Aptix Corporation

Sequence Design, Inc.

Lightspeed Semiconductor Corporation

Gradient Design Automation Inc.

Adi Srinivasan

Ta-Pen Guo

Dana How

Rajit Chandra

Abbas A El Gamal

Shridhar K Mukund

David L Allen

John Yanjiang Shu

Paolo Carnevali

Robert Osann, Jr

Abbas El Gamal

Nanda Gopal

Hong Cai

Ta-Pan Guo

Robert Osann

A thermally aware design automation suite integrates system-level thermal awareness into design of semiconductor chips, performing fine-grain static and/or transient thermal simulations of the chips based on thermal models and boundary conditions. The thermal simulations are performed in accordance with one or more grids, with boundaries and/or resolutions being determined by adaptive and/or hierarchical multi-dimensional techniques. The adaptive grid techniques include material-boundary, rate-of-change, and convergence-information heuristics. For example, a finer grid is used in a region having higher temperature gradients compared to a region having lower temperature gradients. The hierarchical grid techniques are based on critical, intermediate, and boundary regions specified manually or automatically, each region having a respective grid resolution. For example, a critical region is analyzed according to a grid that is finer than a grid of an intermediate region, and resolution of a grid of a boundary region is adapted to boundary conditions.

Thermal simulation using adaptive 3D and hierarchical grid mechanisms

Transient thermal simulation of semiconductor chips uses region-wise variable spatial grids and variable temporal intervals, enabling spatio-temporal thermal analysis of semiconductor chips. Temperature rates of change across a die and/or package of an integrated circuit are computed and tracked versus time. Critical time interval(s) for temperature evaluation are determined. Temperatures of elements, components, devices, and interconnects are updated based on a 3D full chip temperature analysis. Respective power dissipations are updated, as a function of the temperatures, with an automated interface to one or more circuit simulation tools. Subsequently new temperatures are determined as a function of the power dissipations. User definable control and observation parameters enable flexible and efficient transient thermal analysis. The parameters relate to power sources, monitoring, reporting, error tolerances, and output snapshots. Viewing of waveform plots and 3D spatial variations of temperature enable efficient communication of results of the thermal analysis with designers of integrated circuits.

Transient thermal analysis

In a first variation, a thermally aware design automation suite integrates system-level thermal awareness into design of semiconductor chips, performing fine-grain thermal simulations of the chips based on thermal models and boundary conditions. The suite uses results of the simulations to modify thermally significant structures to achieve desired thermal variations across a chip, meet design assertions on selected portions of the chip, and verify overall performance and reliability of the chip over designated operating ranges and manufacturing variations. In a second variation, a discretization approach models chip temperature distributions using heuristics to adaptively grid space in three dimensions. Adaptive and locally variable grid spacing techniques are used to efficiently and accurately converge for steady state and/or transient temperature solutions. The modeling optionally reads a mesh initialization file specifying selected aspects and parameters associated with controlling use and behavior of the variable grid spacing techniques.

Thermally aware design modification

A method and apparatus for using full-chip thermal analysis of semiconductor chip designs to compute thermal conductivity is disclosed. One embodiment of a novel method for analyzing the conductivity of a semiconductor chip design that comprises a plurality of physical layers includes defining at least one thermal layer within the plurality of physical layers, where the thermal layer(s) represents a variance in thermal conductivity relative to a remainder of the semiconductor chip design, and computing a thermal conductivity of the thermal layer(s). As the thermal layer(s) represents variances in thermal conductivity over the semiconductor chip design, the thermal layer(s) does not necessarily correspond one-to-one to the physical layers of the semiconductor chip design. Thus, the thermal conductivities within the semiconductor chip design can be computed from the thermal layers.

Method and apparatus for using full-chip thermal analysis of semiconductor chip designs to compute thermal conductance

A method is provided to optimize delay insertions for reducing timing violations. The method includes inserting a buffer between a driver and a receiver in a timing path and placing the buffer either inside or outside a bounding box that encloses the driver and the receiver. The placement of the buffer inside or outside the bounding box creates the appropriate effective loading on the buffer to generates the required minimum delay to avoid timing violations.

Circuit optimization for minimum path timing violations

A method for optimal driver selection uses a cost function that is based on the non-linear delay characteristics and the stage gain of the candidate drivers. The cost function operates to select an optimal driver for driving the predetermined capacitive load which. simultaneously minimizes the delay and the amount of input capacitance introduced. In one embodiment, a method for selecting a driver for driving a load capacitance from a group of drivers includes: computing for each driver a cost based on a cost function associated with the driver, and selecting the driver having the smallest cost. The cost function is directly proportional to a delay of the driver and inversely proportional to the logarithm of a stage gain of the driver. In another embodiment, the stage gain is an output capacitance driven by the driver (the load capacitance) divided by an input capacitance of the driver.

Method for determining load capacitance

A method for forming an application specific integrated circuit, comprises receiving a circuit design for the application specific integrated circuit from a designer; performing an initial place and route layout of the circuit design which leaves a group of buffer modules unused, based upon a partially predesigned integrated circuit, in which the partially predesigned integrated circuit includes a plurality of logic modules and a plurality of buffer modules uniformly distributed amongst the logic modules; evaluating load and timing characteristics for the initial place and route layout of the circuit design; and integrating buffer modules from the group of unused buffer modules into the circuit design, based on the load and timing characteristics evaluated. A gate array, for forming the application specific integrated circuit in accordance with the invention includes a matrix of function blocks capable of being configured to implement combinational, sequential, and memory modes of operation, as well as providing tri-state drivers and buffers in useful numbers.

Function block architecture for gate array and method for forming an asic

A method for optimal driver selection uses a cost function that is based on the non-linear delay characteristics and the stage gain of the candidate drivers. The cost function operates to select an optimal driver for driving the predetermined capacitive load which simultaneously minimizes the delay and the amount of input capacitance introduced. In one embodiment, a method for selecting a driver for driving a load capacitance from a group of drivers includes: computing for each driver a cost based on a cost function associated with the driver, and selecting the driver having the smallest cost. The cost function is directly proportional to a delay of the driver and inversely proportional to the logarithm of a stage gain of the driver. In another embodiment, the stage gain is an output capacitance driven by the driver (the load capacitance) divided by an input capacitance of the driver.

Method for optimal driver selection

A method for computing a position for a zero-skew driver insertion point in an area occupied by nodes driven by the driver is described. The zero-skew driver insertion point is the position in the area where the spread of the signal arrival times at the nodes driven by the driver is minimized. The method includes: expressing a function describing a distance from each of the nodes to the zero-skew driver insertion point, expressing the variance of the function, minimizing the variance of the function, and solving an equation representative of the minimization of the variance of the function to determine the position of the zero-skew driver insertion point. In one embodiment, the minimizing the variance of the function includes: taking a first derivative of the function with respect to the distance, and setting the first derivative of the function to zero.

Method for determining a zero-skew buffer insertion point

A method for &ldquo;match-delay&rdquo; buffer insertion is provided to add delays at a node without changing the input capacitance of the node as seen by the upstream node. In one embodiment, a method for inserting a delay in a node in an electrical design associated with a logic gate includes: adding the delay at the node by adding a new logic gate before the node where the new logic gate is the same cell type as the logic gate and is positioned near the logic gate. The method may further include: determining if the delay can be added by adding a new logic gate before the node, and if a new logic gate cannot be added before the node, adding the delay by adding a new logic gate after the logic gate where a combination of the logic gate and the new logic gate giving the delay to be added.

Method for match delay buffer insertion

A clock tree insertion method for distributing a clock signal in an integrated circuit design includes providing a physical design representative of the integrated circuit design, specifying a location for a root node of the clock tree in the physical design, constructing an array of buffers as the clock tree where the array of buffers is constructed to minimize the maximum insertion delay from the root node to the clock signal endpoints and to meet a predefined maximum insertion delay constraint, identifying locations in the clock tree where clock skew violations occur and correcting the clock skew violations by introducing delay at buffer locations in the clock tree having the fastest clock signal arrival times, and identifying locations in the clock tree where minimum insertion delay violations occur and correcting the minimum insertion delay violations by slowing down the arrival times of clock signal endpoints of the clock tree.

Method for balanced-delay clock tree insertion

A gate array in accordance with the invention includes a matrix of function blocks capable of being configured to implement combinational, sequential, and memory modes of operation, as well as providing tri-state drivers and buffers in useful numbers. The function block includes a logic circuit with a first bit storage unit, which is selectively configurable to behave as combinational logic or to store a first bit, and a second bit storage unit, which is also selectively configurable to behave as combinational logic or to store a second bit. The matrix of function blocks in accordance with the invention is also useful to properly distribute clocks throughout the gate array.

Function block architecture for gate array

A system for testing an integrated circuit, and particularly a gate array, is disclosed which includes, prior to coupling the array to form a user-designed circuit, predesigned logic that enables testing of the user-designed circuit. The predesigned logic allows logic blocks in the array to operate in &ldquo;freeze&rdquo; mode or to operate in normal mode, where normal mode is defined by the user-designed circuit. When the logic blocks are selected to be frozen, the logic blocks behave as a series of daisy-chained master-slave flip-flops. In normal mode, a logic block can implement combinational, sequential, or other functions and still later be as a master-slave flip-flop. Moreover, each logic block is further equipped for addressable mode control, allowing selected logic blocks to be exercised in isolation once stimulus data is shifted in, simplifying test generation and improving fault coverage.

Method and apparatus for controlling and observing data in a logic block-based ASIC

A customizable ASIC routing architecture is provided. The architecture utilizes the uppermost metal layers of an ASIC composed of an array of function blocks for routing among function blocks while lower layers are used for local interconnections within the function blocks. The second-to-uppermost metal layer is fixed and generally includes a plurality of parallel segmented conductors extending in a first direction. The uppermost metal layer is customizable in a predesignated manner. Metal in the uppermost metal layer is selectively placed in tracks, which are substantially perpendicular to the segmented conductors in the layer below. Vias are provided between the two uppermost layers. One embodiment of the invention permits one-mask customization of an ASIC.

Growing community of inventors

Fremont, CA, United States of America

Adi Srinivasan