Abstract: A voltage source droop compensated clock modulation for microprocessors is described. Specifically, the circuit reduces the clock frequency if a voltage source droop is detected.

Abstract: A voltage source droop compensated clock modulation for microprocessors is described. Specifically, the circuit reduces the clock frequency if a voltage source droop is detected.

Abstract: An apparatus for managing the temperature of an integrated circuit having a multiple core microprocessor is described. Specifically, thermal sensors are placed at potential hot spots throughout each microprocessor core. A thermal management unit monitors the thermal sensors. If a thermal sensor identifies a hot spot, the thermal management unit adjusts the operating frequency and voltage of that microprocessor core accordingly.

Abstract: An apparatus for managing the temperature of an integrated circuit having a multiple core microprocessor is described. Specifically, thermal sensors are placed at potential hot spots throughout each microprocessor core. A thermal management unit monitors the thermal sensors. If a thermal sensor identifies a hot spot, the thermal management unit adjusts the operating frequency and voltage of that microprocessor core accordingly.

Abstract: A method for dynamically varying a clock frequency in a processor. The method of one embodiment comprises driving a clock distribution network with a clock output from a phased locked loop (PLL). An adjustable clock generator is locked with the phased locked loop. The adjustable clock generator is substituted for the PLL on the clock distribution network.

Abstract: A method and an apparatus for dynamically varying a clock frequency in a processor to adapt to VCC voltage changes. The method of one embodiment includes sampling a supply voltage at a plurality of locations. The values of said supply voltage are communicated to a clock generator. A clock frequency of a clock signal generated from the clock generator is adjusted in response to an evaluation of the sampled values of the supply voltage.

Abstract: An apparatus for managing the temperature of an integrated circuit having a multiple core microprocessor is described. Specifically, thermal sensors are placed at potential hot spots throughout each microprocessor core. A thermal management unit monitors the thermal sensors. If a thermal sensor identifies a hot spot, the thermal management unit adjusts the operating frequency and voltage of that microprocessor core accordingly.

Abstract: A method for dynamically varying a clock frequency in a processor to adapt to VCC voltage changes. The method of one embodiment comprises sampling a supply voltage at a plurality of locations. The values of said supply voltage are communicated to a clock generator. A clock frequency of a clock signal generated from the clock generator is adjusted in response to an evaluation of the sampled values of the supply voltage.

Abstract: A method for dynamically varying a clock frequency in a processor. The method of one embodiment comprises driving a clock distribution network with a clock output from a phased locked loop (PLL). An adjustable clock generator is locked with the phased locked loop. The adjustable clock generator is substituted for the PLL on the clock distribution network.

Abstract: A method for dynamically varying a clock frequency in a processor. The method of one embodiment comprises driving a clock distribution network with a clock output from a phased locked loop (PLL). An adjustable clock generator is locked with the phased locked loop. The adjustable clock generator is substituted for the PLL on the clock distribution network.

Abstract: A method for dynamically varying a clock frequency in a processor. The method of one embodiment comprises driving a clock distribution network with a clock output from a phased locked loop (PLL). An adjustable clock generator is locked with the phased locked loop. The adjustable clock generator is substituted for the PLL on the clock distribution network.

Abstract: An on-die clock generator. For one aspect of the invention, the on-die clock generator includes a phase-locked loop (PLL) circuit having a first input coupled to receive an external clock signal and an output coupled to provide an on-die clock signal to be used during a normal operating mode of an integrated circuit. The on-die clock generator also includes a local clock generator circuit having an input coupled to receive the on-die clock signal and an output coupled to provide a local PLL feedback clock signal to a second input of the PLL.

Abstract: A multiple synthesizer based timing signal generation scheme is described that allows accurate data and strobe generation in high speed source synchronous system interfaces. Multiple loop locked clock synthesizers (e.g., phase locked loops, delay locked loops) are used to generate multiple clock signals. Data and strobe signals are triggered off of transitions of one of the clock signals. Because multiple loop locked clock synthesizers are used to generate the clock signals, optimal or near optimal alignment of the data and strobe signals can be achieved. Improved alignment of the data and strobe signals provides improved data transmission rates.

Abstract: The invention comprises, in one aspect, a content addressable memory array having a plurality of memory locations to store tag words. The content addressable memory array includes a parity encoder and a parity comparator. The parity encoder has a first input terminal to receive an input data signal and a first output terminal to deliver a signal representative of the parity of the input data signal. The parity comparator has a second input terminal, a third input terminal connected to the first output terminal, and a plurality of memory cells to store original parities of the tag words. The parity comparator compares the original parity of a first tag word to the parity of the input data signal in response to a receiving a match signal. The content addressable memory array includes a fourth input terminal to receive the input data signal, and a second output terminal to send the match signal in response to one of the tag words matching the input data signal.

Abstract: A method for increasing the accuracy of an analog neural network which computers a sum-of-products between an input vector and a stored weight pattern is described. In one embodiment of the present invention, the method comprises initially training the network by programming the synapses with a certain weight pattern. The training may be carried out using any standard learning algorithm. Preferably, a back-propagation learning algorithm is employed.Next, the network is baked at an elevated temperature to effectuate a change in the weight pattern previously programmed during initial training. This change results from a charge redistribution which occurs within each of the synapses of the network. After baking, the network is then retrained to compensate for the change resulting from the charge redistribution. The baking and retraining steps may be successively repeated to increase the accuracy of the neural network to any desired level.

Abstract: A difference calculating neural network is disclosed having an array of synapse cells arranged in rows and columns along pairs of row and column lines. The cells include a pair of floating gate devices which have their control gates coupled to receive one of a pair of complementary input voltages. The floating gate devices also have complementary threshold voltages such that packets of charge are produced from the synapse cells that are proportional to the difference between the input and voltage threshold. The charge packets are accumulated by the pairs of column lines in the array.

Abstract: In an apparatus for multiplexed operation of multi-cell neural network, the reference vector component values are stored as differential values in pairs of floating gate transistors. A long-tail pair differential transconductance multiplier is synthesized by selectively using the floating gate transistor pairs as the current source. Appropriate transistor pairs are multiplexed into the network for forming a differential output current representative of the product of the input vector component applied to the differential input and the stored reference vector component stored in the multiplexed transistor pair that is switched into the multiplier network to function as the differential current source. Pipelining and output multiplexing is also described in other preferred embodiments for increasing the effective output bandwidth of the network.

Abstract: A method for adaptively setting analog weights in analog cells of a neural network and the like. The process starts by addressing a synapse cell in the network. A target weight for said addressed synapse cell is selected, and the current weight present on the synapse cell is measured. The amplitude and duration of a voltage pulse to be applied to said synapse cell to adjust said synapse cell in the direction of said target weight is calculated using a set of coefficients representing the the physical characteristics of the synapse cell. The voltage pulse is applied to the addressed synapse cell and the new weight of the synapse cell is re-measured. If the synapse cell weight is within acceptable limits of the target weight, the values of the coefficients are saved and the next adjacent synapse cell is addressed until all synapse cells are set. If the synapse cell is not within acceptable limits, new values for the coefficients are calculated in relation to the re-measured weight.

Abstract: A method for increasing the accuracy of an analog neural network which computers a sum-of-products between an input vector and a stored weight pattern is described. In one embodiment of the present invention, the method comprises initially training the network by programming the synapses with a certain weight pattern. The training may be carried out using any standard learning algorithm. Preferably, a back-propagation learning algorithm is employed.Next, network is baked at an elevated temperature to effectuate a change in the weight pattern previously programmed during initial training. This change results from a charge redistribution which occurs within each of the synapses of the network. After baking, the network is then retrained to compensate for the change resulting from the charge redistribution. The baking and retraining steps may be successively repeated to increase the accuracy of the neural network to any desired level.

Abstract: A multi-layer electrically trainable analog neural network employing multiplexed output neurons having inputs organized into two groups, external and recurrent (i.e., feedback). Each layer of the network comprises a matrix of synapse cells which implement a matrix multiplication between an input vector and a weight matrix. In normal operation, an external input vector coupled to the first synaptic array generates a Sigmoid response at the output of a set of neurons. This output is then fed back to the next and subsequent layers of the network as a recurrent input vector. The output of second layer processing is generated by the same neurons used in first layer processing. Thus, the neural network of the present invention can handle N-layer operation by using recurrent connections and a single set of multiplexed output neurons.