The size reduction of computer transistors has been going on since the day they were invented. Thanks to the amazing engineering and theoretical work of thousands of people, we’re all holding in our hands devices that would seem like science fiction just a few decades ago.
But semiconductor manufacturers are never satisfied and keep breaking down barriers to bring us even smaller and more powerful chips. At the nanometer scale we’re at now, things are getting very complicated. Production methods that worked before can be used no longer and new ones have to be developed.
That’s exactly what TSMC is doing right now, as reported by DigiTimes. According to TSMC's latest annual report, the company has begun its research and development efforts towards the 2nm process in 2019, and preliminary studies regarding nodes even smaller than that are underway as well.
If you haven’t been following the developments on the nanometer front, currently, the best smartphone chips are built using the 7nm process. The first ones using the 5nm process are expected later this year (likely Apple’s A14 chip for the new iPhones). After that will come the 3nm one, targeted for 2022-2023, and hopefully, around 2025, we’ll see the first 2nm chips on the market.
In nanometers, the difference is small, but we should look at what it means for transistor density. A 7nm chip can fit 2 times more transistors on the same die as a 14nm one, but a 2nm one will be able to fit 3.5 times more than a 7nm one!
Chips that powerful and that energy-efficient will open up a whole new set of functions our phones will be able to perform. And while that future is still a few years away, it’s steadily coming and it’s exciting!
Scientist managed to change the cubic silicon structure into a hexagonal nanowire structure. (Image Source: Wired)
By changing the cubic silicon structure to a hexagonal nanowire structure, scientists managed to develop photon-based silicon alloy circuitry that may replace electron-based transistors. The photon-based material prevents electron traffic jams, overheating and transmission slowdowns caused by cramming too many transistors.
Intel co-founder Gordon Moore was estimating that CPU transistor counts could double every two years, and so far this proved to be more or less accurate. The problem now is that we are approaching the physical limits of silicon transistors, which cannot really be shrunk down to less than 1 nanometer, so scientists are looking into new materials that can help overcome such restrictions. One of the latest inventions in this field comes from the Netherlands in the form of light-emitting silicon circuitry that could help reduce the number of needed transistors while still improving the performance.
A group of Eindhoven University of Technology scientists led by Erik Bakkers managed to develop silicon alloy nanowires that can emit light. These new materials can potentially be used to build photon-based circuits that, in turn, may replace the electron-based cubic transistor models used in commercial chips nowadays.
The photon-based hexagonal nanowires should greatly improve the computational performance as data transmitted through photons instead of electrons can prevent electron traffic jams, overheating and transmission slowdowns caused by cramming too many transistors in a determined area. Photons move considerably faster than electrons, so they can transmit data more efficiently over multiple channels. With the introduction of photonic circuits, data-intensive applications like machine learning and big data crunching could see tremendous boosts.
Researchers are currently trying to figure out a way to implement a tiny laser to act as the light source within the new materials. It is yet unclear how long it will take for this technology to become a reality. Intel is planning to reach the 1 nm transistor size limit by the end of this decade and TSMC may do it even faster, so scientists ought to come with a decent solution by 2025.
The machine used to fabricate the new hexagonal silicon alloy nanowires (Image Source: Wired)
Simon Segars expects Arm's partners to ship 50 billion chips in the next two years.
Image Credit: Dean Takahashi
Semiconductor and software design company Arm is doubling down on edge AI hardware, a market that’s expected to be worth $1.15 billion by 2023. It today announced two new AI-capable processors — the Arm Cortex-M55 and Ethos-U55, a neural processing unit (NPU) — designed for internet of things (IoT) endpoint devices, alongside supporting software libraries, toolchains, and models. The company claims that the two chips, which are expected to arrive in market in early 2021, together will deliver an uplift of up to 480 times in machine learning performance in certain voice and vision scenarios.
“[Machine learning] processing on low-power endpoint devices is critical to realizing the full potential of AI for IoT,” wrote the company in press materials. “An extended range of advanced hardware capabilities is required to enable innovation and scale.”
The Cortex-M55, the newest member of Arm’s Cortex-M processor portfolio of cost-optimized and power-efficient microcontroller devices, delivers up to a 15 times uplift in AI performance on its own compared with previous Cortex-M generations, as well as custom instructions and configuration options. Like the Ethos-U55, it’s available in a reference design — Corstone-300 — that ships with a number of secure subsystems and a toolkit with which to build secure embedded systems.
The Cortex-M55 also has the distinction of being the first system-on-chip based on Arm’s Helium technology, an extension of the Armv8.1-M architecture that is optimized for low-power silicon and adds over 150 new scalar and vector instructions. The integer Helium enables efficient compute of 8-bit, 16-bit, and 32-bit fixed-point data, the last two of which are widely used in traditional signal processing applications such as audio processing. As for the 8-bit fixed-point format, it’s common in machine learning processing such as neural network computation and image processing, complementing floating point data types including single-precision floats (32-bit) and half-precision floats (16-bit).
According to Arm, Helium enables a performance boost of up to 5 times for digital signal processing and 15 times for machine learning with the Cortex-M55. It additionally allows for advanced memory interfaces to provide speedy access to machine learning data, and it builds in Arm’s TrustZone systemwide embedded security technology.
Ethos-U55 — which is intended to be paired with a Cortex-M processor like the Cortex-M55, Cortex-M33, Cortex-M7, or Cortex-M4 — packs between 32 and 256 configurable compute units capable of achieving up to 32 times machine learning performance uplift compared with the Cortex-M55. Pitted against the Cortex-M7, Arm claims that the Ethos-U55 and Cortex-M55 are up to 50 times faster in terms of speed to inference and up to 25 more energy efficient in tasks like voice activity detection, noise cancellation, two-mic beamforming, echo cancellation, equalizing, mixing, keyword spotting, and automatic speech recognition.
On the software side, both the Ethos-U55 and Cortex-M55 benefit from a unified software development flow that folds embedded code, digital signal processor code, and AI model code into one. Importantly, it plays nicely with popular machine learning frameworks like Google’s TensorFlow and Facebook’s PyTorch, plus Arm’s own solutions.
The unveiling of the Ethos-U55 comes after Arm took the wraps off of the Ethos-N57 and Ethos-N37, which both feature voice recognition and always-on capabilities. (Ethos, which launched recently, is a product suite focused on solving complex AI compute challenges with sensitivity to battery life and cost.) Around the same time in October, the company revealed its roadmap for Mbed OS, the embedded software operating system that’s meant to serve as a foundation for smart and connected devices.
Madikeri is a hill town in southern India. Framed by the Western Ghats mountain range, it’s known for the Raja’s Seat, a simple monument overlooking forests and rice paddies. In the center, the 17th-century Madikeri Fort features 2 stone elephants at the entrance. Nearby, the domed Omkareshwar Temple is dedicated to the Hindu deity Shiva. To the northwest, cascading Abbey Falls is surrounded by spice plantations.
When to visit
Madikeri has a tropical climate. Temperatures are warm year-round, and monsoon rainfall can be heavy Jun–Aug. The relatively dry, mild months of Dec–Apr are a common time to visit. The 10-day Madikeri Dasara festival (dates vary) features colorful processions and temple celebrations throughout the town.
