Lin Song

The RISC-V vs ARM debate has become nothing more than tech tribalism that makes engineers feel virtuous while completely overlooking what actually determines project success in the real world. We've all encountered, 'RISC-V will democratize computing with its open-source model!' But let's be brutally honest: in 80% of commercial deployments, ecosystem availability beats architectural ideals every time. The countless hours spent battling incompatible RISC-V toolchains could have been invested in optimizing ARM-based solutions. And don't get me started on teams that select RISC-V for 'openness' only to find their custom ISA extensions have locked them into a single vendor more tightly than ARM's licensing ever could. I've observed directly projects hemorrhage 6 months just to get basic RISC-V compiler support working reliably for standard peripherals, while comparable ARM implementations reached production in 8 weeks leveraging battle-tested CMSIS drivers. The most ironic part? The 'open-source' narrative conveniently obscures that 70% of production RISC-V systems rely on proprietary extensions and commercial toolchains - you're essentially swapping ARM's licensing costs for RISC-V vendor support agreements. What we genuinely need is ruthless pragmatism: ARM when time-to-market and peripheral support are critical, RISC-V when custom accelerators or extreme cost constraints drive the design. This isn't compromise - it's engineering common sense. When your product launch is next quarter and you need functional WiFi, BLE, and security subsystems, every absent driver in the RISC-V ecosystem directly threatens your schedule. The next time you feel drawn to RISC-V for ideological reasons, ask yourself: Will this choice genuinely accelerate my development timeline, or am I volunteering to be a toolchain guinea pig? ⚖️ What single factor tipped the scales between RISC-V and ARM in your last project? Share your make-or-break decision criteria below - bonus points for how you'd evaluate the trade-off differently today! #RISCV #ARM #EmbeddedSystems #ProcessorArchitecture #EngineeringReality #EcosystemMaturity #TechChoices #DeveloperExperience #FPGA #SoC #CostOptimization #TimeToMarket

166

17 Comments

Proud to Partner with the University of Illinois Urbana-Champaign on Heterogeneous Computing! At AMD we are thrilled to collaborate with the University of Illinois Urbana-Champaign on the Heterogeneous Accelerated Compute Cluster (HACC) program. This innovative program, integrates CPUs, GPUs, and FPGAs, paving the way for groundbreaking research in heterogeneous computing. Why is this collaboration crucial? Addressing Complex Technical Challenges: HACC enables exploration of system interactions, tackling real-world problems with advanced computing solutions. Strengthening Academic-Industry Ties: Our partnership ensures that researchers have the resources and feedback needed to drive impactful research. Championing Open Source: The success of this collaboration is the open-source projects that receives contributions from more than 60 active members and highlights the importance of accessible technology. At AMD, we are dedicated to making AI solutions more accessible and user-friendly, empowering everyone to leverage our advanced computing resources to innovate and make a difference. Join us in celebrating these advancements and the collaborative efforts that make them possible! 🔗 Watch Now https://lnkd.in/g2V9XBV3 #HeterogeneousComputing #Innovation #Research #Education #AMD #OpenEcosystem #TechCollaboration #AI

101

3 Comments

Bloomberg has just published an article about "Manus," an AI application that claims to be the first fully autonomous AI agent in the world and has received tremendous attention recently. The piece includes my comment: “Manus is actually a half-finished product,” said Yiran Chen, an electrical and computer engineering professor at Duke University who has tested the service. The startup likely hopes “that by being first, they can attract investors, despite the product not being fully developed yet.” To clarify, my comment was not a criticism of the startup or its product. Rather, I meant that small companies often have the advantage of bringing new technologies to market before they are fully mature. Their lower trial-and-error costs and faster iteration cycles allow them to take risks that larger companies might avoid. You can read the full article here: https://lnkd.in/eSmxSC84

101

3 Comments

Three observations - from Sam Altman’s blog (https://blog.samaltman.com) (1)The intelligence of an AI model roughly equals the log of the resources used to train and run it. [sub linear scaling with compute resources; need better models now to get to AGI?] 2. The cost to use a given level of AI falls about 10x every 12 months, and lower prices lead to much more use. [unbelievable-> Ai agents everywhere?] 3. The socioeconomic value of linearly increasing intelligence is super-exponential in nature [we need to figure out a linear scaling of AI with compute resources or even better. Increasing investments exponentially is a way but then only few can do it?]

57

The biggest challenge / opportunity in neuromorphic computing is figuring out what the right algorithms should be. Hardware today is scalable and programmable (see Dhireesha Kudithipudi's latest review in this week's Nature). And we and others have shown that this hardware can show real advantages on real problems, but those algorithms often seem ad hoc and not very brain-like. In our most recent pre-print, we (h/t Brad Theilman) describe a neuromorphic algorithm that can solve finite element problems. Not only does this demonstrate that energy-efficient neuromorphic hardware can provide numerically accurate solutions comparable to classic sparse linear solvers, but it does so in a brain-like fashion. This algorithm, which we term NeuroFEM, is a cortex-inspired circuit that iterates using sparse dynamics. The approach is broadly applicable to a wide range of problems that can be described as a solving a linear system (i.e., A x = b). Most importantly, even though the algorithm itself is very brain-like and the neuromorphic implementation is non-trivial, from an application perspective it is transparent: given a matrix A and vector b, it solves for x. In short, application developers won't need a PhD in neuroscience to get the most out of neuromorphic hardware. In our pre-print, we demonstrate this on several examples and show how this scales on Intel's Loihi 2 system (hint: it scales very well!). We believe the next steps may be even more exciting - this both opens up a new outlet for improving numerical methods and understanding how the brain itself processes information. Stay tuned! https://lnkd.in/gPZJu6ph

234

21 Comments

Bridging the Gap: How PITC Accelerates the Industrial Uptake of Photonic Chips. What does it really take to move photonic technologies from university labs into scalable, manufacturable industrial products? At OFC Conference 2025, my Optica colleague Jon Pugh chatted with Akanksha Kapoor and Jan Tatousek to explore how the Netherlands PITC - Photonic Integration Technology Center (PITC) is turning research into real-world solutions. Their conversation highlights PITC’s role as a collaborative hub connecting academia, applied research, and industry. PITC isn’t just another R&D centre. It’s a joint innovation platform tackling early-stage photonic IP challenges that are too risky or capital-intensive for individual companies to handle alone. By pooling expertise from four powerhouse institutions, PITC lowers barriers to adoption and accelerates photonic chip development. 💡 What makes PITC unique? 1. Eindhoven University of Technology (TU/e): Pioneers in Indium Phosphide (InP) PICs for scalable, high-volume production. PITC benefits from TU/e’s Nanolab and deep vertical expertise in laser and amplifier integration. 2. University of Twente (UT): Home to one of the world’s largest integrated photonics communities. UT contributes cutting-edge capabilities in Silicon Nitride (SiN) and nanophotonics via MESA+ and the upcoming New Origin foundry. 3. PhotonDelta: A government-backed ecosystem builder linking the entire photonics supply chain—from startups to foundries to markets. A catalyst for commercial scaling. 4. TNO: Applies its multidisciplinary strength in optomechatronics, packaging, and printed electronics to develop industry-ready, high-reliability PIC modules—especially for aerospace, sensing, and medtech. 🔬 Three IP powerhouses under one roof: ·      InP platform: TU/e + SMART Photonics ·      SiN platform: UT + upcoming New Origin foundry ·      Silicon Photonics: imec in Leuven Why it matters: For startups and corporates alike, the cost of PIC development at low TRLs is a major bottleneck. PITC de-risks that process with infrastructure, expertise, and partnerships tailored to product-market fit. Feel free to engage with Akanksha Kapoor and Jan Tatousek to understand how PITC makes integrated photonics accessible to the industry. Please tell them that Optica sent you. #IntegratedPhotonics #PICs #DeepTech #Photonics #TNO #TUe #UniversityOfTwente #PhotonDelta #PITC #OpticsIndustry #

73

2 Comments

The Intel–NVIDIA partnership marks a pivotal moment for the AI PC. By pairing Intel CPUs with NVIDIA GPUs, we now have the performance headroom to make local LLMs practical at scale. This creates a natural opportunity for the Intel AI Assistant Builder to serve as THE unifying platform—bridging OpenVINO, llama.cpp, and CUDA into a cohesive developer and enterprise solution.

21

1 Comment