International Business Machines Corporation

Globalfoundries Inc.

Samuel I Ward

Benjiman L Goodman

Natarajan Viswanathan

Bao G Truong

Lisa Seacat DeLuca

Kulvir Singh Bhogal

Myung-Chul Kim

Jeremy T Hopkins

Timothy J Eby

William David Ramsour

Linton B Ward, Jr

Zhuo Li

Charles Jay Alpert

Kevin Franklin Reick

Thomas Edward Rosser

Bryan J Robbins

Robert J Shadowen

Manikandan Viswanath

Joshua P Hernandez

Emiliano Lozano

Patrick R Crosby

Jun Zhou

Michael J Lee

Nathaniel Douglas Hieter

David Anthony Papa

Johnny James LeBlanc

Daniel W Cervantes

Julie Ann Rosser

Sertac Cakici

Gahn Wattanadilok Krishnakalin

David Nolan Ault

Systems, methods and articles of manufacture are disclosed for recording events occurring in a virtual world. In one embodiment, properties of events previously recorded and/or attended by a user may be identified. Recording criteria for the user may be derived from the identified properties. Upon identifying an event satisfying the recording criteria, the event may be recorded. The recorded event may be played back at the convenience of the user.

Recording events in a virtual world

An automated method for aligning a critical datapath in an integrated circuit design inserts an artificial alignment net in the netlist which interconnects all cells in the bit stack of the datapath. The cells are placed using a wirelength optimization which assigns weights to wire sections based on the alignment direction. The rate of change of the alignment weighting value can vary during different stages of global placement. The invention is particularly suited for a force-directed placer which uses a linear system solver to obtain a globally optimum solution for placement of the cells having some overlap among the cells, and thereafter spreads the cells to reduce the overlap. Pseudo nets are also inserted which interconnect a cell and an expected location of the cell after spreading for that iteration.

Alignment net insertion for straightening the datapath in a force-directed placer

A datapath extraction tool uses machine-learning models to selectively classify clusters of cells in an integrated circuit design as either datapath logic or non-datapath logic based on cluster features. A support vector machine and a neural network can be used to build compact and run-time efficient models. A cluster is classified as datapath if both the support vector machine and the neural network indicate that it is datapath-like. The cluster features may include automorphism generators for the cell clusters, or physical information based on the cell locations from a previous (e.g., global) placement, such as a ratio of a total cell area for a given cluster to a half-perimeter of a bounding box for the given cluster.

Machine-learning based datapath extraction

Mechanisms are provided for performing placement of cells in a design of a semiconductor device. An initial design of the semiconductor device is generated, the initial design comprising a first placement of cells. A preferred direction of placement associated with the cells is determined. The preferred direction is a direction along which spreading of the cells is preferred. A second design of the semiconductor device is generated by modifying the first placement of the cells to generate a second placement of cells, different from the first placement cells, based on the preferred direction of placement associated with the cells.

Placement of structured nets

A netlist for an integrated circuit design is constrained by virtual or 'soft' pins to control or stabilize the placement of logic such as an architectural logic path. One soft pin is inserted at a fixed location proximate an input net of the path and is interconnected with the input net, and another is inserted at a fixed location proximate the output net and is interconnected with the output net. Cell placement is then optimized while maintaining the virtual pins at their fixed locations. More than two virtual pins may be inserted to bound a cluster of logic. The virtual pins may lie along the input/output nets. Pseudo-net weights are assigned to pseudo-nets formed between a cell and the virtual pins, and the pseudo-net weight can be increased for each placement iteration.

Soft pin insertion during physical design

Spare cells are inserted in a region of an integrated circuit design based on a logic complexity of the region. The logic complexity can be computed based on the number of reachable states of digital logic in the region, and can be correlated to a desired spare cell insertion rate which is then compared to the actual spare cell utilization in the region. The target spare cell rate can further based on logic complexity values for neighboring regions with a proximity penalty.

Spare cell insertion based on reachable state analysis

Bit stacks of an integrated circuit design are identified in a netlist by analyzing cell clusters. Candidate bit stacks are generated for each cluster using cone tracing, and wirelength costs are calculated for the candidate bit stacks based on the cells' locations from a previous (e.g., global) placement. The bit stack partition having a minimum total wirelength cost is selected for the final bit stacks. The invention can find K bit stacks in a cell cluster having N input cells and M output cells, where K, N and M are all different. The method is advantageously made timing aware by weighting connections between cells using weights based on timing information. Once the final bit stacks have been identified, the information can be included in the netlist and passed to a datapath placer for optimized placement.

Generating and selecting bit-stack candidates from a graph using dynamic programming

Datapath placement defines tiers for placement sets of a cell cluster, assigns cells to the tiers constrained by the datapath width, and then orders cells within each tier. Clusters are identified using machine-learning based datapath extraction. Datapath width is determined by computing a size of a bounding box for cells in the cluster. Placement sets are identified using a breadth-first search beginning with input cells for the cluster. Tiers are initially defined using logic depth assignment. A cell may be assigned to a tier by pulling the cell from the next higher tier to fill an empty location or by pushing an excess cell into the next higher tier. Cells are ordered within each tier using greedy cell assignment according to a wirelength cost function. The datapath placement can be part of an iterative process which applies spreading constraints to the cluster based on computed congestion information.

Datapath placement using tiered assignment

Decoupling capacitors (dcaps) are placed in an IC design by assigning different dcap utilization rates to logic cones, applying the rates to corresponding dcap regions surrounding cells in the cones, identifying any overlap of regions from different logic cones, and inserting a dcap at the overlapping region having the highest dcap utilization rate. The best location for the dcap is computed using a hypergraph wherein the cells are edges and the regions are nodes. Any node that is dominated by another node is removed and its edge is extended to the dominating node. The dcap is inserted in the region having the most edges (the edges can be weighted). The process is repeated iteratively, updating the hypergraph by removing nodes connected to dcap location, and inserting the next dcap at a region corresponding to the node which then has the greatest number of connected edges.

Decoupling capacitor insertion using hypergraph connectivity analysis

A method comprises generating a first behavioral model of a circuit describing a physical circuit in a first configuration. The first configuration comprises a first master latch, a first fanout path, and a logic cone. The first master latch couples to the first fanout path and is configured to receive a first data input signal. The first fanout path comprises a plurality of output sinks, each coupled to the logic cone. The first behavioral model is modified to generate a second behavioral model describing the physical circuit in a second configuration. The second configuration comprises an error circuit and an abstract latch clone based on the first master latch. A configuration file is generated based on the second behavioral model. The configuration file comprises information representing a plurality of instantiated latch clones based on the abstract latch clone, each configured to couple to the first data input signal and to one or more output sinks of the plurality of output sinks. The second behavioral model and the configuration file are together configured for input to a synthesis tool.

Circuit design optimization

A method includes generating a first behavioral model of a circuit, the first behavioral model describing a physical circuit in a first configuration. The first configuration comprises a first logic structure configured to generate a first intermediate signal based on a received first plurality of inputs. The first configuration further comprises a logic cone configured to generate a scan output based on the first intermediate signal and a plurality of scan inputs. The first behavioral model is modified to generate a second behavioral model describing the physical circuit in a second configuration. The second configuration comprises an error circuit configured to receive the scan output and the first intermediate signal. A testability model is generated based on the second behavioral model, the testability model comprising a first structural representation of the first logic structure. In the event the first logic structure causes a coverage problem, the testability model is modified to include an inversion structure. The inversion structure is configured based on the first logic structure. The inversion structure is configured to generate an inversion structure output. The testability model is modified to couple the inversion structure output as an input to the error circuit.

Growing community of inventors

Austin, TX, United States of America

Samuel I Ward