Why Verification IP Switching Costs Are A Myth

Many of you may have heard the story about the woodcutter and his blunt axe. The “Switching cost” of sharpening/buying a new axe may seem to be too high when in a time crunch. But a step back to review the situation and switching to a better tool can be life changing!

In today’s world this applies to chip design and verification teams more than ever. A Verification IP plays a key role in controlling verification schedules. Consider a case where tape out schedule is slipping in spite of having both – Internal VIP and External VIP.

Just like a blunt axe will take much longer to fell a tree, a sub standard Verification IP will prolong your IP Development. On the other hand if you “sharpen your axe” i.e. develop/buy a better Verification solution, it may initially seem like its taking longer and others are getting ahead. But, in the long run you will develop your IP faster.

Consider This Scenario:

Supposing you are the Verification Manager. You believe that in the short term, you need a Verification IP to meet schedule deadlines. You select a few Verification IPs from market for evaluation. In parallel you ask an internal team to start working on building a Verification IP for the long run. DUT verification begins in full swing.

Before you know it, chip tape out starts nearing! Even after a lot of effort, functional verification is not converging. The VP starts frequenting to your cubical. All in all life becomes tougher…

Then a discussion starts off between you and verification lead. Verification lead provides list of the issues being faced. He informs you that the Verification IP being developed internally is buggy. Lack of resources is pushing out the internal Verification IP schedule.

Cut to the externally bought Verification IP. There is delay in support turnaround. The test bench bundled with Verification IP turned out to be a “toy” test bench requiring significant internal effort to write the full test bench. Debug is suffering due to insufficient logging. Coverage suite and test suite bundled are not delivering quick value due to lack of customizability. Suddenly all hell breaks loose!

Now your worst nighmare has begun. You already have an external VIP licensed. Internal Verification IP is no where near ready. Verification schedules are still looking way out of control.

Now You Need To Ask Yourself The Million Dollar Question

(Actually, a multi million dollar if you consider revenue loss due to delays)

Should you look for another BETTER external verification solution or continue with what you have and make it work?

The most natural reaction in this situation is to make do with what one has in hand. Keep working the tree with your blunt tool. This is trap. This is what I call the Switching Cost Syndrome. The dilemma of the verification manager is something similar to the dilemma of the woodcutter. Invest time in sharpening the axe or continue to cut the tree with same axe?

You need to step back at this point. Sharpen your axe. Buy a better solution.

So What Does It Take To Overcome The Switching Cost Syndrome?

So how do you know which verification solution would nullify the “Switching Costs”? You need to double down and make sure the next solution you buy plugs all the holes of previously developed internal/ external solutions. You also need to make sure that the new verification solution has inbuilt comprehensive test bench. You should spend absolutely no additional time on making test cases or else it will lead to more delay. In other words, you need a solution that is

1. Complete – ready made test bench with comprehensive test suites and coverage suites
2. Customizable – Does not make your DUT integration itself another project. Comes with hooks to integrate DUT and easily customizable test suites and coverage suites
3. Reuse friendly – Allows easy reuse of already existing DUT specific test bench components along with the vendor supplied verification component
4. Debug-able – Easy to debug the failures with excellent logging and support of revolutionary debug tools
5. Mature – should have been deployed at multiple places
6. Fanatical Support – Promised support turn around time of 24 hours

With a new complete verification environment you have better chance of convergence than continuing to write tests to fill never ending discrete verification holes.

Say YES to sharpening your axe! Get your tape out on time!

Let us know in the comments below what you think about VIP switching costs. Would be great to know your views!

This entry was posted on Monday, October 6th, 2014 at 2:33 am. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.