BY PETE SINGER, EDITOR-IN-CHIEF
On Tuesday morning, Applied Materials hosted a CEO panel discussion with Aart de Geus, PhD, Chairman and co-Chief Executive Officer, Synopsys; Victor Peng, President and CEO, Xilinx; Lisa Su, PhD, President and CEO, AMD; Sanjay Mehrotra, President and CEO, Micron; and Gary Dickerson, President and CEO, Applied Materials. John Markoff, author and former technology writer with The New York Times acted as moderator.
Markoff had some great questions for the panel related to cost per transistor, new materials, life after CMOS, potential surprise technologies, quantum computing, AI and newarchitectures – but his first question was perhaps the most controversial: “Is Moore’s Law dead or alive?” The panel’s mixed response was illuminating:
Art de Geus, Synopsys: It’s completely alive. The discussion on Moore’s Law invariably goes back to the ‘65 document and then people will exactly track where are the economics, where’s the technology? That’s not the way to think about it. I think the way to think about it is Moore’s Law is the behavior of an exponential that had taken economic feedback on the exponential that drove a revolution of what mankind can do. And the reason I’m saying is completely alive is because right now we’re facing another decade or two of amazing opportunities that themselves economically will drive the push for technology. Maybe it’s not exactly the same kind of curve that Moore actually drew, it doesn’t matter. The impact is what matters of the exponential.
Victor Peng, Xilinx: Of course, I have to say the opposite. It’s mostly dead. ‘Not quite dead yet’ as the saying goes. I’m not saying that there won’t be still quite sig- nificant improvements in computing, but it’s going to take a very different form. I’ll talk about Moore’s law in a general sense, where if you take any processor and in every two years you move move it to the next node and you don’t change the archi- tecture, the performance, power and the area will get, we could get better, which is sort of good for cost. Today, you can get definitely you get one, you can get to pick which one you want.., but it’s really hard to get all three. In that more traditional sense of Moore’s law, I think that’s not going to be working. On top of that, it’s been said for a long time, this happens every twoyears, I think clearly the cadence for when the next node comes, it’s taking longer than two years for a variety of reasons, as much economic as technical.
Lisa Su, AMD: John and I were talking this morning about we all used to do. I grew up as a process technologist, so I can’t say Moore’s law is dead, right? It just can’t be (laughter). I think Moore’s law is slowing, but it’s something that we all strive to in terms of how do we get more performance for our products. There are lots of things that you can do in addition to process technology to keep that performance trend going. That’s what we’re all trying to do, whether it’s in hardware or software or systems. That’s the key: how do we keep that performance trend for our industry that’s driving all of the applications going and there’s a tremendous amount of effort to augment Moore’s law.
Sanjay Mehrotra, Micron: For almost a decade, Moore’s law has slowed down significantly and actually has been less relevant. A lot of innovations have been living in memoryand technology, for example, going from 2D NAND to 3D NAND. That has given capability to bring down costs as well as gain more giga- bits per wafer. However, in DRAM, certainly a lot of innovation is happening with high band-width memory, stacking of vertical chips and TSV technologies to gain performance and to get the benefit of higher capacities. But there is no question that Moore’s Law is significantly challenged in memory and storage. If you just look at 10 years ago versus today, what’s happening in NAND as well as in DRAM, year-over-year, year bit growth that you could get from one technology transition to the next technology transition has more than halved now. Certainly, there are challenges. However, engineers are always finding new innovative ways to address memory and technology challenges better from a process point of view or design point of view and architecture point.
Gary Dickerson, Applied Materials: With classic Moore’s law, when you think about shrinking in two dimensions, Victor is correct, it doesn’t deliver it today simultaneous improvements in power, performance area and cost. Over the last few years, that classic approach has really run out of gas. But we’ve talked about five different drivers in this new playbook. One is architecture innovation. Especially with AI and big data and with workload-specific types of applications, in the edge cloud, training and inference, architecture innovation is a huge focus for many different companies. The innovation around new structures – Sanjay talked about 2D and 3D NAND – the playbook in NAND for decades was to shrink in two dimensions and then now we go vertical. It’s a different driver. You hit a physical limit and you hit a cost limit.
Now it’s about new structure and new materials. Also, how you connect the chips together in packaging, heterogeneous integration. That’s another driver. There’s now question that classic 2D scaling has run out of gas, but there are all of these new opportunities in structure innovation, materials innovation architectures and packages. The challenges are significant, but there’s tremen- dous opportunity not looking backwards, at classic Moore’s law 2D scaling, but looking forward with these five year drivers in this new playbook.