Hardware Development Trends

What if a chip didn’t have to be final? That’s the question Ross Freeman asked in 1984—at a time when ASICs ruled the silicon world. Back then, hardware was rigid. Changing logic meant redesigning the entire chip, waiting weeks for fabrication, and burning through budget. But Ross had a radical idea: What if we could build a chip whose function could be reprogrammed even after manufacturing? The result? He co-founded Xilinx and created the world’s first Field-Programmable Gate Array (FPGA). The first chip—XC2064—had just 64 configurable logic blocks (CLBs). It wasn’t fast, it wasn’t cheap, and it certainly wasn’t mainstream. But it worked. Engineers could now implement, test, and modify logic in hardware without going back to the fab. For the first time, hardware had software-like flexibility. In an era dominated by fixed-function chips, that was heresy. FPGAs slowly found a home in applications where change was constant: ◦ Telecom, where protocols evolved rapidly ◦ Aerospace, where reconfigurability mid-mission was vital ◦ Prototyping, where time-to-market could be shortened dramatically Over time, Xilinx expanded its product lines: □ Spartan for cost-sensitive markets □ Virtex for high-performance applications □ Zynq for integrating CPUs and logic fabric But Ross Freeman didn’t live to see the impact. He passed away in 1989—just five years after Xilinx was founded. His invention went on to power: ◦ Telecom backbone equipment ◦ Satellites and spacecraft ◦ Industrial and automotive controllers ◦ Rapid prototyping for nearly every ASIC and SoC team on the planet And in 2022, Xilinx was acquired by AMD for $49 billion—one of the largest deals in semiconductor history. Not bad for an idea everyone thought was too slow, too costly, and too complicated. Sometimes, true innovation isn’t faster or smaller—it’s more flexible. And sometimes, the riskiest ideas become the foundations we all build on. #Semiconductors #Xilinx #FPGA #HardwareDesign #ChipDesign #StartupHistory #VLSI #TechInnovation #EDA

Today, Science Robotics has published our work on the first drone performing fully #neuromorphic vision and control for autonomous flight! 🥳 Deep neural networks have led to amazing progress in Artificial Intelligence and promise to be a game-changer as well for autonomous robots 🤖. A major challenge is that the computing hardware for running deep neural networks can still be quite heavy and power consuming. This is particularly problematic for small robots like lightweight drones, for which most deep nets are currently out of reach. A new type of neuromorphic hardware draws inspiration from the efficiency of animal eyes 👁 and brains 🧠. Neuromorphic cameras do not record images at a fixed frame rate, but instead have the pixels track the brightness over time, sending a signal only when the brightness changes. These signals can now be sent to a neuromorphic processor, in which the neurons communicate with each other via binary spikes, simplifying calculations. The resulting asynchronous, sparse sensing and processing promises to be both quick and energy efficient! 🔋 In our article, we investigated how a spiking neural network (#SNN) can be trained and deployed on a neuromorphic processor for perceiving and controlling drone flight 🚁. Specifically, we split the network in two. First, we trained an SNN to transform the signals from a downward looking neuromorphic camera to estimates of the drone’s own motion. This network was trained on data coming from our drone itself, with self-supervised learning. Second, we used an artificial evolution 🦠🐒🚶♂️ to train another SNN for controlling a simulated drone. This network transformed the simulated drone’s motion into motor commands such as the drone’s orientation. We then merged the two SNNs 👩🏻🤝👩🏻 and deployed the resulting network on Intel Labs’ neuromorphic research chip "Loihi". The merged network immediately worked on the drone, successfully bridging the reality gap. Moreover, the results highlight the promises of neuromorphic sensing and processing: The network ran 10-64x faster 🏎💨 than a comparable network on a traditional embedded GPU and used 3x less energy. I want to first congratulate all co-authors at TU Delft | Aerospace Engineering: Federico Paredes Vallés, Jesse Hagenaars, Julien Dupeyroux, Stein Stroobants, and Yingfu Xu 🎉 Moreover, I would like to thank the Intel Labs' Neuromorphic Computing Lab and the Intel Neuromorphic Research Community (#INRC) for their support with Loihi (among others Mike Davies and Yulia Sandamirskaya). Finally, I would like to thank NWO (Dutch Research Council), the Air Force Office of Scientific Research (AFOSR) and Office of Naval Research Global (ONR Global) for funding this project. All relevant links can be found below. Delft University of Technology, Science Magazine #neuromorphic #spiking #SNN #spikingneuralnetworks #drones #AI #robotics #robot #opticalflow #control #realitygap

The West has a blindspot when it comes to alternative CPU designs. We’re so entrenched in the usual x86, ARM, RISC-V world, that most people have no idea what’s happening over in China. LoongArch is a fully independent ISA that’s sorta MIPS…sorta RISC-V…and sorta x87! — Of course, Loongson (the company) realizes that most software is compiled for x86 and ARM. Thus, they decided to add some hefty translation layers (LBT) built into the hardware. LBT gives you for extra scratch registers, x86+ARM eflags, and an x87(!) stack pointer. — LoongArch is a *hefty* ISA; about ~2,000 instructions. To put it in perspective, base RISC-V is like 50. That said, it’s pretty clean to read. All instructions are 32 bits, and there are only 9 possible formats.  Certainly easier to decipher than modern x86. — I just think it’s fascinating that thousands of schoolchildren in China use a processor architecture that gets almost no global attention! Of course, it’s intended to reduce their reliance on western IP, but it’s still genuinely interesting technically. I wish modern computer architecture classes would at least acknowledge the existence of these alternative ISAs like loongson, elbrus, and others.

A surgeon in Berlin just operated while looking straight through a patient's spine. Not with X-rays. With holograms floating above the body. Helios Berlin-Buch is the first German hospital where surgeons wear AR glasses during spinal surgery. They see organs, bones, and blood vessels in 3D - while operating. What this means: → 40% more accurate implant positioning → Significantly shorter surgery times → Faster recovery for patients Think about that. 1.9 million spine surgeries happen globally each year. 1.9 million people facing potential paralysis. 1.9 million families holding their breath. Now imagine those same surgeries with: • Millimeter precision guided by AI • Surgeons seeing through tissue in real-time • Gesture controls keeping hands sterile The technology that once were for a privileged few? Now spreading to major hospital globally. Here's what changes everything: A spinal implant off by 2mm can mean permanent nerve damage. With AR, surgeons place it perfectly. First time. Every time. By 2025, 20% of surgeons will operate with this superhuman vision. That's 380,000 spine surgeries made safer. 380,000 people with better chances of walking. 380,000 families getting good news. This isn't just better surgery. It's a whole change in healthcare to improve and use the latest technology. The solutions are getting cheaper and more accessible, but still more funding is needed to support doctor's training with AR/VR and the otherwise complicated operations. Follow me, Dr. Martha Boeckenfeld, for more breakthroughs saving lives. ♻️ Share if you believe every surgeon should have superhuman vision. #MedTech #Innovation #FutureOfHealthcare

The first Apple Vision Pro-assisted cataract eye surgery was just completed successfully. An ophthalmologist in San Diego, Dr. Tommy Korn, used Apple Vision Pro during a live cataract procedure as part of a clinical study at Sharp HealthCare. Instead of constantly turning between microscopes, monitors, and different screens during surgery, he used the headset to view critical patient data, surgical imaging, and real-time visual overlays directly in front of him. The setup also included a ZEISS digital surgical microscope and ClearSphere software to create a more immersive operating environment. The goal of the study is simple. See whether spatial computing can improve surgical precision, depth perception, workflow, and surgeon ergonomics while reducing the physical strain doctors have dealt with for decades. As Dr. Korn said, surgeons have been sitting in front of microscopes since 1946, and he doesn’t want to go back to the old way. And honestly… this makes a lot of sense. $3,500 sounds expensive for consumers. For hospitals and surgical centers, that cost is almost nothing if it helps doctors operate more efficiently and potentially improves patient outcomes. We spent years using VR headsets for gaming, entertainment, and watching courtside NBA games. Now we’re starting to see what happens when the same technology enters operating rooms. This is where things get really interesting. Healthcare might be one of the biggest winners of the next wave of consumer tech. Follow Endrit Restelica for more tech stuff.

💡 𝗖𝗵𝗼𝗼𝘀𝗶𝗻𝗴 𝘁𝗵𝗲 𝗿𝗶𝗴𝗵𝘁 𝗺𝗲𝗺𝗼𝗿𝘆 𝗶𝘀 𝗼𝗻𝗲 𝗼𝗳 𝘁𝗵𝗲 𝗺𝗼𝘀𝘁 𝗶𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗮𝗹 𝗱𝗲𝗰𝗶𝘀𝗶𝗼𝗻𝘀 𝗶𝗻 𝗲𝗺𝗯𝗲𝗱𝗱𝗲𝗱 𝘀𝘆𝘀𝘁𝗲𝗺𝘀. Speed, cost, power consumption, endurance, and persistence all depend on which memory you select. 👉 This diagram summarizes the complete memory landscape used in modern embedded systems. Let’s break it down. ⚡ 𝐕𝐨𝐥𝐚𝐭𝐢𝐥𝐞 𝐌𝐞𝐦𝐨𝐫𝐲 (𝐃𝐚𝐭𝐚 𝐥𝐨𝐬𝐭 𝐰𝐡𝐞𝐧 𝐩𝐨𝐰𝐞𝐫 𝐢𝐬 𝐨𝐟𝐟) Used for runtime execution and temporary data. 🧩 𝐂𝐏𝐔 𝐈𝐧𝐭𝐞𝐫𝐧𝐚𝐥 𝐌𝐞𝐦𝐨𝐫𝐲 Integrated directly inside the microcontroller. 𝐑𝐞𝐠𝐢𝐬𝐭𝐞𝐫 𝐌𝐞𝐦𝐨𝐫𝐲  • Fastest memory in the system  • Used directly by CPU instructions  • Stores operands and execution state 𝐂𝐚𝐜𝐡𝐞 𝐌𝐞𝐦𝐨𝐫𝐲  • Stores frequently accessed data  • Reduces external memory latency  • Major performance accelerator 🧠 𝐑𝐀𝐌 — 𝐑𝐮𝐧𝐭𝐢𝐦𝐞 𝐃𝐚𝐭𝐚 𝐒𝐭𝐨𝐫𝐚𝐠𝐞 𝐒𝐑𝐀𝐌 (Static RAM)  • Extremely fast access  • No refresh required  • High silicon cost 👉 Used for CPU cache, buffers, stacks 𝐃𝐑𝐀𝐌 (Dynamic RAM)  • Higher density than SRAM  • Requires periodic refresh 👉 External system memory 𝐒𝐃𝐑𝐀𝐌  • Clock-synchronized DRAM  • Higher bandwidth operation 👉 Main memory for MPU/SoC systems 🔋 𝐍𝐨𝐧-𝐕𝐨𝐥𝐚𝐭𝐢𝐥𝐞 𝐌𝐞𝐦𝐨𝐫𝐲 (𝐃𝐚𝐭𝐚 𝐫𝐞𝐭𝐚𝐢𝐧𝐞𝐝 𝐰𝐢𝐭𝐡𝐨𝐮𝐭 𝐩𝐨𝐰𝐞𝐫) 📘 𝐑𝐎𝐌 𝐅𝐚𝐦𝐢𝐥𝐲 𝐌𝐚𝐬𝐤 𝐑𝐎𝐌  • Programmed during manufacturing  • Not reprogrammable 👉 Mass production devices 𝐏𝐑𝐎𝐌  • Programmable once 👉 One-time configuration 𝐄𝐏𝐑𝐎𝐌  • UV erasable  • Reusable but slow erase process 𝐄𝐄𝐏𝐑𝐎𝐌  • Electrically erasable  • Byte-level write access  • Limited write cycles 👉 Configuration storage ⚙️ 𝐅𝐥𝐚𝐬𝐡 𝐌𝐞𝐦𝐨𝐫𝐲 𝐍𝐎𝐑 𝐅𝐥𝐚𝐬𝐡  • Random read access  • Execute-In-Place (XIP) possible 👉 Firmware storage 𝐍𝐀𝐍𝐃 𝐅𝐥𝐚𝐬𝐡  • Block-based access  • Very high density 👉 Data storage & file systems 𝐞𝐌𝐌𝐂  • NAND Flash + embedded controller  • Wear leveling & management included 👉 Mass embedded storage 💾 𝐍𝐕𝐑𝐀𝐌  • Combines RAM behavior + Flash persistence.  • Byte-level updates  • Fast writes  • Retains data after reset 👉 Ideal for: Retain critical system state across power loss There is no universal best memory. Good embedded architecture balances: Speed, Power consumption, Cost, Endurance, Persistence 💬 Which memory type do you use most in your projects — SRAM, SDRAM, or Flash?

Big news from the tech-world: Google unveils #AndroidXR, a platform for #VR and #AR, along with new #AI-powered smart glasses featuring the #Gemini voice assistant. 💡 I’ve spoken before about how the intersection of AI and XR is not just a futuristic concept but a transformative force that’s already in motion. So this announcement couldn’t have come at a better time - as I’m finalizing my 2025 #predictions, where, spoiler alert, the synergies between AI and VR/AR will play a pivotal role in reshaping industries. With Android XR, Google is positioning itself as a critical player in making this a reality. The potential #usecases are limitless: • Next-gen employee #training through immersive simulations that enhance learning retention and engagement 📚 • Breakthrough R&D applications, using VR environments to prototype and test in real-time without physical constraints 🛠️ • Streamlined #onboarding processes, offering new employees immersive walkthroughs and accelerated knowledge acquisition 🚀 At VRdirect, we’re already realizing many of these use cases, helping businesses unlock the potential of XR for real-world impact. I’ve attached our latest Siemens success story in the comments section, so take a look! This announcement underscores a trend I’ve been closely following: AI’s role as a catalyst for accelerating the adoption and utility of VR/AR. The big question now is whether Google can leverage its ecosystem to lead the XR revolution and compete with the likes of Apple and Meta. 🌐 Let’s discuss: How do you see AI and VR transforming industries in the coming years? 👇 #ExtendedReality #AugmentedReality #VirtualReality #Innovation #TechTrends #DigitalTransformation

I have always been fascinated by the way the brain manages complexity with such elegance. Its ability to balance speed, energy efficiency, and adaptability still surpasses what current machines can deliver—but neuromorphic computing is narrowing that gap. Inspired by neural dynamics, this emerging class of hardware processes information in parallel, consumes exceptionally low energy, and adapts in real time. The shift goes beyond performance gains. It suggests a new paradigm for computation—moving away from rigid sequential logic toward fluid, brain-like interaction. These systems are still in development, but the potential is considerable. From edge devices that make autonomous decisions to AI models that evolve on the fly without retraining, their applications could extend across fields like healthcare, robotics, and beyond. We often track progress by looking at teraflops or model sizes. But perhaps now is the moment to also reflect on how biologically inspired architectures may guide us toward more intelligent and sustainable digital systems. #NeuromorphicComputing #AI #EdgeAI #FutureOfComputing

The real AI moat isn’t the model. It’s the hardware. Models are rapidly commoditizing. AI wrappers are everywhere. What’s becoming scarce isn’t intelligence, it’s access to unique, private, real-world data. And that data lives in devices. Whoever owns the hardware owns the interface and has acess to the data. Whoever owns the sensor layer has access to contextual, high-frequency behavioral data that can’t be scraped from the internet. That data will be used to train the next generation of AI. This is why smart glasses, wearables, pendants, airpods with cameras and on-device AI matter so much. Chat interfaces don’t capture environment, motion, gaze, or spatial AI awareness. Hardware does. If the path forward is AI that truly works in the physical world, we don’t just need better LLMs. We need new models and new architectures like large geospatial models (Niantic Spatial, Inc.) and world models (World Labs) that understand space, context, and cause and effect. The future isn’t Ready Player One. It’s more Jarvis from the Marvel movies. (Remember Tony Stark's EDITH glasses?) The future of computing is spatial, ambient, world-aware intelligence layered onto daily life. I spent half my tech career in XR and spatial hardware and everyone in hardware knows the phrase: "Hardware is hard", because it is, but... That’s exactly why companies like Apple, Meta, Google, Snap, OpenAI, etc's are doubling down on tactile moats like hardware. Do you see what I see? (Pun intended) RiseUp Summit Startup Scene ME #AI #SpatialComputing #Smartglasses #WorldModels #SpatialAI #GeoSpatialAI #Meta #Apple #Snap #GoogleXR #OpenAI #PhysicalAI