Lsi Corporation

Lsi Logic Corporation

Avago Technologies General IP (singapore) Pte. Ltd.

Lsi Logic Corporatiion

Jeffrey Scott Brown

Mark Franklin Turner

Jonathan W Byrn

Chang Ho Jung

Craig R Chafin

Marek J Marasch

Jay Daugherty

Jay D Harker

A method for enhancing signal integrity in an interface between a source device and at least one destination device includes: analyzing two or more consecutive data patterns intended to be conveyed by the interface to determine whether data transitions corresponding to the data patterns are likely to introduce coupling noise and/or simultaneous switching output (SSO) effects on the interface; generating a modified data pattern for transmission by the interface, the modified data pattern reducing coupling noise and/or SSO effects on the interface compared to an original data pattern intended to be conveyed by the interface; and transmitting the modified data pattern and information regarding a manner in which the original data pattern was modified to the destination device to thereby reduce coupling noise and/or SSO effects on the interface.

Parallel interface pattern modification for addressing signal integrity concerns

Disclosed is a technique for providing minimal sequential overhead in a flip-flop circuit. Equalization of setup times is achieved in one embodiment. In addition, delays in clock to Q can be equalized for both rising data transitions and falling data transitions. Large setup times are not required since optimization techniques equalize setup times for both rising and falling data transitions.

Optimization of library slew ratio based circuit

An apparatus for monitoring at least supply voltage in an IC includes a plurality of monitor circuits distributed throughout the integrated circuit. Each of the monitor circuits is operative to receive the supply voltage, or a signal representative thereof, and to generate an output signal indicative of a comparison between the supply voltage and a reference voltage. The apparatus further includes a control circuit coupled to the plurality of monitor circuits. The control circuit is operative to receive the respective output signals from the plurality of monitor circuits and to generate an output of the apparatus which is a function of information conveyed in the respective output signals from the plurality of monitor circuits.

On-chip power supply monitoring using a network of modulators

An apparatus for controlling clock skew in an integrated circuit (IC) includes timing circuitry operative to generate a clock signal for distribution in the IC and at least one buffer circuit operative to receive the clock signal, or a signal indicative of the clock signal, and to generate a delayed version of the clock signal as an output thereof. The buffer circuit includes at least first and second inverter stages and a resistive-capacitive (RC) loading structure. An output of the first inverter stage is connected to an input of the second inverter stage via the RC loading structure. The buffer circuit has a delay associated therewith that is selectively varied as a function of one or more adjustable characteristics of the RC loading structure. Clock skew in the IC is controlled as a function of the delay of the buffer circuit.

Fine-grained clock skew tuning in an integrated circuit

A global power distribution network in an integrated circuit comprising a first layer of conductive material and a second layer of conductive material. The first layer of conductive material may be (i) coupled to one or more power supplies and (ii) configured to form a plurality of first rails of a mesh. The first rails may (a) supply power to one or more components of a core logic of the integrated circuit, (b) be aligned with a first axis of the integrated circuit, and (c) have one or more parameters configured such that the mesh has a uniform voltage gradient from a perimeter of the integrated circuit to a center of the integrated circuit along the first axis. The second layer of conductive material may be (i) coupled to the one or more power supplies and (ii) configured to form a plurality of second rails of the mesh. The second rails may (a) supply power to one or more components of the core logic, (b) be aligned with a second axis of the integrated circuit, and (c) have one or more parameters configured such that the mesh comprises a uniform voltage gradient from the perimeter of the integrated circuit to the center of the integrated circuit along the second axis.

Power grid optimization

Disclosed is a closed-loop feedback system for controlling the soft error rate (SER) due to radiation strikes on electronic circuitry. A variable sensitivity soft error rate detector provides and output corresponding to the soft error rate. This output is supplied to a voltage control. The output of the voltage control is fed back to the sensitivity control of the sensor&#x2014;thus forming a feedback loop. The output of the voltage control may be the power supply of the soft error rate sensor. The output of the soft error rate sensor may also be used to enable and disable fault tolerant schemes or alert a user.

Closed-loop soft error rate sensitivity control

A storage medium recording a cell library having one or more cells that may be readable by a computer and may be used by the computer to design an integrated circuit. The one or more cells may have a physical dimension parameter and a channel width parameter. The physical dimension parameter may be a footprint of the one or more cells. The channel width parameter may have a minimum driver size and a maximum driver size. The channel width parameter may define a range within which a tool varies the channel width between the maximum driver size and the minimum driver size during a design flow of the integrated circuit based upon one or more power criteria without changing the footprint.

Granular channel width for power optimization

An apparatus and method for providing a read-only memory (ROM) bit cell having one each of a PMOS transistor and an NMOS transistor, which has reduced static and dynamic electric power losses, are described. In particular, the bit cell does not require a pre-charge transistor. The sense amplifier for determining the voltages on ROM bit lines may be a digital inverter, address decoding may be simplified since there are no timing requirements with respect to transistor pre-charge, and chips containing a plurality of ROM bit cell may be readily programmed. In one embodiment of the invention, each bit cell includes one PMOS transistor having its source in electrical connection with a voltage source, its drain connected or unconnected to a bit line, and its gate connected to an inverted version of the word line signal; and one NMOS transistor having its source connected to a lower voltage source, its drain connected or disconnected to the bit line, and its gate connected to the word line. The bit cell is programmed during the ROM generation by connecting the drain of either the PMOS (logic level 1) or the NMOS (logic level 0) to the bit line.

Transistor bit cell ROM architecture

A method for conserving power in a device. The method generally includes the steps of (A) generating a polarity signal by analyzing a current one of a plurality of data items having a plurality of data bits, the polarity signal having an inversion bit indicating that the current data item is to be stored in one of (i) an inverted condition and (ii) a non-inverted condition relative to a normal condition such that a majority of the data bits have a first logic state, wherein reading one of the data bits having the first logic state consumes less power than reading one of the data bits having a second logic state, (B) selectively either (i) inverting the current data item or (ii) not inverting current the data item based on the inversion bit and (C) storing the current data item in a plurality of single-ended bit cells in the device.

Energy efficient memory access technique for single ended bit cells

A method to analyze timing in a circuit, generally including (A) simulating reception of an input signal and a clock signal at a first flip-flop, wherein (i) the input signal has a latest transition, (ii) the input signal arrives through a first path and (iii) the clock signal has an active edge, (B) calculating a value of a time difference between the latest transition and the active edge, (C) calculating a delay between the active edge and the latest transition appearing in an output signal, wherein (i) the delay is based on a model responding to the value, (ii) the model characterizes a clock-to-output delay as a function of the time difference and (iii) the characterization covering a range of values, (D) calculating an arrival time of the latest transition at a second flip-flop through a second signal path and (E) storing the arrival time in a recording medium.

Dual path static timing analysis

A method to redistribute current demand is presented. The method includes a first step of determining timing arc data for one or more timing arcs of a circuit design. The method includes a second step of checking the timing arc data for delay shift target cells. The method includes a further step of swapping a delay shift target cell with a delay shift cell.

Redistribution of current demand and reduction of power and DCAP

A memory device using a plurality of enhanced row decode drivers for activating wordlines in a memory array is disclosed. Circuit design attributes of the enhanced row decode drivers are varied as a function of proximity to a source of a row address signal applied to each decode driver. The circuit variations are operable to reduce the leakage power of the driver by degrading performance thereof while maintaining required worst case timing. The worst case timing being defined by the timing and performance requirements for the most distant of the row decode driver circuits relative to the source of the applied row address signals.

Row decode driver gradient design in a memory device

A self-timed memory array is disclosed, in which segmentability and metal-programmability are supported while minimizing layout space. Self-timing row decoder circuits are placed at the top and bottom of the array adjacent to respective I/O blocks. A self-timing signal is routed from the top (resp. bottom) of the array to a point halfway down (resp. up) the memory array and then back to a self-timing row decoder at the top (resp. bottom) of the array. The same approach may also be used to account for the bitline wire delay from the bottom (resp. top) of the array to the sense amplifiers in the I/O block. Further flexibility in wire routing is provided by eliminating metal routing layers from unneeded memory cells, and a programmable gate array may be used to allow an arbitrary word size to be chosen for the memory.

Metal programmable self-timed memories

A method of testing a tristate element by applying a given value to the tristate, applying an opposite value to a keeper element connected at an output of the tristate, capturing a first value at a downstream position of the tristate, evaluating a second value at the output of the tristate using the first value, comparing the second value to the opposite value, and producing a failure code for the tristate when the second value is not equal to the opposite value. Then, applying the opposite value to the tristate, applying the given value to the keeper element, capturing the first value, evaluating the second value using the first value, comparing the second value to the given value, and producing a failure code for the tristate when the second value is not equal to the given value. A passing code for the tristate is produced when a failure code has not been produced.

Growing community of inventors

Fort Collins, CO, United States of America

Jeffrey Scott Brown