Achieving coverage closure ranks as one of the most difficult challenges that product development teams would like solved, as identified by the recent Wilson Research Group Functional Verification Study. To address the coverage closure challenge, many of the industry’s leading functional verification teams have begun turning to a new and emerging technology called Intelligent Testbench Automation (“iTBA”). iTBA combines the high quality of directed testing with the high quantity of constrained random testing, and can be easily integrated into existing verification environments.
In June of this year, we released a new module in the Verification Academy titled Intelligent Testbench Automation, which provides an excellent introduction to the topic and offers a good foundation of understanding on how to integrate iTBA in your own verification environments. The Intelligent Testbench Automation module provides a complete introduction to iTBA, showing you how you can achieve your coverage goals >10X faster, leaving you the option to reduce your verification time, expand your coverage targets even further, or both.
Although many engineers have been intrigued by the possibility of achieving >10x faster time to verification closure, they still have concerns about the investment they have in their existing simulation testbench infrastructure. Yet, it turns out that it is possible to achieve the iTBA revolutionary results with simple evolutionary steps. In other words, you do not have to throw out your legacy testbench infrastructure. The iTBA technology can coexist with your current testbench infrastructure. To demonstrate how this can be accomplished, we are expanding the Verification Academy Intelligent Testbench Automation module with three additional sessions that demonstrate how to integrate iTBA into your SystemC environment, or your directed test environment, or even into an “e” environment. In addition, we have introduced three more advanced topics related to iTBA covering topics such as reactive graphs, configurable rule graphs, and distributed simulation.
As always, I look forward to hearing your feedback and suggestions on how we can continue to improve and expand the Verification Academy content.
- Harry Foster
Harry Foster is the Lead Verification Academy Subject Matter Expert (SME) and a Chief Verification Scientist at Mentor Graphics.
Academy Forums
Have an UVM/OVM or SystemVerilog question? Post it in the Academy Forums and our Verification Methodologists will answer it.
UVM/OVM Cookbook
Looking for more information on hierarchical sequence generation? Login and check out the UVM/OVM Online Cookbook sequence articles.
Academy Spotlight - Design & Verification in the SoC Era
Tuesday, October 18th | San Jose - DoubleTree Hotel
At this seminar, learn how you can dramatically boost productivity, radically decrease time-to-coverage, and more effectively and efficiently manage your design and verification processes from transactions to gates.
Keynote Presenters include:
John Goodenough, Vice President of Design Technology and Automation at ARM Inc.
Harry Foster, Chief Verification Scientist at Mentor Graphics.
This session describes integrating Intelligent Testbench Automation into a directed test environment, re-using existing directed test code, and achieving >10X more functional coverage.
This session describes integrating Intelligent Testbench Automation into an ‘e’ environment, re-using existing eVCs, and achieving functional coverage >10X faster.
This session describes how Intelligent Testbench Automation can distribute its process across a network of simulation servers, to achieve functional coverage goals even faster. iTBA realizes nearly linear results (i.e. 100 servers achieves better than 95X improvement), as overhead is less than 5%, and no redundant work is performed by each server. Automatic allocation of work, load balancing, and coverage monitoring will also be discussed.
This session describe show configurable graphs can be developed to support broad re-use and to allow stimulus to be easily focused on specific areas without modifying the graph source.