Logo image
An Effective ECO Methodology for Reducing Back-side Design Rule Violations in Double-sided Signal Routing
Conference paper   Open access

An Effective ECO Methodology for Reducing Back-side Design Rule Violations in Double-sided Signal Routing

Che-Ping Tsai, Fang-Yu Hsu, Wai-Kei Mak and 廷基 王
IEEE/ACM International Conference on Computer-Aided Design, Digest of Technical Papers, ICCAD
2024

Abstract

power delivery network

In response to the continuous scaling of technology nodes, a dedicated back-side metal stack for the power delivery network (PDN) has been proposed to enhance performance. To fully utilize the remaining routing resources left after the back-side PDN construction, netlist planning has been introduced to transform single-sided netlists into double-sided ones, thus enabling double-sided signal routing. However, since netlist planning precedes routing, it may overlook certain elements, leading to design rule violations (DRVs) on the back side. Addressing these DRVs during the implementation of engineering change orders (ECOs) requires designers to correct all violations remaining after place-and-route (P&R), a process that demands substantial manual effort. It has been observed that back-side DRVs mainly stem from excessive competition among back-side nets for limited routing resources. These DRVs cannot be easily resolved by merely refining P&R on the back side; adjustments to the netlists on both sides are crucial as well. Therefore, this work introduces an innovative ECO methodology that integrates netlist adjustments with P&R refinement on both sides to effectively reduce back-side DRVs in double-sided signal routing while preventing the creation of new front-side DRVs. This marks a pioneering effort in tackling this issue. Experiments show that our methodology resolves 100% back-side DRVs from various netlist planning strategies across 14 test cases while keeping similar routing quality.

url
https://doi.org/10.1145/3676536.3676824View
Published (Version of record) Open

Related links

Metrics

1 Record Views

Details

Logo image