Protocol-directed trace signal selection for post-silicon validation

Post-silicon validation

Sharma, Abhishek ; Vasudevan ; Shobha

关键词: Post-silicon validation;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/95447/SHARMA-THESIS-2016.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Due to the increasing complexity of modern digital designs using NoC (network-on-chip) communication, post-silicon validation has become an arduous task that consumes much of the development time of the product. The process of finding the root cause of bugs during post-silicon validation is very difficult because of the lack of observability of all signals on the chip. To increase observability for post-silicon validation, an effective silicon debug technique is to use an on-chip trace buffer to monitor and capture the circuit response of certain selected signals during its post-silicon operation. However, because of area limitations for debug structures on chip and routing concerns, the signals that are selected to be traced are a very small subset of all available signals. Traditionally, these trace signals were chosen manually by system designers who determined what signals may be needed for debug once the design reaches post-silicon. However, because modern digital designs have become very complex with many concurrent processes, this method is no longer reliable. Recent work has concentrated on automating the selection of low-level signals from a gate-level analysis. But none of them has ever been able to interpret the trace signals as high-level meaningful debugging information. In this work, we present an automated protocol-directed trace selection where the guiding force is the set of system-level protocols. We use a probabilistic formulation to select messages for tracing and then further analyze these solutions. This method produces traces that allow a debugger to observe when behavior has deviated from the correct path of execution and localize this incorrect behavior for further analysis. Most importantly, unlike the previous gate-level analysis based methods, this method can be applied during the chip design phase when most of the debug features are also designed. In addition, this method drastically reduces the time needed to select signals, as we automate a currently manual process.