CANDE Predictions for the Mid 1990's

(Made in April 1991)

Commentary (1996):

For the first time, we included top thirty predictions from just over a hundred distinct suggestions made in 1991 for the mid-latter part of 1990's. In addition, we have included the "Bottom Ten" of the list, as a control. At the time these predictions were made, the Committee felth that the Top Ten were worthy of endorsement.

When these predictions were made, "MCM CAD" was one of the topics of the workshop, perhaps encouraging the prominence of Predictions (3) and (4). In the final voting, Predictions (1) and (2) led by a large margin, garnering at least one vote from almost all present. I think, we can agree that they are both accurate, although the support cost is no longer as "hidden" as it might have been -- it is a big business! Prediction (8) is arguably quite accurate, and Prediction (10) is not.

Prediction (11) is a miss, while (12) and (13) are accurate. Prediction (14) is accurate at the process level and growing rapidly. Prediction (15) ranked lower than (13); and Prediction (16) is just beginning to take on significant importance. Prediction (17) is certainly true for new codes, while Prediction (18) will received another test this year, and Prediction (19) probably should have said "governmental and industrial consortia" (Ed: This commentary was written almost four years before GSRC. Perhaps Richard Newton had seen it coming!). Predictions (20) to (25) are right on, and with the recent successful IPO for Pure Software Prediction (26) is in the bag. Certainly (28) is true today and despite the successful Metasoftware (H-SPICE) IPO, users still report (29) as very true. And number (30)? As commonplace as we could hope for.

Although not listed below, Prediction (88) was "Ethical Issues Will Surface in CAD" -- a topic of some current interest in the press. (Ed: And once again in 2001, with the Avanti saga..)

Predictions:

1. Hardware/software co-design will be one of the most important design problems.

2. Support will still be the biggest hidden cost for both CAD vendors and customers.

3. MCM CAD becomes a reality.

4. MCM will enable new CAD and semiconductor business.

5. Internal CAD will make a come-back.

6. There will be tools for validation of specifications.

7. Partitioning will emerge as a commercial product.

8. The telecommunications industry will provide the most challenging problems in CAD.

9. SPICE algorithms still dominate circuit simulation.

10. Frameworks will be provided by computer vendors

Commentary: (9/2001)

MCMs did not make it to the mainstream due to cost reasons, along with all the EDA innovations and our predictions (3) (4) related to MCMs. Frameworks are still evolving while hopes for advances in partitioning (especially at higher levels) seem to have been misplaced (or too early).

However, the best part is reading some of the less popular predictions..

11. Logic synthesis performed by ASIC manufacturers, not customers (HDL hand-off).
12. High-level specification and verification will be the largest open problem.
13. There will be no industry-standard framework.
14. Statistical CAD is universally accepted and in use.
15. A single framework will dominate.
16. Packaging will be taken into account by most CAD tools.
17. C++ will dominate.
18. CANDE will have better dinners.
19. CAD R&D will be dictated largely by government consortia.
20. Noise and power distribution will become as important as timing.
21. Analog HDL (AHDL) just became a standard.
22. Automatic layout will still be an active research problem.
23. Formal verification will be for real.
24. 3D analysis of interconnect will become routine.
25. Low-power and hand-held electronics will be an important area.
26. CASE tools will be used widely for CAD tool development.
27. Increased use of design-process-management.
28. Analog CAD experience will help formulate new high frequency digital CAD tools.
29. SPICE will still have convergence problems.
30. Synthesis will make 100K-gate designs commonplace.

In the bottom ten, all except Prediction (92) received votes from at most one person (its proponent?), with (98) through (101) receiving no support at all -- Prediction (92) was voted for by four participants, but was low on all of their lists.

91. CAD companies will still be trying to provide practical multi-level simulation.
92. CFI will start to become a reality.
93. Mentor announces system 8.0 will be finished next quarter.
94. A logic synthesis breakthrough will occur outside of Berkeley.
95. Routing will be a solved problem and no one will worry about it any more!
96. SPICE models will prevail but no simulator will be able to handle them.
97. CAE vendor products will become commodity items.
98. "Simulator backplanes" was an idea whose time has passed.
99. Micromechanical CAD will emerge as the next big CAD area.
100. People will finally understand the difference between mixed-level and mixed-mode simulation.
101. SPECTRE will prelace SPICE.