What Interested You In 2020

In business you are always told to follow the money, but for us it is more important to follow the readership. If we are not writing what you want to read, then we are missing the mark. I like to review the ones that have garnered the most attention, in part to see if that will influence what I write about for 2021, but also to find out where the industry is looking for the most help.

As Semiconductor Engineering grows and we expand the number of channels, it becomes increasingly difficult to keep track of what is being read across all of them, and so I will concentrate on the two channels that I write for – Low Power-High Performance and Systems and Design.

As always, I am surprised that this process does not favor articles published earlier in the year, but it seems to be a fairly common practice that articles get most of their hits within the first month after publication and then slowly drift up over time thanks to search, as well as focused research on various topics within our Knowledge Center.

There are always exceptions, and it is surprising when an old article suddenly re-ignites, often within a different group of people and outside the initial readership. That says to me that we still have a lot of room to expand.

Systems and Design

Fifth place goes to “New Architectures, Much Faster Chips,” published in August. This piece, written by Ed Sperling and Ann Steffora Mutschler, looked at the innovations that are driving orders of magnitude improvement in performance.

Fourth place goes to a February post entitled “The Challenges Of Building Inferencing Chips,” which Ann Steffora Mutschler wrote. It examined advancements in AI and the different approaches to inferencing that are being developed. It posited that not all of them will work.

Third place, and all the ones above it, approached the subject of open source. “Open-Source Hardware Momentum Builds” was written by me in June and examined the impact that RISC-V is having on the open-source hardware market and contrasts it to open-source software.

Second place was a March article, also written by me, entitled “Why It’s So Hard To Create New Processors.” It explored the interest in creating new processors following the success of RISC-V and the potentially daunting verification challenges that go along with that.

The grand prize for 2020 goes to a September article written by Ed Sperling, entitled “RISC-V: What’s Missing And Who’s Competing.” This was an ‘Experts at the Table’ discussion that looked at the state of the tool suite and the uncertainties in the business model.

Low Power-High Performance

Fifth place goes to a March article entitled “HBM Issues In AI Systems.” It looked at the advancements that have been made possible by adoption of HBM and the basic relationships between compute and memory access.

Fourth place was for the June article “Interconnect Challenges Grow, Tools Lag.” It examined the historical limitations cause by interconnect systems and how it is happening again today.

Third place goes to Ann Steffora Mutschler’s May article, entitled “‘More Than Moore’ Reality Check.” She examined some of the gaps in tooling and methodologies that would enable the industry to make full use of multi-die packages.

Second place was for “Increase In Analog Problems,” published in October. It examined what may be an anomaly in survey data, or an indicator that the industry is hitting a major issue with analog failures.

First place, and by a very wide margin, goes to another June article titled “Aging Problems At 5nm And Below.” It posited that semiconductor aging has moved from being a foundry issue to a user problem. At 5nm and below, vectorless methodologies become too inaccurate.

Conclusion

There are three subject areas that dominated the rankings: Artificial Intelligence (AI), advanced packaging, and RISC-V. I predict that AI will be even hotter in 2021 due to the number of new AI accelerator chips that will be coming to market. We had the first few in 2020 and as we see more of the architectures and the tradeoffs that have been made, I am sure that will prompt a lot more discussion.

Advanced packaging is becoming inevitable for many cutting-edge designs and while it may still be too early to call 2021 the year of the chiplets, I think we will get a lot closer to that becoming a reality. We still have to overcome a bunch of hurdles to get there.

The third subject area was RISC-V. I believe that it will go down in the readership polls next year because people will have their heads down trying to make it work. There will be successes and failures, but lower reader interest does not imply less success. In fact it may well become the really hot topic for 2022, when those efforts start showing more results.

Are there other areas that you would like to see us do a deep dive on, or to get commentary from the industry about? I would love to know what your biggest concerns are and where you might like help or guidance that we can provide by having a wider view of the industry. And as always — thanks for reading.