Upload
minjonet-roussel
View
16
Download
1
Embed Size (px)
DESCRIPTION
Teaching Functional Verification: Lab Mechanics. Design Automation Conference Sunday, June 9, 2002. Agenda. Administration Lab 1 Lab 2 Lab 3 Grading. Administration. Length of labs Lab 1 – 1 week Lab 2 – 4 weeks Lab 3 – 4-6 weeks Instructor/TA Availability - PowerPoint PPT Presentation
Citation preview
Teaching Functional Verification: Lab Mechanics
Design Automation Conference
Sunday, June 9, 2002
Agenda Administration Lab 1 Lab 2 Lab 3 Grading
Administration Length of labs
Lab 1 – 1 week Lab 2 – 4 weeks Lab 3 – 4-6 weeks
Instructor/TA Availability Lab 1 – e-mail/office hours Lab 2 – e-mail/office hours/1-2 lab days Lab 3 – e-mail/office hours/3-4 lab days
Administration (continued) Website with lab specifications Student Message Board
Students helping students Monitored/Administrated by Instructor
Student Team Sizes Lab 1 – Individual Lab 2 – Individual Lab 3 – Individual or groups of 2
Individual requires more time
Lab 1 Focused on directed type testing Black box approach
No visibility into RTL Input to output type tests Bus Functional Models (BFM) provided Students need only to use the BFM’s Bugs fixed by setting “error” input signal Students hand in bug analysis
Post Lab 1 exercise Discuss deficiencies of lab 1
Specification Directed nature
Discuss better approach Pseudo-random stimulus Self-checking
Discuss nature of bugs
Lab 2 Build upon lab 1 and lecture material Given starting point reference from
solution (better approach) of lab 1 Introduce High Level Verification
Languages (HVL’s) Grey box approach Provide a testplan from lab 1 Introduce students to debug
Lab 2 (continued) Bug fixes are done by “forcing” internal
nets If using VHDL, use procedures with FLI calls
This reduces instructors maintenance work load on design
If using Verilog/HVL, force nets directly Introduce students to “regression”
Provide ability for students to regress Students hand in testplan and bug
analysis
Post Lab 2 exercise Discuss deficiencies of lab 2
When are we done? Coverage
Maintenance of code in environment Lab 1 solution not well commented
Reuse Discuss nature of bugs Establish Coding guidelines for
regression purposes (pass, fail, etc)
Lab 3 Builds upon lab 2 and lecture material Provide lab 2 solution Coverage oriented
Provide mechanism for code coverage Provide functional coverage goals
Testplan becomes more of a functional coverage plan
Same approach as lab 2 for bug fixes and debug
Students hand in testplan
Grading Lab 1
Bug analysis Lab 2
Testplan and bug analysis Lab 3
Perform “escape analysis” for each group “Panel” session What bugs were found What techniques were used Poke holes in environment Ensure environment followed testplan
Grading (continued) Complete bug discovery isn’t
necessary for “A” Testplan Environment Methodology
Summary Each lab builds upon itself
Emphasizes reuse and maintenance Labs emphasize lecture material
Testplan Pseudo-random Self-checking Regressions Coverage
Summary (continued) Labs are time intensive
Students time Instructors time May opt to have some classes as “lab
sessions” depending on schedule and syllabus