From RISC-V to 3D NAND Technology—Looking Back at the Trends of 2022
Catch up on some of the technology and industry trends we’ve noticed from 2022.
This last year, 2022, has been filled with tons of new technology and events that made the year fly by. Sometimes, it’s important to look back at everything that happened this year to get a picture of where next year may be heading or to see what the industry is focusing on.
In this article, we’ll take a look at some of the top trends we’ve seen this year.
A Rivalry Over 3D NAND Layers
Sometimes trends spread out over the year, but one certainly stood out above the rest, with companies releasing back-to-back news on NAND technology. This trend came as a bit of a rivalry, as companies one-upped each other on the number of layers and innovations they could pack into NAND memory.
The first release to hit our radar was Micron’s 3D NAND flash memory chip, which had 232 layers packed onto it. This chip came as a bit of a surprise, considering the last time Micron announced how many NAND layers it had reached was 176 back in 2020.
A rendering of Micron's latest 3D NAND with 232 layers. Image used courtesy of Micron
What was even more of a surprise was that, a week later, SK hynix pushed beyond Micron by creating a 238-layer 4D NAND. The term “4D” is coined by SK hynix, and the difference is that, in a traditional 3D NAND, the periphery circuit is next to the array, while 4D NAND places it underneath.
An overview of SK hynix's 4D NAND structure vs. a traditional 3D NAND. Image used courtesy of SK hynix
Building up this trend, another release that same week was Yangtze Memory Technologies Co Ltd’s (YMTC) 232-layered X3-9070 3D NAND chip. A week after that, Samsung announced its V-NAND-based solid-state drives (SSDs) for gaming applications. To learn more about V-NAND, you can check out this article.
All of these releases show a clear interest in the industry's investment in NAND memory technology. As companies seek to increase memory density, it appears that they are looking towards 3D NAND because of the ability to space out transistor cells to avoid leakages and interference from adjacent cells. This can allow designers to increase memory density without sacrificing performance, thus side-stepping the performance limitations of scaling technology.
With so many companies racing to push beyond NAND limitations, it will be interesting to see if more innovations pop up in 2023. I wonder how many layers the next release will be.
RISC-V Takes Huge Leap Forward in 2022
Since its creation in 2010, RISC-V, an open-source ISA (instruction set architecture), has taken the world by storm. This architecture has found its way into all sorts of nooks and crannies across the electronics industry. From helping ease the ongoing chip shortage to finding its way into space-rated technology, RISC-V is proving to be an ongoing trend year over year, and it doesn’t look like that's about to change anytime soon.
Block diagram of SiFive's RISC-V architecture, which NASA will use in its high-performance spaceflight computer (HPSC) project. Image used courtesy of SiFive
One company that made massive strides for RISC-V this year was SiFive. Over the years, All About Circuits has covered SiFive since 2016. What started with an article or two a year has ramped up this year thanks to a plethora of new technology being released. As mentioned a moment ago, RISC-V found its use in space technology with NASA tapping SiFive to leverage its X280, an 8-core RISC-V processor, in a new project.
Adding to SiFive’s announcements for the year, it also released:
- A new version of the X280, which was teased at Embedded World 2022
- Two processors—the P670 and P470—which added to its Performance line of RISC-V IP-based processors
- A portfolio of automotive RISC-V processor cores
With all of these innovations coming out of SiFive this year, it feels obvious that this startup is working toward becoming a major cornerstone of RISC-V. Seeing how quickly this technology is growing, major companies are jumping on board this year as well. Take Intel, for example, which is showing its support by launching the new Intel Pathfinder program. This program is designed to help engineers in the pre-silicon development of RISC-V devices.
Another major advancement in RISC-V history this year was the announcement of Alibaba’s RISC-V SoC (system-on-a-chip) powering the ROMA development laptop, thus making this the industry's first laptop using a RISC-V chip.
The RISC-V community ended another year with a bang in the form of its annual RISC-V Summit event. Our All About Circuits team (virtually) attended the event. This enabled us to cover interesting RISC-V Summit keynotes, a RISC-V processor core announcement from startup Ventana, and other RISC-V news unveiled at the event.
Looking back at all of these RISC-V news items from this year (there were still plenty more I didn't touch on here), it's apparent that RISC-V made some enormous leaps in becoming a mainstream technology. Entering into 2023, I can't see its momentum slowing down anytime soon.
Quantum Computing Aims for Commercialization
One trend that has been consistently growing year over year is quantum computing. What was once largely based on university research (which is still pumping out new quantum tech) has become part of many big-name companies’ strategies for the future. Take Intel, for example, which took a “software first” approach this year to quantum computing using its platform based on Loihi 2.
Another big name, IBM, which is one of the biggest players in the quantum computing world, also had its fair share of the spotlight this year. Notably, in November, IBM announced that it had created a 433-qubit quantum processor called “Osprey.” This processor claims to have tripled the number of qubits over IBM’s Eagle processor released in 2021. Along with this processor, IBM also revealed its latest roadmap and goals for 2023, which you can read about in this article.
IBM's Osprey quantum processor. Image used courtesy of IBM
Seeing a year-over-year improvement like this example from IBM helps show that quantum computing isn’t just a trendy new technology; it is quickly gaining ground in the mainstream market thanks to some startups. Throwing their hats into the quantum computing ring this year, many startups made waves that could lead to more advancements in 2023, especially concerning the commercialization of quantum computing.
Take one U.K.-based company, Quantum Motion, which, back in October, demonstrated that its “Bloomsbury” quantum chip could be created using the same manufacturing processes for standard semiconductors. This announcement followed Intel’s notification that same month that it hit a new milestone toward producing silicon-based qubits. These efforts kick off the idea that quantum computing technology is steadily heading toward mass production.
As the commercialization and manufacturing limitations of quantum computing hardware continue to be at the forefront of the industry, it will be exciting to see how this unfolds throughout the new year.
Wi-Fi 6/6E Keeps on Trucking While Wi-Fi 7 Hardware Debuts
Released back in 2021, the Wi-Fi 7 standard had a pivotal year in 2022. Though companies are still creating and focusing on technology working within the Wi-Fi 6 standard (released in 2019), some companies like Qualcomm are already looking to the future.
One reason, as contributor Robert Keim points out, is that, though Wi-Fi 6 focused on driving data to as many devices as possible and had advanced capabilities, it has some shortcommings when it comes to latency and data rates.
As mentioned, Qualcomm has been a driving force in pushing towards Wi-Fi 7 with its Wi-Fi 7 Networking Pro platform, which was released back in May. Along with that platform, Qualcomm also announced new RFFE (radio frequency front end) modules to fit within this standard. Another company, MediaTek, is also getting on board with the release of its Filogic 880 and Filogic 380 Wi-Fi 7 platforms.
An overview of Qualcomm's Wi-Fi 7 platform. Image used courtesy of Qualcomm
Since the shift towards Wi-Fi 7 has barely begun, companies are still innovating around Wi-Fi 6. Back in January of this year, the Wi-Fi Alliance released the Wi-Fi CERTIFIED 6 Release 2, which affects the Wi-Fi 6E version.
Overall, this release was aimed at the Internet of Things (IoT) to increase power efficiency and performance, and designers can now leverage this standard in their devices while Wi-Fi 7 technology is still trying to break into the industry. All in all, 2023 is sure to have more news about both Wi-Fi 6/6E and Wi-Fi 7.
In-memory Compute Shakes Off Memory Limitations
A final trend we’ve seen in 2022 is in-memory compute. This one has been big the past few years and has continued into this year. As artificial intelligence (AI) continues to be integrated into more and more devices, it comes with a few limitations, namely attributes like speed, processing, and storage capacity. To combat these challenges, engineers are looking for better memory architectures.
Though there is still a lot to be accomplished with in-memory computing, some universities are trying to break the mold with their research.
Take the NeuRRAM from UCSD, Stanford, Tsinghua, and Notre Dame. The NeuRRAM is a resistive random access memory (RRAM) compute-in-memory (CIM) neuromorphic chip. This chip made strides in the space of in-memory processing by having “48 neurosynaptic cores, 256 CMOS neurons, and 65,536 RRAM cells,” all of which can do parallel processing and support data and model parallelism.
The NeuRRAM chip. Image used courtesy of UCSD
Another release targeting in-memory compute that happened this year came from the University of Pennsylvania (UPenn). Researchers from UPenn showed how analog in-memory compute circuits could be created using specialized materials and without transistors. This architecture leveraged aluminum scandium nitride (AlScN) to create memory blocks and a convolution neural network (CNN) using a ferroelectric diode (FeD) array. This combo helped the researchers demonstrate that in-memory computing is reachable and that the von Neumann bottleneck can be bypassed.
Aside from university research into in-memory compute, some companies are trying to take it to a more mainstream level. Take the startup Axelera AI, which got funding this fall to commercialize its AI acceleration platform, combining multicore in-memory compute with a specialized dataflow architecture. You can read up on this company in a startup spotlight article here.
As researchers and companies focus on making in-memory compute a more tangible reality, I’m sure we will see even more developments in 2023.
The Outlook for 2023 is Promising
I know I only pulled apart just a few trends of 2022, but after looking back at even just these five, it’s exciting to see all of the technology that helped us get to another year. These five trends (and many more) have barely started to crescendo. Next year is sure to keep the momentum going and will bring us many new innovations, some we might never have thought previously possible.
Thank you for spending another year with us here on All About Circuits! We couldn't do it without you!
What technology made the biggest impression on you this year? Let us know in the comments down below!