The Past Predicting The Future

It is often said that you cannot predict the future by looking at the past, but that isn’t always correct. There are many clues provided by digging into change. Those changes are a prelude to what may happen in the future. One way we can do that here at Semiconductor Engineering is by looking at changes in reading habits. What types of articles are attracting the most attention? This is a sure indication that these are the topics that are impacting the work that readers are doing today, that will lead to products in the future. Also, just as interesting are topic areas that are seeing less readership than in the past.

Declining interest

Let’s start with the topics that have dropped. It is a surprising list that includes RISC-V, power optimization, verification, and open source. My guess is that these are going through a consolidation phase after being very hot topics for the past few years.

Having said that, one article on RISC-V saw the fourth-highest readership numbers, High-Level Synthesis For RISC-V. It is the extensibility of RISC-V that is driving a lot of design activity. The challenge is designing and implementing custom processors without having to re-implement them every time at the register transfer level (RTL). Abstraction is the key to custom processor design and verification, but defining the right language and tool flow is a work in progress.

Power ticks up a little in the rankings when it also involved AI, with 11 Ways To Reduce AI Energy Consumption. As the machine-learning industry evolves, the focus has expanded from merely solving the problem to solving the problem better. “Better” often has meant accuracy or speed, but as data-center energy budgets explode and machine learning moves to the edge, energy consumption has taken its place alongside accuracy and speed as a critical issue.

A positive outlier in verification was The Verification Mindset. The practice of semiconductor verification has changed substantially over the years, and will continue to do so. The skillset needed for functional verification 20 years ago is hardly recognizable as a verification skillset today. The same should be expected moving forward, as design and verification become more abstract, the boundary of what is implemented in hardware versus firmware and software continues to move, along with the adoption of new technologies. A move toward a complete verification spectrum is a constant response to the steady increase in complexity and time-to-market pressures.

Staying constant

Some categories continue at a similar pace compared to previous years. Those include AI and ML, semiconductor advances, analog, and articles about the general market. This category had the highest number of highly read articles, but few that got into the top 10. The one with the highest interest level was Architectural Considerations For AI. Custom chips, labeled as artificial intelligence (AI) or machine learning (ML), are appearing on a weekly basis, each claiming to be 10X faster than existing devices or consume 1/10th the power. Whether that is enough to dethrone existing architectures, such as GPUs and FPGAs, or whether they will survive alongside those architectures, isn’t clear yet.

Chiplets continue to be a subject that many are keeping an eye on, such as with Chiplets For The Masses. Chiplets are a compelling technology, but so far they are available only to a select few players in the industry. That’s changing, and the industry has taken little steps to get there, but timing for when you will be able to buy a chiplet to integrate into your system remains uncertain.

Big gainers

Two categories were very big this year. One of those was computer architecture, which had the most articles that appeared in the top list, and some of the highest numbers. The other subject was memory, which also had a lot of highly read articles, especially for what used to be a very sleepy subject.

The two articles related to computer architecture held the number one and two positions for the year. The top article was Von Neumann Is Struggling. The world has changed from being control-centric to one that is data-centric, pushing processor architectures to evolve. Venture money is flooding into domain-specific architectures (DSA), but traditional processors also are evolving. For many markets, they continue to provide an effective solution.

In the second spot for the year was Data Centers On Wheels. Automotive architectures are evolving quickly from domain-based to zonal, leveraging the same kind of high-performance computing now found in data centers to make split-second decisions on the road.

Also in the top 10 was Designing Chips In A ‘Lawless’ Industry. The guideposts for designing chips are disappearing or becoming less relevant. While engineers today have many more options for customizing a design, they have little direction about what works best for specific applications or what the return on investment will be for those efforts.

Another very popular article was Software-Hardware Co-Design Becomes Real. For the past 20 years, the industry has sought to deploy hardware/software co-design concepts. While it is making progress, software/hardware co-design appears to have a much brighter future. Automatic mapping of software onto existing hardware, or using software to drive hardware design, are highly desired but very difficult.

The last article to spotlight in this category is Servers Are Becoming More Heterogeneous. The number of CPUs in a server is growing, and so is the number of vendors that make those processors. In the most recent Top500 supercomputer list, 140 of the supercomputers had Nvidia GPU co-processors, and that number will only grow. Within the next 5 to 10 years, general servers will be shipping with x86 processors, GPUs, FPGAs, Arm cores, AI co-processors, 5G modems, and networking accelerators.

Two articles on memories stood above the rest. One is, Has Computational Storage Finally Arrived? In a nutshell, computational storage brings processing power to the storage level. It eliminates the need to load data from the storage system into memory for processing. Moving data between storage and compute resources is inefficient and computational systems, while growing rapidly, can’t keep up with ever-expanding datasets.

The second article is, Will Monolithic 3D DRAM Happen? As DRAM scaling slows, the industry will need to look for other ways to keep pushing for more and cheaper bits of memory. The most common way of escaping the limits of planar scaling is to add the third dimension to the architecture. There are two ways to accomplish that. One is in a package, which is already happening. The second is to scale the die into the Z axis, which has been a topic of discussion for a while.

There are also some articles that don’t fit neatly into general categories, but were in the top 10. One was Impact Of GAA Transistors At 3/2nm. The chip industry is poised for another change in transistor structure as gate-all-around (GAA) FETs replace finFETs at 3nm and below, creating a new set of challenges for design teams that will need to be fully understood and addressed.

Reliability is becoming a concern for an increasing number of people, as evidenced in Design For Reliability. Circuit aging is emerging as a mandatory design concern across a swath of end markets, particularly in markets where advanced-node chips are expected to last for more than a few years. Some chipmakers view this as a competitive opportunity, but others are unsure we fully understand how those devices will age.

Finally, and third place for the year, was Chipmakers Getting Serious About Integrated Photonics. Integrating photonics into semiconductors is gaining traction, particularly in heterogeneous multi-die packages, as chipmakers search for new ways to overcome power limitations and deal with increasing volumes of data.

As always, there is a slight bias toward articles that appeared earlier in the year, but in general that does not appear to disadvantage articles that much, except perhaps for those that only got published a couple of weeks ago.