Timing Violation Predictor (TVP) using Instruction Level Path Sensitization Analysis

Technical Summary

Sanghamitra Roy and Koushik Chakraborty, computer scientists at Utah State University, have developed a method of predicting timing violations in technology nodes, a transistor in silicon chips. Timing violations occur sporadically in high-performance pipelined microprocessors due to rapid technology scaling. 

The inventors discovered that an instruction in a microprocessor, identified by its Program Counter (PC), is an excellent predictor of an upcoming timing violation. Using PC, it is possible to predict several clock cycles in advance, a timing violation in a later pipeline stage. This early indication of an upcoming timing violation can enable a host of robust system design mechanisms, otherwise impossible with existing state-of-the-art techniques. Popular error detecting techniques are unable to predict timing violations in advance, but this technology’s Timing Violation Predictor can sense upcoming violations.

Using full-system architectural simulation, the inventors demonstrated average performance improvements of 20 to 90 percent and 14 to 80 percent over other state-of-art techniques under two diverse faulty environments.


Commercial Applications

Commercial partners would include computer designers and chip design fmanufacturers. 


Competitive Advantages

•  System predicts and tolerates timing violations seamlessly in a pipelined microprocessor 

•  Includes a low-overhead instruction-level Timing Violation Predictor (TVP) that predicts upcoming timing violations in the decode stage

•  Design can effectively handle timing violations in complex microprocessor designs with forwarding logic and back-to-back dependent instruction scheduling



•  See U.S. Patent 9,058,070

•  Predicting Timing Violations Through Instruction Level Path Sensitization Analysis, Design Automation Conference (DAC), 2012 49th ACM/EDAC/IEEE

Patent Information:
Computer Science
For Information, Contact:
Christian Iverson
Utah State University
Koushik Chakraborty Sanghamitra Roy