摘要:
A method for noise comprising synthesizing blocks of sequential latches, e.g., a pipeline circuit architecture or clocking domain, which comprises combinational logic, synthesizing a root or a master clock and at least one phase-shifted sub-domain clock for each block, assigning primary inputs and primary outputs of the block to the root clock, assigning non-primary inputs and non-primary outputs of the block to the sub-domain clock, splitting root clock inputs into root clock inputs and phase-shifted sub-domain clock inputs, assigning each of the blocks a different phase-shifted sub-domain clock phase offset, creating a clock generation circuitry for the root clocks and the phase-shifted sub-domain clocks.
摘要:
A fast/slow state machine latch is provided that generates fast and slow select signals for a single toggle, low power multiplexer circuit. In accordance with an embodiment of the present invention, the fast/slow state machine latch includes a first latch with a delayed output, a second latch with an undelayed output, an inverter for coupling the delayed output of the first latch to an input of the second latch, and an exclusive-OR (XOR) gate coupled to the delayed output of the first latch and a data input, an output of the XOR gate coupled to an input of the first latch. A method for incorporating low power multiplexer circuits into a circuit design with minimal input from a circuit designer is also provided.
摘要:
An integrated circuit (IC) chip (100) containing a plurality of voltage islands (124I-M) containing corresponding functional blocks (104I-M) that can be selectively fenced, i.e., powered down, while saving the states of the corresponding inputs, and unfenced in order to manage power consumption of the chip. Each fencable functional block includes a power switch (140I-M) and state-saving circuitry (148I-M) for saving the state of the inputs to that functional block. A power modulation unit (PMU) (132) generates fencing signals (144I-M) that control the power switches and state-saving circuitries so as to selectively fence the corresponding functional blocks. The PMU generates the fencing signals as a function of one or more operating arguments.
摘要:
In a first aspect, a first method is provided for allocating memory bandwidth. The first method includes the steps of (1) assigning a fixed priority of access to the memory bandwidth to one or more direct memory access (DMA) machines; and (2) assigning a programmable priority of access to the memory bandwidth to a processing unit. The programmable priority of the processing unit allows priority allocation between the one or more DMA machines and the processing unit to be adjusted dynamically. Numerous other aspects are provided.
摘要:
A structure and associated method of processing data on a semi-conductor device comprising an input island, a processing island, and an output island formed on the semiconductor device. The input island is adapted to accept a specified amount of data and enable a means for providing a first specified voltage for powering the processing island after accepting the specified amount of data. The processing island is adapted to receive and process the specified amount of data from the input island upon powering the processing island by the first specified voltage. The output island is adapted to be powered by a second specified voltage. The processing island is further adapted to transmit the processed data to the output island upon said powering by the second specified voltage. The first specified voltage is adapted to be disabled thereby removing power from processing island upon completion of transmission of the processed data to the output island.
摘要:
An integrated circuit, method and system providing finer granularity dynamic voltage control without performance loss. The invention provides a means for dynamically changing a voltage level of at least one stage on a critical path for a particular cycle. In this way, optimum voltages can be provided to the stages for the given expectation.
摘要:
A method for reducing leakage power of a logic network comprising the steps of: using (observability) don't care information to identify “sleep states” for individual nets; determining based on probabilistic analysis at least one net in which expected power consumption will be reduced by forcing a net to a particular value during at least a portion of a “sleep state”; and forcing the determined net to the determined value determined portion of that “sleep state”.
摘要:
Redundant chip sections held in standby are substituted for chip sections that are at risk of over heating based on certain sensor signals. When these signals are received operations of the chip section at risk IS transferred to a redundant chip section and the chip section at risk is shut down. After the original chip section has cooled, it becomes available as a replacement chip section itself. The sensor signals may be based on temperature values, elapsed operation time, and number or rate of operations within a chip section.
摘要:
A multi-port memory is provided that includes means for receiving synchronous memory requests, means for receiving asynchronous memory requests, and means for processing the received synchronous and asynchronous memory requests simultaneously. Systems and methods that employ the multi-port memory are also provided.
摘要:
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 toggling that commonly occurs in many types of logic circuits. The preferred embodiment reduces unneeded node toggling in a circuit by holding a portion of the device at the previous output until the all the inputs have stabilized to their final value during each clock cycle. This reduces power consumption in the device that would normally occur due to unnecessary node toggling.