Abstract: A processor which optimizes performance opportunistically by using a hierarchy of variables comprising voltage, clocking and the operations being performed by the processor or its system. The invention accomplishes performance optimization by defining various states with the goal that the processor stays in an optimal performance state of accelerated voltage and clock when the processor executional units are operating. The states are selected by a logic network based on information that is provided by temperature sensors and a performance control. The logic network can be envisioned as an UP-DOWN counter. The counter can be advanced UP or DOWN the state "ladder" as the conditions warrant.

Abstract: According to the preferred embodiment, a device and method for reducing power consumption by reducing unneeded node toggling is provided. The preferred embodiment reduces unneeded node toggling in a circuit by utilizing either a pull-up or pull-down transistor to pull the input of the circuit to a state that minimizes power consumption during periods in which the circuit is inactive. By tying the circuit input high or low during inactivity, node toggling is reduced or eliminated in that circuit. In the preferred embodiment, the inputs to the circuit all pulled after a time of inactivity which is proportional to the leakage current of the leakiest transistor in the circuit. By timing the input pulling proportional to the leakage current, the power consumption is minimized without excessive power loss caused by the pulling itself.

Abstract: A voltage limiting circuit for fuse technology. The voltage limiting circuit is coupled to the two terminals. The voltage limiting circuit is responsive to a fuse blow through a low impedance sensing circuit, and then minimizes the voltage across the fuse gap that is created by the fuse blow. Thus, the invention prevents dendritic growth and corrosion in copper or similar types of fuses.

Abstract: According to the present invention, an anticipating cache memory loader is provided to "pre-load" the cache with the data and instructions most likely to be needed by the CPU once the currently executing task is completed or interrupted. The data and instructions most likely to be needed after the currently executing task is completed or executed is the same data and instructions that were loaded into the cache at the time the next scheduled task was last preempted or interrupted. By creating and storing an index to the contents of the cache for various tasks at the point in time the tasks are interrupted, the data and instructions previously swapped out of the cache can be retrieved from main memory and restored to the cache when needed. By using available bandwidth to pre-load the cache for the next scheduled task, the CPU can begin processing the next scheduled task more quickly and efficiently than if the present invention were not utilized.

Abstract: An integrated circuit implements simple and efficient normal power to low power and low power to normal power transitions. Dedicated shadow latch circuits are added, each having a corresponding system latch. The state of the system latches is transferred to the shadow latches upon a transition from normal to low power mode and the stored information is transferred back to the system latches on the transition from low power to normal power operation. The shadow latches are optimized to minimize power usage during low power operation.

Abstract: A soft latch circuit having a first and second inverter is disclosed. The output of the first inverter is connected to the input of the second inverter, and the output of the second inverter is connected to the input of the first inverter. The first inverter includes a complimentary pair of field-effect transistors (FETs). The second inverter includes either a complimentary pair of current mirrors, or a current mirror and a complimentary FET, the latter providing improved noise immunity characteristics when the soft latch is set in only one direction.

Abstract: An apparatus and method for pre-loading a cache memory based on information contained in a compiler generated program map are disclosed. The program map is generated by the compiler at the time source code is compiled into object code. For each application program, the user would have this program map stored with the object file. At the beginning of the program execution cycle, the operating system will determine whether or not a program map exists for the application. If a program map exists, the operating system will load the program map into an area of RAM designated as the program map random access memory (RAM). The program map will be used to pre-load the cache with the appropriate data and instructions for the central processing unit (CPU) to process. The program mapping would be the address location of each jump/branch target that the CPU might encounter during the execution cycle. Each of these locations represent a starting point for a new code sequence.

Abstract: A high speed/narrow I/O DRAM device comprises both a data input/output (I/O) port as well as a command port for receiving commands used to control the operations of the DRAM. The command port is defined as input only (i.e., for inputting command data). The present invention comprises multiplexing write data to be written and stored in the DRAM onto the command port with command data packets. The data I/O port can then become dedicated to streaming out seamless data since it no longer needs to flip between input and output data. Even greater bus efficiency can be realized if, during a command packet transfer, data writes to the DRAM are switched back to the data I/O port. With this input port switching protocol, greater bus efficiency and increased memory performance can be realized.

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: An input buffer circuit has a switching point accurately set according to the input logic level, even when the input buffer circuit has a low supply voltage. The switching point is set according to an internal reference voltage of equal magnitude to the desired switching point that is applied to a current source. The current source accurately sources (or sinks) a current matching the current flowing in an input inverter when the input logic level substantially equals the reference voltage. At that point, the voltage at the output of the input inverter is substantially equal one half of the supply voltage. When the input logic level is slightly below or above the reference voltage, the output of the input inverter is near the supply or ground rail, respectively. Hysteresis is added to compensate for noise that may exist on the input logic signal.

Abstract: A processor which optimizes performance opportunistically by using a hierarchy of variables comprising voltage, clocking and the operations being performed by the processor or its system. The invention accomplishes performance optimization by defining various states with the goal that the processor stays in an optimal performance state of accelerated voltage and clock when the processor executional units are operating. The states are selected by a logic network based on information that is provided by temperature sensors and a performance control. The logic network can be envisioned as an UP-DOWN counter. The counter can be advanced UP or DOWN the state "ladder" as the conditions warrant.

Abstract: A compound cascode amplifier comprising first and second FET input transistors, the gates of which are coupled to a differential input, and first and second FET cascode transistors. The sources of the first and second cascode transistors are coupled respectively to the drains of the first and second input transistors. The gate of the first cascode transistor is coupled to a reference voltage V Ref. The drain of the first cascode transistor is coupled to the gate of the second cascode transistor, and the drain of the second cascode transistor forms the output of the circuit. In operation the drive to the gate of the second cascode transistor arrives in synchronization with the drive to the source thereof, such that the drive to the gate arrives in anticipation of a voltage swing at the output of the second cascode transistor amplifier, not in reaction to it. Accordingly, this arrangement does not introduce a delay or an additional pole in the frequency response of the circuit.

Abstract: A calibrated multi-voltage level system is disclosed having a network of devices, including a first and a second device. The first device comprises a processor for generating data, an encoding unit for encoding the data into a first data signal having multiple voltage levels, and a transmitting unit for transmitting the encoded data signal to the second device. The first device also comprises a calibration unit for sending a first calibration signal to the second device, and for storing a second calibration signal from the second device; and an adaptation unit for correcting the second data signal from the second device with the stored second calibration signal.

Abstract: A structure and method for removing and recovering an anodically bonded glass device from a substrate using a metal interlayer interposed between the glass and the substrate is provided. As used in semiconductor mask fabrication, the structure comprises a silicon wafer substrate coated with a membrane on which a metal interlayer is disposed. The metal interlayer and a glass device are anodically bonded together. Recovery of the glass device is accomplished by chemically and mechanically removing the wafer and its membrane from the metal interlayer. The membrane is preferably removed using reactive ion etching to which the metal interlayer is resistant. The metal interlayer is then removed from the glass device using a highly corrosive chemical solution. The recovered glass device may then be reused.

Abstract: A CMOS inverter circuit having a resistive bias device is disclosed. The CMOS inverter circuit comprises a pair of inverter transistors for receiving an input signal. At least one pair of compensating transistors is coupled to the inverter transistors for providing nonlinearity to the input signal. An inverter, coupled to the drains of the inverter transistors at a first node, receives the nonlinear signal as an input. The resistive bias device, coupled to the output of the inverter and to the compensation transistors, provides adjustable reference voltages to the compensation transistors, which allow for an improved noise immunity and high transition gain. The output, taken from the first node, provides for an improvement in the performance of the circuit.

Abstract: A relatively low frequency oscillator in junction with a much higher frequency oscillator is used to produce a clock that is both accurate and minimizes power consumption. The high frequency oscillator is enabled only during a small portion of the clock's operation and is used to gauge the output of the low frequency oscillator. The output of the high frequency oscillator is counted during its operation period, and the amount counted is accumulated for subsequent time periods. When the accumulated count reaches a predetermined value, a clock output is provided.

Abstract: According to the preferred embodiment, a buffer amplifier is provided that provides improved linearity while providing increased control over the gain without unduly limiting the amplifier frequency response. The amplifier preferably includes a series pair of transistors with their gates connected to the amplifier input and their drains connected to the amplifier output. The amplifier further includes a pair of feedback transistors connected in series with the series pair. The gates of the feedback transistors are connected to the amplifier output through a pair of feedback networks. Each network includes at least one impedance element. The impedance elements are preferably selected to maximize the linearity of the amplifier response. Furthermore, the impedance elements can be selected to modify the gain of the amplifier, increasing the amplifier gain if needed.

Abstract: A semiconductor memory produced in a unipolar technology includes a cell which has an inversion capacitor with one terminal connected to a bit/sense line, the other terminal is coupled to a source of charges by a pulse from a word line. To provide a word organized array of these cells, each word includes a source of charges produced at the surface of a semiconductor substrate and plurality of inversion capacitors are formed also at the surface of the semiconductor in spaced apart relationship from the charge source. Information is written into the capacitors by applying voltages of two different magnitudes, representing 1 and 0 bits of information, to one terminal of the capacitors while a word pulse produces inversion layers at the surface of the substrate between the capacitors to interconnect serially the charge source with each of the capacitors. The capacitors having the larger voltage store the greater amount of charge.

Abstract: An integrated circuit memory system includes capacitive storage memory cells capable of storing n bits of information on n capacitors associated with multiple emitters of a bilaterally conductive bipolar transistor. Each capacitor is coupled to a separate bit/sense line. Access of a storage cell is achieved by forward biasing the common base/collector junction of the bipolar transistor. Writing is achieved by driving the bit/sense lines to charge or discharge the storage capacitors during an access cycle. In reading, or sensing, the charged state of each storage capacitor is determined by sensing potential changes on the bit/sense lines during access. Fabrication of memory arrays is possible by any one of several different techniques, all of which are compatible with high speed bipolar logic circuits.