Issue link: https://iconnect007.uberflip.com/i/1299286
12 PCB007 MAGAZINE I OCTOBER 2020 think we learned a lot along the way. But it's not unusual for a stan- dard to have a short life or even be skipped. It may take a long time; it may have a very short cycle or du- ration, and then the market may jump to adopting the next genera- tion quickly. I saw that in PCI Express. It took almost six or seven years to come up with one generation, and the next generation came within a year and a half because there was so much detail learning that was done while its previous generation was being defined. Moving from one generation to the next was so challenging that they had to do a lot of due diligence to get all the information that was used for the generation after. Then, they started looking at all the different angles, and those areas opened up new insight in terms of improvement and maybe even skip- ping a generation. It's not unusual for some of these interfaces or definitions to be short-lived after it's taken so long to implement or define, and then the next step takes over quickly. Holden: Why did the IEEE invest all this time in creating the HIR? Because I don't remember IEEE roadmaps before this. The overall con- sensus is that this is so important and going in many different directions that we need to have standards, and the IEEE pioneered and champi- oned the standards. And whatever those stan- dards are, the HIR roadmap indicates a greater need for those standards. Otherwise, we're go- ing to have a Tower of Babel, and costs aren't going to come down. Horner: Correct. If you look at the HIR, there's a lot of collaboration, even between the sub- topics. Each has its own focus, but it's all un- der the same umbrella so that they can benefit from each other's knowledge and shared learn- ings. Holden: If we're going to have multinational standards, without this kind of cooperation, people will create their own. Horner: Like when people can't speak the same language, misun- derstanding happens. If the two ends of the interface don't follow the same specification, communi- cation will not be effective. Dan Feinberg: And just like with any other international group, that doesn't mean that what they're doing is going to be agreed upon by everybody. Horner: Our CEO, Aart de Geus, gave a key- note at SEMICON West talking about global- ization and coming together, and he said that we all need to collaborate and make this world a better place. There are many angles on many fronts. Nolan Johnson: The collaboration that's neces- sary to make HIR type components work in the future becomes a part of the whole design pro- cess to make the design work. That's going to change the role of the fab house, and they're going to have to be more consultative and in- volved in the design process. Horner: This die-to-die integration packag- ing is not going to be cheap. It's going to get more expensive as it becomes more complex. It needs to be collaborative to get the optimal solution, and that's the main driver for col- laboration. The IC designer can design in their own IC silo, the board designer can design their board, and the package designer can de- sign their package independent of the other disciplines. You cannot get to the market fast enough if you have to wait for the IC to be de- fined, designed, and characterized, and then design a package for it. That's how it used to be done. We can't afford to wait for a sequen- tial implementation. Johnson: That's going to change the way we do business. OEMs will need to line up with the fabricators early, be a part of the engineer- ing team, and then manufacture it once they're done. Rita Horner