Close

Presentation

Logfile Error Analysis: How to make sense of millions of lines of logfiles using CHEW_LOGS (Check Errors and Warning in Logfiles)
TimeMonday, July 11th5pm - 6pm PDT
LocationLevel 2 Exhibit Hall
Event Type
Engineering Track Poster
Engineering Tracks
Topics
AI
Back-End Design
Cloud
Embedded Systems
Front-End Design
IP
DescriptionIn the weeks before taping out a chip, running timing jobs on dozens or hundreds of scenarios can produce millions of lines of logfiles. To avoid having an important error or warning message go unheeded, lucky person needs to go through all the messages and see which ones might be a problem.

To help this, we wrote Chew_logs: Check Errors and Warnings in Logfiles. Chew_logs categorizes and summarizes all errors and warnings and any other targeted message (which we call collectively MOCs, or Messages Of Concern) across multiple logfiles. We start running it early in the project cycle, so that we can instantly see if a new MOC pops up.

While originally written for timing logfiles, it is completely configurable with token-based config files, so it can be used for any EDA tool in any flow. Each MOC can be waived or sub-categorized based on timing mode, process corner, pin name, or anything else that appears in the message or the logfile path.

There are many utilities to help the user debug: view a sample MOC in context, quickly jump to a given MOC line in the original logfile, see which logfiles contain any given MOC, compare two Chew_logs runs, etc.