Always Be Curious #211: Neuromorphic computing, artificial vision, and the new Atlas robot
This week in ABC: Intel announced a new neuromorphic computer, brain implants could help restore vision, and Boston Dynamics unleashed a stronger new Atlas robot
While most of the news in this industry revolves around the fierce competition for the nanometers, there are plenty of other cutting-edge developments to keep a close eye on. This week, Intel reveiled "Hala Point”, a neuromorphic computer that contains over 1 billion ‘artificial neurons’ and 128 billion ‘artificial synapses’, spread out across 1,152 specially designed chips. Despite this being an experimental computer, we should still pay close attention, because neuromorphic computers harness an entirely different architecture from the traditional Von Neumann computer archicture. Time to explain! 🍿
Traditional computing has been built on the ideas of John von Neumann, a computer pioneer at Princeton University in the 1940s. In his computing architecture, there’s a clear separation between the central processing unit (CPU), memory storage, and the data bus that connects them. The CPU performs logical computations, while the memory holds both the program instructions and the data needed by those instructions. However, the data bus between the CPU and memory can only transport a limited amount of data at a time. As computers and software became more advanced over time, the CPU's ability to process data increased much faster than the ability to move data in and out of memory. This discrepancy created the (you guessed it) “von Neumann bottleneck”. Simply put, it slows down computation because the processor often has to wait for data to be delivered from memory. Lots of innovations that you might catch in the news, like advanced packaging and High Bandwidth Memory—they help to address this particular challenge.
Neuromorphic computing architecture inherently avoids the von Neumann bottleneck. That’s because neuromorphic systems are inspired by the human brain's structure and function. The human brain is remarkably efficient at processing complex tasks such as learning, pattern recognition, and sensory data processing with minimal energy consumption. Neuromorphic computing attempts to emulate the brain’s neural network structure through artificial neurons and synapses. Neurons are the basic computational units in neuromorphic computing, designed to simulate the functionality of biological neurons. So they receive signals, process them, and can trigger signals to other artificial neurons, creating a network of interconnected units. Synapses form the connections between artificial neurons. In neuromorphic systems, synapses can change their "strength" or weight, much like the synaptic plasticity of the human brain. This ability enables the system to learn, remember and adapt from the data it processes, much like how the brain learns from experiences.
Intel says Hala Point can do 20 quadrillion operations per second. That means it can handle AI tasks 50 times faster and use 100 times less energy than the computers we're used to. It’s still in the experimental stages, but it will now be able to get some real-life practice at Sandia National Laboratories, where it will be cracking physics problems.
Perhaps neuromorphic computing holds the key to a future where we can build advanced AI, maybe even Artificial General Intelligence, in a smarter and more sustainable way than we do now. In biology, there’s a term that might apply here: “Mimicry”, or an “evolved resemblance between an organism and another object, often an organism of another species.” If technology is our next nature, and humanity keeps developing and integrating with technology as we have done over the last 70 years—wouldn’t the most logical next step be that our computers mimic the most remarkable computer of all, the human brain?
Have a good week, stay safe and sound,
👨💻The round-up in sci-tech💡
👁️ The next frontier for brain implants is artificial vision (WIRED)
Elon Musk’s Neuralink and others are developing devices that could provide blind people with a crude sense of sight.
💅 She dreams of pink planets and alien dinosaurs (The New York Times)
Lisa Kaltenegger, director of the Carl Sagan Institute at Cornell University, hunts for aliens in space by studying Earth across time. Nifty!
🔥 The tiny ultrabright laser that can melt steel (IEEE Spectrum)
Photonic crystals are the key to the brightest semiconductor laser ever.
🇳🇱 ‘Muziek is m’n geliefde metafoor’ / ‘Music is my beloved metaphor’ (NRC) 🔐
NRC sat down with Nobel laureate Anne L’Huillier, who visited ASML’s campus in Veldhoven for a presentation about the collaborative research efforts.
🇮🇱 How Israel defended against Iran's drone and missile attack (WIRED)
In today’s world, everybody could use an Iron Dome.
🔮 AI is tech’s ‘greatest contribution to social elevation’ (NVIDIA)
AI promises to bring the full benefits of the digital revolution to billions across the globe, NVIDIA CEO Jensen Huang told Oregon State University students.
🤓This week in chips⚠
🔐 Why chipmakers are investing billions into ‘advanced packaging’ (Financial Times)
By integrating or “packaging” multiple chips — whether of the same kind or different varieties — more closely together, chipmakers can increase speed and efficiency while circumventing the limits of miniaturisation.
💪 Intel completes assembly of first commercial High-NA EUV chipmaking tool — addresses cost concerns, preps for 1.4nm process development in 2025 (Tom’s Hardware)
Transistor shrinking intensifies.
🇳🇱 Semiconductor equipment maker ASML ships second 'High NA' EUV machine (Reuters)
ASML, the biggest supplier of equipment to computer chip makers, said on Wednesday said it has shipped one of its newest "High NA" EUV lithography systems to a second customer.
🧠 Intel builds world’s largest neuromorphic system to enable more sustainable AI (Intel)
Hala Point, the industry’s first 1.15 billion neuron neuromorphic system, builds a path toward more efficient and scalable AI.
🇺🇸 Intel reveals world's biggest 'brain-inspired' neuromorphic computer (New Scientist)
A computer intended to mimic the way the brain processes and stores data could potentially improve the efficiency and capabilities of artificial intelligence models
💰 Samsung pouring $45B into Austin area — one of largest deals in US history (Austin Business Journal)
Samsung is making a big play in Texas.
💪 AI cloud startup TensorWave bets AMD can beat Nvidia (The Register)
Starts racking MI300X systems - because you can actually buy them and they beat the H100 on many specs.
🥇 AMD expands commercial AI PC portfolio to deliver leadership performance across professional mobile and desktop systems (AMD)
AMD announced new products that will expand its commercial mobile and desktop “AI PC” portfolio.
🇰🇷 A sneak peek at HBM cold war between Samsung and SK hynix (EDN)
A fierce competition kicks off between Samsung and SK hynix with transition from HBM3 to HBM3e memory technology.
🔥 Samsung develops industry’s fastest 10.7Gbps LPDDR5X DRAM, optimized for AI applications (Samsung)
Industry-leading features come with 25% higher performance, 30% more capacity and 25% higher power efficiency The new LPDDR5X is the optimal solution for future on-device applications and is expected to expand adoption into PCs, accelerators, servers and automobiles.
🇨🇳🇰🇷 Scientists in China, South Korea tell of ‘crucial stride’ in chip tech (South China Morning Post)
Newly devised strategy could make commercially viable amorphous p-type semiconductors a reality, team in Chengdu and Pohang says in Nature paper.
📈By the numbers📉
📊 ASML announces Q1 2024 financial results (ASML)
"Our first-quarter total net sales came in at €5.3 billion, at the midpoint of our guidance, with a gross margin of 51.0% which is above guidance, primarily driven by product mix and one-offs. We expect second-quarter total net sales between €5.7 billion and €6.2 billion with a gross margin between 50% and 51%. ASML expects R&D costs of around €1,070 million and SG&A costs of around €295 million. Our outlook for the full year 2024 is unchanged, with the second half of the year expected to be stronger than the first half, in line with the industry's continued recovery from the downturn. We see 2024 as a transition year with continued investments in both capacity ramp and technology, to be ready for the turn in the cycle," said ASML President and Chief Executive Officer Peter Wennink.
🥇 TSMC climbs to #1, NVIDIA posts triple-digit growth in final 2023 ranking of top 25 semi suppliers (TechInsights)
No new suppliers broke into the top 25 list in 2023, but there were significant changes in where companies were ranked.
📊 TSMC reports Q1 2024 financial results (TSMC)
“Our business in the first quarter was impacted by smartphone seasonality, partially offset by continued HPC-related demand,” said Wendell Huang, Senior VP and Chief Financial Officer of TSMC. “Moving into second quarter 2024, we expect our business to be supported by strong demand for our industry-leading 3nm and 5nm technologies, partially offset by continued smartphone seasonality.”
📈 TSMC expects 6% revenue growth in 2Q24 (Digitimes)
TSMC expects to post revenues of between US$19.6 billion and US$20.4 billion in the second quarter of 2024, representing a 6% sequential growth and a 27.6% year-over-year increase at the midpoint. Gross margin and operating margin for the first quarter of 2024 are estimated at 51-53% and 40-42%, respectively.
🔮 The 2024 AI Index Report (Stanford University)
This year’s AI Index Report measures and evaluates the rapid rate of AI advancement from research and development to technical performance and ethics, the economy and education, AI policy and governance, diversity, public opinion and more.
❤️For the love of tech❤️
🤖 The Atlas robot is dead. Long live the Atlas robot (WIRED)
Before the dear old model could even power down, Boston Dynamics unleashed a stronger new Atlas robot that can move in ways us puny humans never can.
Always Be Curious is the personal newsletter of Sander Hofman, Senior Creative Content Strategist at ASML. Opinions expressed in this curated newsletter are my own and do not necessarily reflect those of my employer.