International Business Machines Corporation

David Anthony Papa

Zhuo Li

Charles Jay Alpert

Chin Ngai Sze

Gi-Joon Nam

Natarajan Viswanathan

Jarrod Alexander Roy

Wael R El-Essawy

Michael David Moffitt

Samuel I Ward

Brian C Wilson

Jeremy T Hopkins

Wael R Ei-Essawy

Tao Luo

A mechanism is provided for automatically routing network interconnects in a data processing system. A processor in a node of a plurality of nodes receives network topology from neighboring nodes in the plurality of nodes within the data processing system. The processor constructs a system node map that identifies a physical connectivity between the node and the neighboring nodes. The processor programs a switch in the node with a connectivity map that indicates a set of point-to-point connections with the neighboring nodes. The set of point-to-point connections comprise locally-connected connections and pass-through connections.

Automatically routing super-compute interconnects

A mechanism is provided for determining connectivity while minimizing wiring in an electronic system. The mechanism identifies a configuration of the electronic system, a location of each module in a plurality of modules within the electronic system and at least one constraint with regard to wiring the electronic system, the location of each module being identified using three-dimensional coordinates. The mechanism routes a separate cable from each module in the plurality of modules to each of the other modules in the plurality of modules without violating any constraints, thereby forming a plurality of cables. The mechanism then generates a cabling list indicating how each cable in the plurality of cables is to be routed in the electronic system in order to not violate any constraints and provide connectivity while minimizing wiring.

Minimizing the maximum required link capacity for three-dimensional interconnect routing

A method, system, and computer usable program product for latch clustering with proximity to local clock buffers (LCBs) where an algorithm is used to cluster a plurality of latches into a first plurality of groups in an integrated circuit. A number of groups in the first plurality of groups of clustered latches is determined. A plurality of LCBs are added where a number of added LCBs is the same as the number of groups in the first plurality of groups. A cluster radius for a subset of the first plurality of groups of clustered latches is determined, a group in the subset having a cluster radius that is a maximum cluster radius in the subset. The plurality of latches are reclustered into a second plurality of groups responsive to the maximum cluster radius exceeding a radius threshold, the second plurality of groups exceeding the first plurality of groups by one.

Latch clustering with proximity to local clock buffers

A physical synthesis tool for dock optimization with local clock buffer control optimization is provided. The physical synthesis flow consists of delaying the exposure of clock routes until after the clock optimization placement stage. The physical synthesis tool clones first local clock buffers. Then, the physical synthesis tool runs timing analysis on the whole design to compute the impact of this necessarily disruptive step. After cloning local clock buffers, the physical synthesis tool adds an extra optimization step to target the control signals that drive the local clock buffers. This optimization step may includes latch cloning, timing-driven placement, buffer insertion, and repowering. The flow alleviates high-fanout nets and produces significantly better timing going into clock optimization placement. After placement, the physical synthesis tool fixes latches and local clock buffers in place, inserts clock routes, and repowers local clock buffers.

Clock optimization with local clock buffer control optimization

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

The input slew at a selected gate of an integrated circuit design is computed by assigning a default slew rate to the output gate of a previous logic stage which is greater than a median slew rate for the design. This default slew rate is propagated through the logic stage to generate an input slew rate at the selected gate. The default slew rate corresponds to a predetermined percentile applied to a limited sample of preliminary slew rates for randomly selected gates in the design. The default slew rate is adjusted as a function of known characteristics of the wirelength from the output gate to a first gate in the second logic stage. The delay of the selected gate is calculated based on the input slew rate. The input slew rate can be stored during one optimization iteration and used as a default slew rate during a later optimization iteration.

Accuracy pin-slew mode for gate delay calculation

A portion of a gate-level netlist representing an integrated circuit design is selected for optimization. A timing window representing the selected portion is made comprising one or more copies of the selected portion. A checkpoint is created for the timing window and stored in a transaction history. One or more changes are then made to the timing window and stored in the transaction history. The changed elements are marked as dirty and stored in the transaction history. After the one or more changes have been made, the timing window is queried for current timing conditions and compared with the checkpoint. If the one or more changes are an improvement, the one or more changes are committed by replicating the one or more changes to the portion of the gate-level netlist. If the one or more changes are not an improvement, the timing window may be restored to the checkpoint.

Method for bounded transactional timing analysis

A timing-driven cloning method iteratively partitions sinks of the net into different sets of clusters and for each set computes a figure of merit for a cloned gate location which optimizes timing based on linear delay, that is, a delay proportional to the distance between the cloned gate location and the sinks. The set having the highest figure of merit is selected as the best solution. The original gate may also be moved to a timing-optimized location. The sinks are advantageously partitioned using boundaries of Voronoi polygons defined by a diamond region surrounding the original gate, or vice versa. The figure of merit may be for example worst slack, a sum of slacks at the sinks in the second cluster, or a linear combination of worst slack and sum of the slacks.

Optimal timing-driven cloning under linear delay model

A method, data processing system and computer program product for optimizing the placement of logic gates of a subcircuit in a physical synthesis flow. A Pyramids utility identifies and selects movable gate(s) for timing-driven optimization. A delay pyramid and a required arrival time (RAT) surface are generated for each net in the selected subcircuit. A slack pyramid for each net is generated from the difference between the RAT surface and delay pyramid of each net. The slack pyramids are grown and tested using test points to generate a worst-case slack region based on a plurality of slack pyramids in the selected subcircuit. The worst-case slack region is mapped on a placement region and a set of coordinates representing the optimal locations of the movable element(s) in the placement region are determined and outputted.

Method for incremental, timing-driven, physical-synthesis optimization under a linear delay model

A method, data processing system and computer program product for optimizing the placement of logic gates of a subcircuit in a physical synthesis flow. A Path Smoothing utility identifies one or more movable gates based on at least one selection criteria. A set of legalized candidate locations corresponding to one or more identified movable gates is generated. A disjunctive timing graph based on the generated set of legalized candidate locations is then generated. An optimal location of one or more movable gate(s) is determined using a recursive branch-and-bound search and stored in the computing device.

Growing community of inventors

Austin, TX, United States of America

David Anthony Papa