Abstract: Selector circuits and systems for single and multilevel selection within one clock cycle having a static switching factor on the output of a dynamic logic circuit. A logic device for single and multilevel selection having a dynamic logic circuit portion and a static logic circuit portion is implemented. In this way, an output logic state is maintained so long as the value of the Boolean operation being performed by the logic device does not change. Additionally, static logic elements may perform the inversions necessary to output both logic senses, mitigating the need to provide for dual-rail dynamic logic implementations. An asymmetric clock permits a concomitant decrease in the size of the precharge transistors thus ameliorating the area required by the logic element and obviating a need for keeper device.

Abstract: A voltage controlled oscillator (VCO) is constructed using a series ring connection of an odd number K of logic inverters where K is greater than three. Each sequence of three of the logic inverters has voltage controlled feed-forward conduction circuit coupled in parallel. Each of the feed-forward circuits has the same phase between its input and output as the path it parallels. The control voltage of the feed-forward circuits operates to decrease the path delay of the logic inverters when they are conducting. Selectable inverters are connected in parallel with each logic inverter using a P and an N channel field effect transistor (FET). The N channel FET is controlled with a Mode signal and the P channel FET is controlled by a Modeb signal which is generated by inverting the Mode signal. The Mode and Modeb signals control the connection of the selectable inverters are in parallel with the logic inverters thus increasing the drive capability of the parallel combination of inverters.

Abstract: Selector circuits and systems for single and multilevel selection within one clock cycle having a static switching factor on the output of a dynamic logic circuit. A logic device for single and multilevel selection having a dynamic logic circuit portion and a static logic circuit portion is implemented. In this way, an output logic state is maintained so long as the value of the Boolean operation being performed by the logic device does not change. Additionally, static logic elements may perform the inversions necessary to output both logic senses, mitigating the need to provide for dual-rail dynamic logic implementations. An asymmetric clock permits a concomitant decrease in the size of the precharge transistors thus ameliorating the area required by the logic element and obviating a need for keeper device.

Abstract: Circuits and systems for producing a static switching factor on the output lines of dynamic logic devices. A logic device having a plurality of dynamic logic circuits each performing a Boolean function on a plurality of inputs and generating an output on a dynamic node. The corresponding plurality of dynamic outputs are coupled to a static logic circuit which performs an additional Boolean function of the plurality of dynamic outputs. The static logic circuit operates to generate an output logic state that is maintained so long as the value of the Boolean operations being performed by the logic device do not change. Additionally, static logic elements may perform the inversions necessary to output both logic senses, mitigating the need to provide for dual-rail dynamic logic implementations. An asymmetric clock permits a concomitant decrease in the size of the precharge transistors, thus ameliorating the area required by the logic element.

Abstract: A method and apparatus performing rotate operations using cascaded multiplexers provides a scalable rotator circuit having a sub-field rotate capability that requires no additional interconnects at the sub-field endpoints. The rotator performs bit field swap operations at each stage of a series of cascaded multiplexers. The bit field size increases monotonically from a single bit to half of the rotator operand size. The control logic selects swap operations for each individual bit field at each stage, in order to arrange a desired rotated output vector.

Abstract: A voltage controlled oscillator (VCO) is constructed using a series ring connection of an odd number K of logic inverters where K is greater than three. Each sequence of three of the logic inverters has voltage controlled feed-forward conduction circuit coupled in parallel. Each of the feed-forward circuits has the same phase between its input and output as the path it parallels. The control voltage of the feed-forward circuits operates to decrease the path delay of the logic inverters when they are conducting. Selectable inverters are connected in parallel with each logic inverter using a P and an N channel field effect transistor (FET). The N channel FET is controlled with a Mode signal and the P channel FET is controlled by a Modeb signal which is generated by inverting the Mode signal. The Mode and Modeb signals control the connection of the selectable inverters are in parallel with the logic inverters thus increasing the drive capability of the parallel combination of inverters.

Abstract: A charge pump has two charge pump nodes. The first charge pump node has a first current source (CS) with a source terminal connected to a positive supply voltage and an output terminal connected to the first charge pump node with a P channel metal oxide silicon transistor (PFET) controlled by a first control signal. The first charge pump node is also connected to a second CS with a source terminal connected to the ground supply voltage and an output terminal connected to the second CS with an NFET controlled by a second control signal. The second charge pump node has a third CS with a source terminal connected to the positive supply voltage and an output terminal connected to the second charge pump node with a PFET controlled by a third control signal. The second charge pump node is also connected to a fourth CS with a source terminal connected to the ground supply voltage and an output terminal connected to the second CS with an NFET controlled by a fourth control signal.

Abstract: A reference signal and a voltage controlled oscillator (VCO) output are compared for relative phase and frequency differences. A lead error signal is generated if the reference signal leads the VCO output and a lag error signal is generated if the reference signal lags the VCO output the lead and lag error may result from a combination for phase and frequency differences between the reference signal and the VCO output. A time window is used to sample the polarity of the lead and lag error signals by incrementing and decrementing a phase error signal. If the phase error signal reaches a threshold value within the time window, a Reset Delta pulse is generated and if the phase error signals does not reach the maximum delta value within the time window a Reset Total pulse is generated. A variable first gain signal is increased on each Reset Delta pulse and decreased on each Reset Total pulse and limited to a value between predetermined maximum and minimum values.

Abstract: A phase locked loop (PLL) circuit uses a programmable frequency divider (PRFD) to generate a feedback clock from the PLL output clock. The PLL power supply voltage and a PLL reference current are generated by regulating the scalable logic supply voltage of the system in using regulator circuits. The PLL power supply voltage is regulated to a level lower than the lowest level of the scalable logic supply voltage used by the system. The PLL generates a PLL output clock whose frequency is higher than the highest frequency of operation of the system using the highest level of the scalable logic power supply voltage. The PLL output clock is divided is a second PRFD to generate a divided PLL clock. The PLL clock and a fixed auxiliary clock are selected in a glitch-free multiplexer (MUX) as the system clock for the system. The system clock frequency may be dynamically scaled by programming the divisor in the second PRFD dividing the PLL clock.

Abstract: A glitch-free clock selector selects between asynchronous clock signals. In one embodiment a select signal has two logic states corresponding to the two clock signals. A clock output signal is gated with a latched compare signal which compares a new select signal state to a stored current select signal state. A multiplexer (MUX) selects between the two clock signals in response to a select latch output signal. If the new and current select signals do not compare the clock output signal is forced to a logic zero by the output of a compare latch which latches the compare signal when the MUX output (present selected clock signal) goes to a logic zero. While the present clock signal is held low, the MUX switches to the new clock signal. The new clock signal (MUX output) latches the new select state as the current select state causing the new and current select signal to compare.

Abstract: A clock divider circuit receives a first clock signal and frequency divides the first clock signal to generate a second clock signal. The first and second clocks are inputs to multiplexer (MUX) that selects one of the clocks as the MUX output based on the state of a MUX control signal. The MUX output is combined with a latched Freeze clock signal in an AND gate to generate a clock output signal. The state of Bypass logic signal determines which clock signal is selected as the clock output signal. The Bypass logic signal is received in a latch circuit which latches the Bypass logic signal in two series latches one latch latching on the positive edge and one on the negative edge of the MUX output. The output of the second latch is latched in a third latch on when a NOR logic gate signals the first and second clock are concurrently at a logic zero assuring that the MUX output is changed only when both clocks are low. The clock output signal is gated with the latched Freeze clock signal in the AND logic gate.

Abstract: An apparatus for performing ADD and ROTATE as a single instruction within a processor is disclosed. In accordance with a preferred embodiment of the present invention, the apparatus comprises an adder and a rotator. The adder is utilized for adding a first number to a second number in a multiple stages to yield a carry-out and a sum output. During each of these stages, the adder produces a group generate value and a group propagate value. The rotator is utilized for rotating the group propagate value and the group generate value at each of the stages before the yielding of the carry-out and the sum output. As such, both ADD and ROTATE instructions can be completed within a single processor cycle.

Abstract: A system implementing a methodology for determining the exponent in parallel with determining the fractional shift during normalization according to partitioning the exponent into partial exponent groups according to the fractional shift data flow, determining all possible partial exponent values for each partial exponent group according to the fractional data flow, and providing the exponent by selectively combining possible partial exponents from each partial exponent group according to the fractional data flow. There is also provided a system implementing a methodology for generating the sticky bit during normalization. Sticky bit information is precalculated and multiplexed according to the fractional dataflow. In an embodiment of the invention, group sticky signals are calculated in tree form, each group sticky having a number of possible sticky bits corresponding to the shift increment amount of the multiplexing.

Abstract: In a multiprocessor system, a plurality of data processors are each equipped with a local, level 1, cache and have access to a main memory through memory access circuit having a level 2 cache and a single register arbiter. The single register includes a primary queue defining priority of requests from the plurality of processors and a secondary queue defining processor requiring access to main memory. The register contains one position for each of the processors served and employs a pointer for demarcation between the primary and secondary queues. When a request is detected, the highest priority processor in the primary queue is served and when the requested memory address is in the level 2 cache, it will be retrieved and the identity of the served processor will be moved to the low end of the primary queue as defined by the pointer.