Abstract: The present invention comprises method for optimizing an integrated circuit design that includes computing of capacities and delays of an integrated circuit design, resynthesizing said integrated circuit design utilizing a plurality of local optimization procedures, and removing overlap the local optimization procedures can include a local resynthesis of logic trees procedure that utilizes multiple cost functions, a dynamic buffer and inverter tree optimization procedure, and a cell resizing procedure. Generally, faster local optimization procedures are applied first and slower, more thorough procedures are applied to areas where the faster procedures have not solved the optimization tasks.
Type:
Grant
Filed:
October 2, 2000
Date of Patent:
May 13, 2003
Assignee:
LSI Logic Corporation
Inventors:
Andrej A. Zolotykh, Elyar E. Gasanov, Alexander S. Podkolzin, Valery B. Kudryavtsev
Abstract: A modeling apparatus for producing a model by lamination of sheet material including a support platform which can be raised and lowered, a sheet feed mechanism for feeding sheet material over the platform, a superstructure mounted over the platform, a cutting mechanism for cutting the sheet material to form shaped portions, an adhesive applicator for applying adhesive to the shaped portions, and a control mechanism.
Abstract: Provided are systems and techniques for optimizing an integrated circuit design, in which a critical zone is identified in an integrated circuit design and a plurality of alternative identities are applied in the critical zone in order to obtain a corresponding plurality of outcomes. Alternative representations are then identified as those of the plurality of outcomes pursuant to which at least one of ramptime and timing are improved, and a best one of the alternative representations is selected to replace into the critical zone based on specified priorities which include: (i) selecting based on reduction in ramptime violation; (ii) selecting from among alternative representations that preserve cell area based on timing improvement; and (iii) if all alternative representations increase cell area, selecting based on an evaluation of a relationship between timing decrement and area increment.
Type:
Grant
Filed:
October 2, 2000
Date of Patent:
April 1, 2003
Assignee:
LSI Logic Corporation
Inventors:
Andrej A. Zolotykh, Elyar E. Gasanov, Alexander S. Podkolzin, Valery B. Kudryavtsev
Abstract: Methods for calculating a total capacitance of a metal wire in an integrated circuit is disclosed. In the present invention, a library containing predetermined wiring topologies is created. Each of the wiring topologies has an associated capacitive value. After extracting a layout topology of a segment of the metal wire, the layout topology is used to find and extract one of the predetermined wiring topologies in the library that corresponds to the layout topology. The associated capacitive value for the extracted wiring topology is used to calculate the total capacitance of the metal wire.
Abstract: Provided is an analog-to-digital converter that includes a first analog-to-digital conversion (ADC) stage connected to input a first analog signal and a second ADC stage connected to input a second analog signal produced by the first ADC stage. A tone detector enables the second ADC stage from a disabled state when a first condition indicating the presence of a high-level interference tone is satisfied and disables the second ADC stage from an enabled state when a second condition indicating the absence of a high-level interference tone is satisfied.
Abstract: Constant pins are determined in a combinational circuit by associating an input of a combinational circuit with a first variable and a second variable, with the second variable being the complement of the first variable. For a first logical cell interconnected to such input, a first mathematical representation and a second mathematical representation are computed. The first mathematical representation is a function of the operation of the first logical cell and a function of the first variable, and the second mathematical representation is a function of the operation of the first logical cell and a function of the second variable. A determination is then made as to whether one of the first and second mathematical representations is equal to a constant.
Type:
Grant
Filed:
July 27, 2000
Date of Patent:
March 18, 2003
Assignee:
LSI Logic Corporation
Inventors:
Alexander Andreev, Ranko Scepanovic, Anatoli Bolotov
Abstract: The present invention selects parts of an integrated circuit description for resynthesis and then prepares those parts for resynthesis. Initially, a resynthesis goal is input, with the resynthesis goal having been selected from a set of possible resynthesis goals. Plural buffer and/or logic trees in the integrated circuit description are then selected based on the resynthesis goal, and information for each of the selected trees is obtained and stored. The tree information includes: (i) a description of each tree cell, including cell types, cell coordinates, and flips and angles of the tree cell, (ii) a description of each input net, (iii) a signal arrival time for each input net as a function of a capacity of such input net, (iv) coordinates of each pin driving each input net, and (v) a maximum capacity of each input net that will prevent such input net from having a timing violation.
Type:
Grant
Filed:
October 2, 2000
Date of Patent:
March 11, 2003
Assignee:
LSI Logic Corporation
Inventors:
Andrej A. Zolotykh, Elyar E. Gasanov, Alexander S. Podkolzin, Valery B. Kudryavtsev
Abstract: The present invention involves a method for determining constant pins in a combinational circuit. The method comprises the steps of associating an input of a combinational circuit with a first variable and a second variable, wherein said second variable is the compliment of said first variable, computing for a first logical cell interconnected to said input a first canonical representation, wherein said first canonical representation is a function of the operation of said first logical cell and a function of said first value, computing for said first logical cell a second canonical representation, wherein said second canonical representation is a function of the operation of said first logical cell and a function of said second value, determining whether one of said first and second canonical representations is equal to zero.
Type:
Grant
Filed:
October 2, 2000
Date of Patent:
March 4, 2003
Assignee:
LSI Logic Corporation
Inventors:
Alexander Andreev, Ranko Scepanovic, Anatoli Bolotov
Abstract: The present invention involves a method for reducing delay of a net. The method includes constructing a time-space grid, said time-space grid corresponding to a net, passing a wave through the time-space grid, said wave having a wave value, and inserting a buffer at a point on said time-space grid where insertion of the buffer increases a wave value. The buffer can be a negative buffer or positive buffer. Generally, a second wave is passed through the time-space grid simultaneously with the first wave. Typically, the second wave and the first wave are inverted.
Type:
Grant
Filed:
October 10, 2000
Date of Patent:
February 11, 2003
Assignee:
LSI Logic Corporation
Inventors:
Alexander Andreev, Ranko Scepanovic, Anatoli Bolotov
Abstract: Methods for calculating delays for cells in ASICs are disclosed. In the present invention, delays are computed by considering not only the process (P), voltage (V), temperature (T) but also input ramptime (R) and output load or fanout (F) of the cells by fitting the delay at four corner points for derated PVT condition into a non-linear equation which is a function of P, V, T, R and F. Thus, the delay is a five dimensional characterization, and the characterization is split into (P,V,T) characterization and (R,T) characterization to reduce the characterization time and resources. The present invention provides for accurate calculation of delays for cells in ASICs.
Abstract: Provided is a technique for designing an integrated circuit die which includes a semiconductor layer, a primary metal layer, a horizontal metal layer and a vertical metal layer. Electronic components are laid out on the semiconductor layer, and a primary power distribution network for distributing power to the electronic components is laid out on the primary metal layer. Then, a uniform trunk width is calculated for all trunks in a power mesh based on a desired maximum voltage drop for the generated electronic component layout. Finally, horizontal power trunks are laid out on the horizontal metal layer and vertical power trunks are laid out on the vertical metal layer using the calculated uniform trunk width, so as to form the power mesh, and an electrical connection is specified between the power mesh and the primary power distribution network.
Abstract: The present invention is directed to laying out connection points and routing a signal from an input terminal to the connection points for use in signal distribution to control skew. A pattern of diamond-shaped rings is constructed, each of the diamond-shaped rings including an inner diamond, an outer diamond and all space between the inner diamond and the outer diamond, where the diamond-shaped rings are arranged such that each point within a specified area is included within at least one diamond-shaped ring. A center point is identified for each of the diamond-shaped rings and each center point is designated as a connection point. A connection is then routed between the input terminal and each of plural of the connection points. The invention also is directed to routing a signal from one of plural connection points to a pin so as to control skew.
Abstract: Initially, a graph is electronically displayed, the graph including a historical portion that includes historical values of the variable over time and also including a future portion. Then, a participant is permitted to designate a point on the future portion of the graph (e.g., by using an input device such as a mouse, a touch-sensitive display screen or the like) and the designated point is converted into a predicted value for the variable at a realization time.
Type:
Grant
Filed:
September 8, 1999
Date of Patent:
October 29, 2002
Assignee:
C4Cast.Com, Inc.
Inventors:
G. Michael Phillips, M. Chapman Findlay, III, William P. Jennings, Stephen A. Klein, Mark E. Rice
Abstract: A method for laying out input/output (I/O) pairs, each including an I/O cell and a pad, on an integrated circuit die. Size information is obtained for each of a first I/O pair and a second I/O pair. A minimum pad spacing criterion is obtained which specifies a minimum distance between the pad in the first I/O pair and an element of the second I/O pair, and the first I/O pair and the second I/O pair are laid out so as to satisfy the minimum pad spacing criterion. Also provided is a method for laying out pads for input/output (I/O) cells on an integrated circuit die in which size information is obtained for each of a first I/O cell pad and a second I/O cell pad. A minimum pad spacing criterion is obtained, and the first I/O cell pad and the second I/O cell pad are laid out so as to satisfy the minimum pad spacing criterion.
Abstract: Provided is a technique for circuit delay prediction in which blocks (preferably, non-overlapping blocks) are specified, each of the blocks including a portion of the circuit. Delay calculation collars (DCCs) are then defined for the blocks, the DCCs including complete dependency information required to calculate delay within the blocks. Next, delay is calculated for the blocks based on the DCCs and delay is calculated for the circuit based on the DCCs. The DCCs are then modified as necessary based on results of either or both of the delay calculation for the blocks or the circuit. The delay calculation and DCC modification steps are then repeated.