European CAVE Workshop: Predictions for the Early 90s

(Made in Early 1987)

Commentary (1996):

Inspired by the efforts of their CANDE colleagues, the European CAVE Workshop participants made a similar set of predictions in 1987 for the early 1990's (1992 being the target year.) Included below are the Top 20 predictions, in order. Prediction (5) is of particular note, since Synopsys, Trimeter, and Silc were just off the ground and Cadence, Mentor and Valid were not yet convinced!

Predictions:

1. ICs will be designed by systems people.

2. Silicon Compilers (including analog) will go from a behavioural specification to architecture specific compilers and will be widely used.

3. UNIX will dominate, whether we like it or not.

4. Simulations at the highest level will be a major problem.

5. Logic synthesis will be generally applied.

6. One month Silicon will be available from European Semiconductor companies.

7. Smart architectures will be more important than fast technology.

8. ASICs will become the next VLSI wave.

9. Tools for higher levels of abstraction will be more important.

10. General Purpose parallel computers will replace special purpose accellerators and will be "wildly" used.

11. Silicon Compilers will "think" about testing.

12. Lack of professionals in CAD and design still a problem.

13. CAD systems will be a hierarchy of tools combining rule-based and procedural approaches.

14. Ocean of gates will replace standard gate arrays.

15. Foundaries and design houses will separate.

16. Test solved by hiding the problem from the designer.

17. ASICs will still be small - 95% < 10K gates.

18. CAD tools targeted at systems designer.

19. First real use of formal verification to reduce simulation effort (but both).

20. Time to market will dominate area and performance consideration.

Commentary: (9/2001)

Looking at these predictions from the post dot-com crash world of 2001, two things seem to stand out: (a) our European colleagues seem to have been looking much farther ahead than the traditional five-year window at CANDE. System design and higher levels of abstractions seem to be much more on their minds (Predictions: 1, 4, 7, 9). Ten years later, we are beginning to see a genuine call for complete system designs on chip, simulation at system levels is complicated by the heterogeneity of components (Analog, RF/Baseband, chemical/mechanical, going forward). (b) Some of the less popular predictions (10-20) seem to have been right on the money. For future predictions we should list all the predictions, just to make sure that a contrarian view is presented.