DocumentCode
33255
Title
Compositional Performance Verification of Network-on-Chip Designs
Author
Holcomb, Daniel E. ; Seshia, Sanjit A.
Author_Institution
Dept. of Electr. Eng. & Comput. Sci., Univ. of California, Berkeley, Berkeley, CA, USA
Volume
33
Issue
9
fYear
2014
fDate
Sept. 2014
Firstpage
1370
Lastpage
1383
Abstract
This paper presents a compositional approach to formally verify quality-of-service properties of network-on-chip designs. A major challenge to scalability is the need to verify worst-case latency bounds for hundreds to thousands of cycles, which are beyond the capacity of state-of-the-art model checkers. The scalability challenge is addressed using a compositional model checking approach. The overall latency bound problem is divided into a number of smaller sub-problems, termed latency lemmas. The sub-problems imply the overall latency bound, but are easier to prove on account of being inductive. A method is presented for computing these lemmas based on the topology of the network and a subset of relevant state, and the latency lemmas are verified using k-induction. The effectiveness of this compositional technique is demonstrated on illustrative examples and an industrial ring interconnection network. In the ring network, a latency bound that cannot be verified in 10 000 s without lemmas is proved inductively in just 75 s when the lemmas are used.
Keywords
integrated circuit interconnections; network-on-chip; quality of service; compositional performance verification; industrial ring interconnection network; k-induction; latency lemmas; network-on-chip designs; overall latency bound problem; quality-of-service properties; Microarchitecture; Model checking; Network topology; Network-on-chip; Quality of service; Topology; Upper bound; On-chip networks; quality-of-service; verification;
fLanguage
English
Journal_Title
Computer-Aided Design of Integrated Circuits and Systems, IEEE Transactions on
Publisher
ieee
ISSN
0278-0070
Type
jour
DOI
10.1109/TCAD.2014.2331342
Filename
6879626
Link To Document