Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS20080134122 A1
Publication typeApplication
Application numberUS 12/027,024
Publication dateJun 5, 2008
Filing dateFeb 6, 2008
Priority dateFeb 17, 2006
Also published asUS7376921, US20070198960
Publication number027024, 12027024, US 2008/0134122 A1, US 2008/134122 A1, US 20080134122 A1, US 20080134122A1, US 2008134122 A1, US 2008134122A1, US-A1-20080134122, US-A1-2008134122, US2008/0134122A1, US2008/134122A1, US20080134122 A1, US20080134122A1, US2008134122 A1, US2008134122A1
InventorsDimitris K. Fotakis, Milan F. Jukl
Original AssigneeAthena Design Systems, Inc.
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Methods for Tiling Integrated Circuit Designs
US 20080134122 A1
Abstract
Methods for routing in the design of integrated circuits (ICs) to simplify the routing task. The method includes dividing a given IC design into a limited number of non-overlapping tiles, and then routing all tiles in parallel, each tile being independently routed by a standard router. Thereafter, routed tiles are assembled to form a routing solution for the entire IC. Details of exemplary methods are disclosed.
Images(9)
Previous page
Next page
Claims(18)
1. A method for tiling an integrated circuit (IC) design, said method comprising:
computing a routing capacity of each metal lithographic layer of said IC design;
constructing a set of horizontal and vertical wire segments;
bi-sectioning said IC design for identifying a set of tiles; and
partitioning the connectivity of said IC design based on the identified tiles.
2. The method of claim 1, further comprising:
distributing said tiles to a plurality of distributed routers;
simultaneously routing said tiles on the plurality of distributed routers; and
assembling the routed tiles to form a routed IC design.
3. The method of claim 1, wherein said IC design is provided by at least one of: a placement tool, a global router; and a fine router.
4. The method of claim 1, wherein constructing said set of horizontal and vertical wire segments comprises:
receiving said IC design data from a placement tool;
for each net in said IC design, constructing an enclosing rectangle, wherein said enclosing rectangle includes all ports of said net; and
setting two opposite sides of said enclosing rectangle as a wire segment.
5. The method of claim 1, wherein bi-sectioning said IC design comprises:
receiving a rectangular portion of said IC design, said rectangular portion includes at least one horizontal wire segment and at least one vertical wire segment;
positioning a cut line perpendicularly to a longer side of said rectangular portion;
searching for a location that said cut line cuts a minimum number of wire segments; and
dividing said rectangular portion in the location with minimum cuts of wire segments, and thereby forming two smaller rectangular portions.
6-13. (canceled)
14. A machine-readable medium that provides instructions to implement a method for accelerating the generation of a physical layout of an integrated circuit (IC) design, which instructions, when executed by a set of processors, cause said set of processors to perform operations comprising:
computing a routing capacity of each metal lithographic layer of said IC design;
constructing a set of horizontal and vertical wire segments;
bi-sectioning said IC design for identifying a set of tiles; and
partitioning the connectivity of said IC design based on the identified tiles.
15. The machine-readable medium of claim 14, further comprising:
distributing said tiles to a plurality of distributed routers;
simultaneously routing said tiles on a set of distributed routers; and
assembling the routed tiles to form a routed IC design.
16. The machine-readable medium of claim 15, wherein said IC design is provided by at least one of: a placement tool, a global router, and a fine router.
17. The machine-readable medium of claim 15, wherein constructing said set of horizontal and vertical wire segments comprises:
receiving said IC design data from a placement tool;
for each net in said IC design, constructing an enclosing rectangle, wherein said enclosing rectangle includes all ports of said net; and
setting two opposite sides of said enclosing rectangle as a wire segment.
18. The machine-readable medium of claim 14, wherein bi-sectioning said IC design comprising:
receiving a rectangular portion of said IC design, said rectangular portion includes at least one horizontal wire segment and at least one vertical wire segment;
positioning a cut line perpendicularly to a longer side of said rectangular portion;
searching for a location that said cut line cuts a minimum number of wire segments; and
dividing said rectangular portion in the location with minimum cuts of wire segments, and thereby forming two smaller rectangular portions.
19-26. (canceled)
27. A distributed system for accelerating the routing of an integrated circuit (IC) design, said system comprising:
a main computing node having at least a multi-processing agent for tiling said IC design;
a plurality of remote processing nodes coupled to said main computing node and programmed for simultaneously executing the routing of the tiles on a set of distributed routers; and
a communication network for communication between said main computing node and said plurality of remote processing nodes.
28. The distributed system of claim 27, wherein each of said distributed routers is at least one of: a global router, a fine router.
29. The distributed system of claim 27, where said multi-processing agent is further capable of:
distributing said tiles to said plurality of remote processing nodes; and
assembling the routed tiles by said main computing nodes.
30. The distributed system of claim 29, wherein tiling said IC design comprises:
computing a routing capacity of each metal lithographic layer of said IC design;
constructing a set of horizontal and vertical wire segments;
bi-sectioning said IC design for identifying a set of tiles; and
partitioning the connectivity of said IC design based on the identified tiles.
31. The distributed system of claim 30, wherein constructing said set of horizontal and vertical wire segments comprises:
receiving said IC design data from a placement tool;
for each net in said IC design, constructing an enclosing rectangle, wherein said enclosing rectangle includes all ports of said net; and
setting two opposite sides of said enclosing rectangle as a wire segment.
32. The distributed system of claim 31, wherein bi-sectioning said IC design comprising:
receiving a rectangular portion of said IC design, said rectangular portion includes at least one horizontal wire segment and at least one vertical wire segment;
positioning a cut line perpendicularly to a longer side of said rectangular portion;
searching for a location that said cut line cuts a minimum number of wire segments; and
dividing said rectangular portion in the location with minimum cuts of wire segments, and thereby forming two smaller rectangular portions.
Description
BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates generally to the design of integrated circuits (ICs), and more particularly to the routing phase of an IC design.

2. Prior Art

Due to the ever increasing complexity of integrated circuit (IC) designs, IC designers become more and more reliant on electronic design automation (EDA) tools. An IC is fabricated by a series of lithographic steps that may be abstracted as a construction of a multilayered stack of materials, each layer consisting of a large set of simple geometries. In the related art, EDA tools that manipulate modular sets (e.g., component libraries, IP blocks, and so on) of simple geometries are provided. Each such modular set consists of geometries that exist on several or all lithographic layers of the IC and contains a logic function, ranging from a simple inverter library cell to an IP block module that may hold a microprocessor.

Generally, the processing steps taken by an EDA tool to obtain an IC layout are: a) mapping of the logic for an IC to existing blocks and further partitioning the circuit into blocks of modules or circuits; b) floor planning, which finds the alignment and relative orientation of the circuit blocks; c) placement, which determines more precisely the positions of the circuit blocks and their component blocks; d) routing, which completes the interconnects among electrical components; and e) verification, which checks the layout to ensure that it meets design and functional requirements.

Routing is a key operation in the physical design cycle. Routing consists of establishing a set of electrical conductors that may be constructed on almost any of the geometries and most of the lithographic layers. However, due to the ever-increasing electrical constraints, modern ICs bound all routing geometries to metal layers, wherever such layers have not been utilized by the connections within the library modules or IP building blocks. Given the fact that modern technology includes many metal layers, placement procedures construct a complete packing of the building IP modules or blocks with no space reserved for routing. All routing is then constructed from geometries that exist on the layers above the building blocks contents.

A set of terminal points to be connected is commonly known as a net. Geometries that may have been utilized in the construction of the already placed building blocks are represented as a list of obstacles for the routing problem. Routing is usually performed in two phases: global and detailed. In the global routing, the routing of the nets of the circuit disregards the exact geometric details of each wire and terminal. For each wire, a global router finds a topology that represents the wires of a given net. That is, the global routing specifies the loose route of a wire through the routing space, attempting to reconcile net demand on routing space. The global routing is followed by a detailed routing that completes the point-to-point connections, thereby realizing the connectivity of each net. Global routing usually includes detailed information, such as layer assignment of wire segments, widths of wires, and so on. The detailed routing is performed using a fine router.

Due to the fact that a typical IC consists of millions of nets, a routing of a single net is a NP-hard problem. This fact indicates a potential of extremely high demands on computational resources. Therefore, it would be advantageous to provide a solution that simplifies the routing of IC designs and allows routing tasks to be executed by standard computational resources.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a flowchart describing a method for tiling an IC design in accordance with one embodiment of the present invention.

FIGS. 2A, 2B, 2C and 2D are schematic diagrams of an IC design used to exemplify the techniques of the tiling method of FIG. 1.

FIG. 3 is a flowchart describing the execution of the bisection task in accordance with one embodiment of the present invention.

FIG. 4 is a flowchart describing the process for partitioning the IC connectivity.

FIG. 5 is a flowchart describing the task of identifying exact positions of exit locations in accordance with one embodiment of the present invention.

FIG. 6 is a block diagram of an exemplary distributed multi-processing system used to carry out the techniques disclosed in accordance with the present invention.

FIGS. 7A and 7B are exemplary layouts before and after the execution of the tiling method of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The present invention comprises a method that allows the simplification of the routing phase in the design of integrated circuits (ICs). Using the disclosed method, the time expended on the routing task is significantly reduced. The method includes dividing a given IC design into a limited number of non-overlapping tiles, and then routing all tiles in parallel, each tile being independently routed by a standard router. Thereafter, routed tiles are assembled to form a routing solution for the entire IC.

FIG. 1 shows an exemplary and non-limiting flowchart 100 describing a method for tiling an IC design in accordance with one embodiment of the present invention. At 110, the files of an IC design to be tiled are received. The input design may be a result of a placement or routing tool. Input from a global router can be utilized to tile the design and then routing the tiles using fine routers. Input from a fine router can be utilized for tiling the design for the purpose of optimization, calculating delays, and RC extraction.

FIG. 2A shows an exemplary IC design 200 used to describe the techniques of the disclosed method. IC 200 includes four cells 210-1 through 210-4. Each cell 210 may represent a single logic element, such as a gate, or several logic elements interconnected in a standardized manner to perform a specific function. Each of cells 210 has at least one port (or a terminal points) 220, each of which may be connected, by wires, to one or more other ports 220 of IC 200. The wires connecting the ports of the IC are also formed on the surface of the IC. IC 200 includes two nets 230-1 and 230-2. Net 230-1 is a two-port net comprising ports 220-4 and 220-5 and net 230-2 is a three-port net consisting of ports 220-2, 220-6, and 220-8. A cell may further include one or more input and output ports (not shown in FIG. 2A).

At 120, a routing capacity of each metal lithographic layer is determined. Specifically, layers are sorted into vertical and horizontal metal tracks. Then the capacity of the horizontal and vertical metal tracks is computed. This is performed by examining each layer's properties from data provided by an IC manufacturer (foundry). At 125, the method constructs a model of the connectivity for each net with respect to its horizontal and vertical routing resource demands. This is done by constructing a set of horizontal and vertical wire segments to be realized on the IC area. The construction is based on the input IC provided in 110 and the routing resources determined at 120. Specifically, if the input IC is provided by a global routing tool, the horizontal and vertical segments are the resultant wiring of this tool. If the nets of the IC were fine routed, the segments generated by the fine router are utilized. If the input IC provided by a placement tool, i.e., no routing (global or fine) data is available, for each net an enclosing rectangle including all ports of a net is constructed. The two opposite (parallel) sides of such a rectangle is the wire segment. FIG. 2B shows the IC 200 where a vertical wire segment 240-1 and a horizontal wire segment 240-2 are constructed. The vertical and horizontal segments 240-1 and 240-2 include nets 230-1 and 230-2, respectively.

At 130, potential tiles in the input IC design are determined by bi-sectioning the input design. FIG. 3 shows the execution of the bisection process in greater detail. At 310, a first parameter (Tiles_Num) and a second parameter (Nets_Num) are set to their initial values. At 320, a rectangle to be divided is received as an input. A rectangle is a part or the entire IC design and includes its respective horizontal and vertical wire segments. At 330, the longer side of the rectangle is found, and then, at 340, a cut line is perpendicularly positioned to that side. This is performed in order to generate a rectangle tile with an aspect ratio close to one. The cut line is also perpendicular to a direction of given horizontal or vertical sets of segments. At 350, the process looks for a cut line position where the cut line will cut a minimum number of wire segments running in the orthogonal direction. In order to achieve tiles with a reasonable size, the position of the cut line should be within a predefined range (e.g., between 30% and 70%) of the perpendicular dimension of the rectangle being cut. At 360, in the position determined at 350, the original rectangle is divided into two rectangles. FIG. 2C shows IC 200 that includes a cut line 260 that is perpendicular to horizontal segment 240-2. As can be shown, cut line 260 crosses segment 240-2 in one location, i.e., only one net is divided. As a result, cut line 260 is the partition of IC 200 and forms two rectangles 270-1 and 270-2. At 370, a check is made to determine if the number of rectangles equals to the value of the Tiles_Num parameter, and if so execution terminates; otherwise, execution continues with 380. At 380, another check is performed to determine if the number of nets in each rectangle is below the value of the Nets_Num parameter, and if so execution ends; otherwise, execution returns to 320 where the execution is repeated with the rectangles constructed at 360.

At 140, the IC connectivity is partitioned, based on the constructed rectangles, and thereby forming an overlay of the IC area. FIG. 4 shows the execution of the partitioning task in greater detail. At 410, all nets in the input IC design are sorted into two groups: Local and Global. A Local net is entirely bounded in a single tile, whereas Global nets reside in two or more tiles. At 420, exact positions of exit locations of each Global net on the boundaries of each tile are identified. This is performed to allow the separation of a routing task of the entire IC into multiple independent tasks. The execution of the process performed at 420 is described in greater detail below with reference to FIG. 5. At 430, Global nets are fragmented into fragments, where exit locations and terminal points of a given net inside a tile form a new net (hereinafter “tile net”). At 440, terminals of Local nets are copied as a single tile net into a tile that the Local nets belong to. At 450, for each tile, obstacle and fixed geometries that are in the area of the tile are copied to the tile. That is, a tile includes all information of the piece of an IC that it covers. At 460, for each tile, one track obstacle of a guard band is generated. This is performed to communicate existence of features inside adjacent tiles that may be near the boundaries of the tile. The result of the partitioning process is a set of tiles that together form an overlay of the IC area. In a typical IC, but not by way of limitation, the desired number of such tiles is between 1 and 100. FIG. 2D depicts IC 200 after the partitioning the connectivity. As shown, two tiles 270-1 and 270-2 are constructed were the tiles boundaries are on top of cells 210. The net 230-2 has connection points in both tiles 270, and thus this net is divided to form the adjacent abutting exit locations 280-1 and 280-2.

It should be noted that all data on tile boundaries is defined as fixed and any changes made to this data requires the reconstruction of the respective tile and its neighbor tiles. The reconstruction of tiles can be carried out by repeating the execution of the partitioning task discussed with reference to FIG. 4.

Upon completion of the tiling of the input IC design, each tile is sent to a router. This allows the routing of up to all, or at least two tiles, on a set of distributed routers simultaneously. At 150, once the routing of each tile is completed, its resulting wires are assembled (i.e., copied and merged) together to form a routing solution for the entire IC. It should be appreciated by a person skilled in the art that by routing the tiles independently and simultaneously, the time required for completing the routing phase is reduced typically by an order of magnitude.

FIG. 5 is a non-limiting flowchart 430 describing the process of identifying positions of exit locations in accordance with one embodiment of the present invention. The exact positions of exit locations are identified for all Global nets on the boundaries of all tiles. At 510, an additional routing tile (hereinafter a “Globaltile”) having a size of the entire IC is generated. The Globaltile includes all Global nets and building block geometries that are in the vicinity of tile boundaries that Global nets reside in. The Globaltile does not contain any information on the tiles. At 520, the Globaltile is routed as a first tile of the parallel routing and the result is a fine routing of all Global nets included in the Globaltile. Common wisdom of the art shows that the Global nets account for between ten (10) and fifteen (15) percent of the total nets presenting the IC design. Generally, this sparse problem can be easily routed by any fine router. At 530, upon completion of the routing of the Globaltile, each wire of each of the Global nets is extracted, and then, at 540, intersection points where the extracted wires intersect the boundaries of all the tiles are identified. These intersection points are the exit locations of Global nets from the tiles.

It should be noted that the tiling of an IC design is a valuable asset for handling of engineering changes to a completed IC. That is, the tiling approach restricts the work of realizing the change only in tiles that are affected while preserving the rest of the completed design. This is particularly important for very large designs, where a change may affect only a small part of a design, often contained in a single tile.

FIG. 6 shows a diagram of an exemplary distributed multi-processing system 600 that can be utilized to execute the tiling process in accordance with one embodiment of the present invention. System 600 comprises a main computing node 610 and a plurality of distributed remote processing nodes 630. The main computing node 610 includes a main database 611 for holding design information, a script engine 612 for propagating scripts to be executed by remote processing nodes 630, a data streamer 613 for transferring binary data streams to remote processing nodes 630, and a multi-processing agent (MPA) 620. MPA 620 is the infrastructure that enables the distributed parallel processing. Specifically, MPA 620 manages the distributed processing resources and the transfers of pluralities of data streams to and from remote processing nodes 630. In addition, main computing node 610 preferably includes a central processing unit (CPU) 615 for executing various of the processing tasks.

Each of the remote processing nodes 630 includes a remote script engine 631, a remote data streamer 632 for receiving and transforming data streams, a remote database 633 for maintaining block information, and a third party interface 634. The third party interface 634 interfaces with at least a router 640. A remote processing node 130 preferably includes a CPU 635 having its own operating system and being capable of performing various processing tasks. Remote processing nodes 630 are part of a computer farm where workload management for achieving the maximum utilization of computing resources is performed by MPA 620. The communication between main computing node 610 and a remote processing node 630 is performed over a network 605. The architecture and the operation of system 600 is described in greater detail in U.S. patent application Ser. No. 11/315,892 entitled “System for Performing Parallel Distributed Processing for Physical Layout Generation” assigned to the common assignee and is hereby incorporated by reference for all that it contains.

In accordance with the present invention, the tiling method discussed in greater detail above is executed by main computing node 610. That is, main computing node 610 breaks an IC design, saved in main database 611, into non-overlapping tiles. Each tile is transferred as a data stream to remote processing nodes 630 and saved in remote databases 633. Each of nodes 630 receives the data stream that encapsulates tile information and routes the tile using router 640. In fact, geometries on all layers are copied into the tiles that are sent to routers. The distribution of tasks to remote processing nodes 640 is performed by MPA 620 in a way that ensures optimized performance and load balancing. Once the routing of a tile has completed, only incremental routed data is sent back to main computing node 610 as a data stream. The pieces of incremental routed data received from remote processing nodes 630 are merged and saved in main database 611.

It should be emphasized that the tiling method disclosed herein is a geometric algorithm that divides the physical layout into non-overlapping tiles, the tiles may or may not be of equal size. The number of tiles is preferably on an order of ten (10). FIG. 7A shows an exemplary layout 700 of an IC after placement. Executing the tiling method on layout 700 would result layout 710, shown in FIG. 7B, that includes 17 tiles. As can be shown these tiles are not overlapped and of different tile sizes. In accordance with one embodiment of the present invention, each generated tile can be further used for the purposes of mask analysis and mask generation. This is possible, as a tile represents a small fraction of the complex full IC nevertheless still having the ability to have mask operations performed on it in seclusion from the rest of the tiles. Furthermore, each tile can be used as a data unit that incrementally adjusts routing and mask generation tasks with minimum perturbation to the entire IC.

Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US7934189 *Jan 25, 2008Apr 26, 2011Infineon Technologies AgMethod of making an integrated circuit including simplifying metal shapes
US8607183Jan 20, 2011Dec 10, 2013Infineon Technologies AgMethod of making in an integrated circuit including simplifying metal shapes
Classifications
U.S. Classification716/129, 716/130, 716/131
International ClassificationG06F17/50
Cooperative ClassificationG06F17/5077, G06F17/5068
European ClassificationG06F17/50L2, G06F17/50L