Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Metal Density Aware Signal Routing Method

IP.com Disclosure Number: IPCOM000184127D
Original Publication Date: 2009-Jun-12
Included in the Prior Art Database: 2009-Jun-12
Document File: 6 page(s) / 133K

Publishing Venue

IBM

Abstract

Disclosed is a signal routing method that is metal density aware, so that it routes the nets and determines the optimal metal fill pattern such that timing is not degraded by a large factor and metal density requirements are met. Given the maximum and minimum density requirement, window size, step size, instance count threshold in the window and the type of metal fill (connected and floating, square and rectangle, staggered and non-staggered), the signal routing method takes the optimal metal fill pattern. This results in exact capacitance consideration due to metal fill during post-route timing optimization, and not the estimated capacitance value.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 41% of the total text.

Page 1 of 6

Metal Density Aware Signal Routing Method

Authors

Veena Prabhu, Karan B Koti

The dielectric layers in chip designs vary in thickness due to the different patterns of metal on successive metal layers. The thickness variations reduce yield and impact chip performance. Adding inactive metal segments called metal fill to open areas of the design, reduces the variations. The metal fill makes the topology of the metal layers more uniform, which reduces the variations in metal density. Metal fill addition is done as a final Physical Design Step, after signal routing. The traditional signal routing method does not consider the metal density requirements during routing of nets. The additional metal increases cross-coupling capacitance. The increase in cross-coupling capacitance adversely affects timing. Since the existing routing method does not foresee the timing impact due to metal fills, it involves estimating the timing margin to be considered for post- route timing optimization which may not be accurate [1].

The existing signal routing methods take the cell delays, the resistance & capacitance of the metal layers from the technology library files, and do optimal routing along the available routing tracks to meet timing. After signal routing, extraction of the resistance and capacitance of the nets assumes a percentage of metal fills, so that the cross-coupling capacitance due to metal fills is also considered. If timing is met, metal fill addition is done, and resistance and capacitance extraction that follows considers the actual coupling-capacitance due to metal fills. If timing is not met, timing optimization is done where timing issues are fixed, and nets are re-routed. Then RC extraction assumes a metal fills percentage, followed by timing checks. This process is repeated till finally timing is fixed. Fig. 1 illustrates the above process as a flow chart:

1

Page 2 of 6

Figure 1

The disclosed signal routing method does optimal signal routing along the available routing tracks to meet timing, and then determines the metal fill pattern around these routes such that timing impact is minimal due to this metal fill pattern. Metal density specifications like maximum and minimum density requirement, window size, step size, instance count threshold in the window and the type of metal fill (connected and floating, square and rectangle, staggered and non-staggered) are input to the signal routing method, along with the cell delays and the resistance and capacitance of metal layers.

The obtained metal fill pattern is saved separately. The extraction of resistance and capacitance uses the metal fill pattern along with the routed design, for subsequent timing analysis. If timing is met, the saved metal fill pattern is implemented. Else, timing optimization is done to fix timing issues. Timing optimization could involve routin...