plurality of columns. Decode logic controls the first and second multiplexers, and the decode logic accesses a particular column and cell in one of the first and second pluralities during each memory access. A sense amplifier is configured to read the bit and nbit connections of the first and second pluralities via respectively the first and second multiplexers.