About the Cover

The cover of the first edition of Writing Testbenches featured a photograph of the collapse of the Quebec bridge (the cantilever steel bridge on the left) in 1907. The ultimate cause of the collapse was a major change in the design specification that was not verified. To save on construction cost, the engineer in charge of the project increased the span of the bridge from 1600 to 1800 feet, turning the project into the longest bridge in the world, without recalculating weights and stresses.

In those days, engineers felt they could span any distances, as ever longer bridges were being successfully built. But each technology eventually reaches its limits. Almost 100 years after its completion in 1918 (after a complete re-design and a second collapse!), the Quebec bridge is still the longest cantilever bridge in the world. Even with all of the advances in civil engineering and composite material, cantilever bridging technology had reached its limits.

You cannot realistically hope to keep applying the same solution to ever increasing problems. Even an evolving technology has its limit. Eventually, you will have to face and survive a revolution that will provide a solution that is faster and cheaper.

Replacing the Quebec bridge with another cantilever structure is estimated to cost over $600 million today. When it was decided to span the St-Lawrence river once more in 1970, the high cost of a cantilever structure caused a different technology to be used: a suspension bridge. The Pierre Laporte Bridge, visible on the right, has a span of 2,200 feet and was built at a cost of $45 million. It provides more lanes of traffic over a longer span at a lower cost and weight. It is better, faster and cheaper. The suspension bridge technology has replaced cantilever structures in all but the shortest spans.

Directed testcases, as described in the first edition, were the cantilever bridges of verification. Coverage-driven constrained-random transaction-level self-checking testbenches are the suspension bridges. This methodology revolution, introduced by hardware verification languages such as e and OpenVera and as described in the second edition of Writing Testbenches, make verifying a design better, faster and cheaper. Hardwave verification languages have demonstrated their productivity in verifying today’s multi-million gate designs.

SystemVerilog brings the HVL technology to the masses, as a true industry standard, with consistent syntax and simulation semantics and built in the simulators you already own. It is no longer necessary to acquire additional tools nor integrate different languages. Like the Pierre Laporte Bridge, which today carries almost all the traffic across the river, you should use these productive methods for writing the majority of your testbenches.

I’m hoping, with this new book, to facilitate your transition from ad-hoc, directed testcase verification to a state-of-the-art verification methodology using a language you probably have at your fingertip.

The author, age 1.5, at the Quebec Bridge in the spring of 1965.