Abstract: An asynchronous switching apparatus is enabled to reshape data pulses and eliminate skewing problems as data is transmitted through the switch. The switching apparatus still functions asynchronously and maintains all the advantages of asynchronous operation, such as, not requiring the alignment and distribution of a central clock, having no central point failure mechanisms, and allowing each node of the parallel system to function free of synchronization requirements with other nodes.

Abstract: A method of providing a predetermined level and state of stress in a film deposited on a surface of a substrate. In one embodiment, a layer of crystalline material is deposited on a surface of a substrate and then a layer of amorphous material is deposited on the layer of crystalline material. Then, the layers are heated, causing the amorphous material to crystallize. Such crystallization reduces, or even changes the state of, stress in the amorphous layer, which in turn alters the forces applied by the layer to adjacent regions of the substrate. The method may be used for filling a deep-trench capacitor of the type used in trench-storage DRAMs.

Abstract: A process and structure are disclosed for a programmable array for use in a read-only memory comprising diode elements and shorted diode elements. The elements are connected across bit and wordlines. The invention utilizes lateral polysilicon diodes and metal silicide layer bridging the junction of pre-selected diodes to short pre-selected diode elements. Programming is accomplished by either forming the silicide layer across the junctions of pre-selected diodes or removing the silicide layer from the junctions of pre-selected diodes.

Abstract: An improved differential source follower with negligible input/output mismatch over a wide range of input signal magnitudes. A pair of FETs and current sources provides bias current control for each differential output that cancels the inherent body-source voltage variation of the source followers which acts to attenuate the unity gain output signal.

Abstract: A PLA whose slowest product terms are located as close as possible to the true/complement generators or input buffers. The associated input buffers and product terms are partitioned into two or more sections. A modified gap cell recombines the product terms before propagating the signal into an array.

Abstract: A superscalar uniprocessor that performs concurrent multi-task processing is provided. The processor of the present invention maintains a complete set of program address, memory control and general data registers for each task executing concurrently within the microprocessor, allowing independent control of the program flows. Each set of registers are associated with only one task and are utilized by the memory control and execution units to execute the associated task. The processor includes an instruction fetcher and memory management unit that retrieves an instruction from memory for a given task, as directed by the task's address and control registers, and attaches a task tag to the retrieved instruction that identifies that task. The superscalar processor has a plurality of execution units that can execute a plurality of tasks simultaneously, and a dispatch unit that sends a retrieved instruction and its attached task tag to one of the plurality of execution units for execution.

Abstract: An integrated circuit chip having circuitry to adjust its threshold voltage between a plurality of threshold voltages for interfacing to integrated circuit chips having different supply voltages. The integrated circuit chip also includes circuitry for communicating its threshold voltage level to a second integrated circuit such that the second integrated circuit may set its threshold voltage prior to receiving logic communications from the integrated circuit. The present invention also discloses an integrated circuit that detects the logic level of incoming logic communications and adjusts its threshold voltage accordingly.

Abstract: A process and structure are disclosed for a programmable array for use in a read-only memory comprising diode elements and shorted diode elements. The elements are connected across bit and wordlines. The invention utilizes lateral polysilicon diodes and metal silicide layer bridging the junction of pre-selected diodes to short pre-selected diode elements. Programming is accomplished by either forming the silicide layer across the junctions of pre-selected diodes or removing the silicide layer from the junctions of pre-selected diodes.

Abstract: Integrated circuit chips are screened for susceptibility to latch-up by first applying power and ground to the chips to be tested while limiting current flow to a non-destructive compliance value. Next, the chips are irradiated with a pulse of radiation having an energy dose calibrated to trigger latch-up in latch-up sensitive chips. Upon termination of the radiation, the current is detected. Chips having current persisting at the compliance value are indicated as failing. The current in passing chips returns approximately to the original standby current value. In the preferred embodiment, the radiation is visible light and the radiation energy dose is selected to cause a percentage of chips to latch-up approximating the percentage of failures expected at burn-in.

Abstract: The present invention addresses the limitations of prior art ALLNODE switches by including dual priority, adaptive, path seeking, and flash-flood functionalities in a single ALLNODE switch. The switch of the present invention further includes a selection device responsive to a selection signal for enabling the selection of the mode of switch operation from any one of the foregoing functionalities. The selection signal is applied to the switch in a number of different ways including: the transmission of a command over the data path interface to the switch; the transmission of a command over special purpose serial or parallel control lines; or via hardwiring. Thus, the selection of functionality for the switch is capable of being made in either a dynamic or static fashion. The present invention further comprises two new high performance networks utilizing the selectable function ALLNODE switch.

Abstract: A power efficient implementation of a single-pulse generator requiring less chip area and fewer circuit devices.

Abstract: An apparatus for redirecting late-entering PLA input signals to avoid holding up the entire PLA while the late-entering signals are processed, and a method for designing the same.

Abstract: An improved current mirror circuit with isolation of the output leg for improved stabilization of the circuit even when heavily loaded.

Abstract: A method and hardware apparatus provide a fault tolerant and flexible multi-stage network addressing scheme for transmitting a message with a header containing control bits for selecting from various destination checking functions to be performed. Upon arrival of the message at a node, destination checking is performed or not in response to the massage's header. If destination checking is not performed, or if destination checking is performed and indicates that the node is the desired destination for the message, the message is accepted. If destination checking is performed and indicates that the node is not the desired destination for the message, the message is rejected. Destination checking is disabled during address assignment, broadcasting and multi-casting, and replaced with one's complement-based verification of the sending node.

Abstract: A circuit for converting received input signals to highly symmetrical CMOS level outputs having fast slew rates. The circuit can accept a differential input signal with a wide range of common mode voltages. A first stage level shifts the input signals to provide a ground-based common mode output to a second stage level shifter which centers the input signals around the midpoint between V.sub.cc and ground and which increases their voltage swing. The final stage provides a full, highly symmetric, rail-to-rail output capable of driving highly capacitive loads at high rates and which is immune to temperature, V.sub.cc, and process variations.

Abstract: A method for wiring IC chips such that electromigration criteria are met while minimizing the effect on overall chip wireability. A technique to optimize the width of automatically routed wire segments so that these widths are adequate to support the electromigration current on that net as a function of the capacitive loading of the net itself.

Abstract: An integrated cache architecture that has low power consumption, high noise immunity, and full support of an integrated validity/least recently used (LRU) cache write mode. The cache stores TAG, index and LRU information directly on a master word line, and cache line data on local word lines. The access information is made available early in the cycle, allowing the cache to disable local word lines that are not needed. The master word lines and local word lines having approximately the same cycle time. By laying out the master and local word lines in a metal layer that substantially renders the stored data immune to overlaying noise sources, high frequency interconnections can be made over the cache without disturbing the stored data. The architecture circuitry efficiently updates the stored LRU information, such that a combined data validity/full LRU cache update protocol is supported.

Abstract: An apparatus for redirecting late-entering PLA input signals to avoid holding up the entire PLA while the late-entering signals are processed, and a method for designing the same.

Abstract: An integrated cache architecture that has low power consumption, high noise immunity, and full support of an integrated validity/LRU cache write mode. The cache stores TAG, index and LRU information directly on a master word line, and cache line data on local word lines. The access information is made available early in the cycle, allowing the cache to disable local word lines that are not needed. By laying out the master and local word lines in a metal layer that substantially renders the stored data immune to overlaying noise sources, high frequency interconnections can be made over the cache without disturbing the stored data. The cache includes circuitry for efficiently updating the stored LRU information, such that a combined data validity/full LRU cache update protocol is supported.

Abstract: A two-port memory cell design which permits simultaneous reading and writing of cells which are on the same wordline but on different Bit Select lines without increase in Read Access Time, and while maintaining memory functionality at low voltages. The memory cell uses a standard 6 transistor design to provide a differential read for fast access plus another three transistors are added to each cell to provide a means of differentially writing the cell and de-gating the write if the bit-select is not active. This cell design has applicability to multi-port memories as well.