Julie Willey

5 Reasons Why Legacy Nodes Remain Crucial in the 1nm Era? Legacy nodes Typically refer to older, well-established fabrication processes, often at 45 nm or 28nm and larger. 🚀 Cost-Effectiveness: Manufacturing chips at advanced nodes like 1nm is expensive due to the complexities involved. Legacy nodes are a better choice for cost-sensitive applications. Imagine building a high-performance gaming PC vs. a basic office computer. The gaming PC (1nm chip) is expensive with all the latest tech. But a basic computer (legacy node) is much cheaper to make, perfect for everyday tasks like browsing the web. 🚀 Power Efficiency: Smaller transistors in advanced nodes enable faster performance but often come at the cost of higher power consumption. Legacy nodes, with their larger transistors, can be more power-efficient, a critical factor for battery-powered devices like smartphones and wearables. Tiny transistors in new chips are like powerful engines in cars - they work great but use a lot of fuel (battery power). Bigger transistors in legacy nodes are like fuel-efficient car engines - they might not be the fastest, but they use less battery, ideal for things like smartphones and smartwatches that need to last all day. 🚀 Maturity and Reliability: Legacy nodes have been around for longer and have undergone extensive refinement, leading to highly optimized and reliable processes. This maturity ensures consistent performance and fewer bugs, crucial for applications demanding stability. Think of a recipe you've made many times. You know exactly how much of each ingredient to use and how long to cook it for. Legacy nodes are like that tried-and-tested recipe - they're reliable and we know how to make them work perfectly for everyday electronics. 🚀 Compatibility: Not all applications require the bleeding-edge performance offered by 1nm chips. Legacy nodes can integrate well with newer technologies, enabling efficient design of complex systems with a mix of performance requirements. Imagine building a house - you might need different materials for the roof, walls, and foundation. Legacy nodes can be like those different materials - they work well with newer technologies to create all sorts of electronic devices, like a phone with a powerful processor (1nm chip) but a low-power Bluetooth chip (legacy node) for connecting to headphones. 🚀 Internet of Things (IoT): The vast and growing world of IoT devices, many of them simple sensors, often doesn't need the processing power of 1nm chips. Legacy nodes are perfectly suited for these low-power, low-cost applications. Tiny sensors in things like fitness trackers or smart thermostats don't need the processing power of a fancy 1nm chip. Legacy nodes are the perfect tool for these smaller jobs, keeping costs low and battery life long. Guess, Which Country is investing huge in legacy Nodes? Tell us in the comments. A detailed post is in comments. For all semiconductor and AI content, follow TechoVedas

77

3 Comments

I have finally delved into this year's "100 Silicon Startups Worth Watching in 2024," recently published by EE Times | Electronic Engineering Times. One insight from Peter Clarke in his preface is the interesting shift in the geographical distribution of these top 100 companies. While there's a slight decline in California and North American representation (is it real?), the most striking changes are the sharp decrease in Chinese companies and the meaningful rise in European entries. Peter attributes the Chinese decline to trade tensions: "#China continues to have high levels of startup activity, but its reduced presence on the Silicon 100 reflects the trade tensions between the U.S. and China. That friction has intensified in recent years and has hit a number of Chinese startups that may have had international aspirations but have been denied access to leading-edge silicon out of foundry TSMC." #Europe, on the other hand, has seen an increase from 23 to 29 startups, with the U.K. leading the charge: "The U.K. has 12 startups on the current list, up from nine startups in version 23. The U.K., with longstanding entrepreneurial bases in Cambridge, Oxford, Bristol, and Edinburgh, has enjoyed strong representation for many years." In my view, this growth is a testament to #Europe's robust academic system, coupled with government support, investments, and incubators like Silicon Catalyst.UK. Israel continues to excel: "Israel has always punched way above its weight by population and continues to do so," with its representation growing slightly from 7 to 8 this year, including NeuReality*, POLYN Technology, NeoLogic*, Speedata.io, Pliops, Chain Reaction Ltd.*, Hailo, .Quantum Machines . (*=new additions to the 2024 list). This list provides a good overview of this dynamic and evolving industry, blending both emerging and more established technologies. It's a must-read for anyone interested in the future of #semiconductor #innovation. For the full list: https://lnkd.in/dgfrNUWX

50

6 Comments

Keeping your data close to the GPUs is essential, so we offer VAST Data as part of our #GPUasaService AI Platform. Let us help you #StepontheGaaS and move your #AI strategy forward. #Fast #Secure #Private Dominic Ken Michael Bill Traci

10

1 Comment

PSA: TUNE YOUR #VLLM SETTINGS. Kind of crazy how different the results can be, especially on the AMD MI300X which is showing top numbers. Mako auto-tuner does this for you btw! Also Nvidia friends, why is off-the-shelf vLLM so slow on H100 + #llama 8B model. I feel like this can't be right. 500 prompts, benchmark_serving.py with shareGPT (250 tok in / 200 tok out)

34

4 Comments

In my two decades of recruiting, I've rarely seen the investment in hardware startups like that surrounding AI. Its rapid rise has brought many newcomers vying to make compute more efficient. But, this makes it tough for customers to separate the "signal from the noise". The harsh reality is that some offer little more than concepts—essentially, vaporware. This undermines years of work by patent-holding researchers but it’s the operating reality for startups like Ayar Labs, whose collective time invested in R&D can be measured in centuries rather than decades. What started in 2015 as the brainchild of Professor Milos Popovic, Alex Wright-Gladstein, and five eminent researchers soon evolved into a company that showed immense promise for integrating photonics and processors to accelerate computing. They took a deliberate path, ignoring hungry VCs and choosing strategic investors like GlobalFoundries, Intel, and Applied Materials, allowing them to focus on advancing their technology and mission. Under their first commercial CEO Charlie Wuischpard, he led them to a key design win in 2022 that fueled investor confidence and resulted in an additional $130M in funding. But if you ask any photonics engineer or architect, they’ll tell you using this technology to accelerate computing is BRUTAL. Combining photons and electrons requires a Herculean level of precision and deep insight into physics and engineering. Recognizing this, Ayar’s Board looked within and tapped co-founder and Chief Scientist Mark Wade to be their new CEO in 2023. This was a stroke of brilliance. Not only has Mark helped shepherd the company through their most painstaking technical developments, but he’s also been an evangelist for optical interconnectivity, attending what seems like every top conference to share its potential. Under Mark, Ayar has progressed to have not one, but two revenue-generating products and they’re shipping tens of thousands. BUT (and excuse me for framing it that way) in our industry, you’ve got to be in the “hundred million club” to make it; frankly, the clock is ticking and competitors are circling. Ayar turns ten next year and the game has changed. That AI gold rush I mentioned has made it harder to get in front of customers with a salient value proposition, and the same goes for attracting talented engineers and leaders. Those we recruit are leery of startups making big claims. My advice to Mark and the rest of the team is to go BIG or go HOME. Secure the next round of funding and expand the team swiftly and holistically. Do an entire team lift-out! You’ll save time and accelerate faster through a group with proven synergy rather than hiring piecemeal. I’ve said it before: 2025 is primed to be the year photonics revolutionizes the data center so make your moves now because you don’t want to be left on the sidelines when all the key players have already been picked. #semiconductorindustry #photonics #artificialintelligence

71

3 Comments

Nvidia and Intel go way back. They had kindof a truce where they stuck to their own markets. But PCIe decided otherwise. Xeon was using PCIe gen3 while AMD Epyc was already on PCIe gen4. Nvidia was used to integrate Xeons all the time. In return, Intel never brought out a discrete GPU to directly compete with Nvidia as AMD was doing. Then the PCIe thing happened, and if anything is crucial it is the bandwidth of the PCIe link between CPU and GPU. Similar to the stage where Intel wasn't interested in low power cpus (phones and tablets where ARM was allowed to grow into a giant), this truce hampered Intel. So they only started to develop a discrete GPU after the relationship took a hit. Of course they had the on chip gpu but that is just minimal in comparison with the beasts we got from AMD and Nvidia. In hindsight, everything is quite obvious, the truce worked for a long time. And not going low power probably made sense back then for whatever real reason. I'm just observing to understand more about the complex nature of semi giants and their strategies. For make benefit great person myself. https://lnkd.in/gqBMYDWi

43

10 Comments

POLYN’s analog chips are thinking outside the box—literally. With 100X less power and 1000X lower data latency, they’re making us question who’s really in charge here. Invest now. https://lnkd.in/dyvbMAK3

13

Sorry for the delay in last week's AI news - just got back from Taiwan and all of its Jensenity! 🔙 𝐓𝐡𝐞 𝐁𝐚𝐜𝐤𝐰𝐚𝐫𝐝 𝐏𝐚𝐬𝐬: 𝐑𝐞𝐯𝐞𝐫𝐬𝐢𝐧𝐠 𝐓𝐡𝐫𝐨𝐮𝐠𝐡 𝐭𝐡𝐞 𝐀𝐈 𝐖𝐞𝐞𝐤 Week 42 | June 3-9 JENSENITY at Computex Industry: 🚨AMD unveiled its latest AI processors at the Computex trade show. It also detailed plans to compete with Nvidia (https://lnkd.in/ggvTrCre) 🚨 NVIDIA is said to be prepping AI PC chip with next-gen Arm cores and its Blackwell GPU architecture (https://lnkd.in/gS6Qv2rm?) Financing and M&A: 🚨 Text-to-video platform Pika raises $80M led by Spark Capital with participation from Greycroft, Lightspeed, and others (https://pika.art/blog) 🚨 Twelve Labs, a leader in the multi-modal foundational model space, raised $50m led by New Enterprise Associates (NEA) and nVentures. The platform is being used by companies in the media & entertainment, advertising and automotive sectors (https://lnkd.in/gvwRkxAr) 🚨 Tektonic AI raises $10m led by Point72 and Madrona to build AI agents to help automate business operations. CEO Nic Surpatanu previously held leadership roles at Tanium, UiPath and Microsoft (https://lnkd.in/gGfxiNW6?). 🚨 Hoop, a next-generation task management platform, raises $5M seed round from Index Ventures 🚨 Greptile raises $4m seed round for code understanding (https://lnkd.in/gDF8MD-C?) 🚨 Cartwheel, a generative 3d animation tool raises $5.6m seed led by Accel and KV Research & Development: 🚨 Qwen2 from Alibaba launched. According to internal released benchmarks, the language model demonstrates superior performance over Llama3-70B and Mixtral 8x22B (https://lnkd.in/gEJi2shn?) 🚨LlamaCare: A Large Medical Language Model for Enhancing Healthcare Knowledge Sharing (https://lnkd.in/g44Q8Pyv?) Other Interesting Resources: 🚨 A good article on robotics in the new wave of LLMs and foundational models. Highly recommend a quick read through for anyone interested in robotics (https://lnkd.in/gn5gbrTR) 🚨Interesting article from John Luttig from Founders Fund on open-source vs. closed-source 🚨Interesting article from Anthropic on Claude's character based on Anthropic's constitutional AI (https://lnkd.in/gK43G7kF?) 🚨Jensen keynote address at Computex: https://lnkd.in/g_eBy-he

76

In the fiercely competitive semiconductor industry, companies are in a constant battle to retain top talent because in the heart of every good engineer is a passionate problem solver, looking for their next challenge. For them, innovation is the currency of success. Yet companies like Qualcomm, Intel, and Broadcom attract tens of thousands every year by dangling lucrative compensation packages to lure them from competitors. Startup leaders don’t have those resources, so they have to work harder to attract the best talent and then fight like hell to retain them. But some of the most successful startup leaders I know take a different approach. Rather than viewing attraction and retention as a fight, they leverage the inherent qualities of their companies as their advantage. They use their nimbleness, bias toward innovation, and their collaborative cultures like superpowers to attract the right people and those usually end up being the reasons why people stay. Take Richard Fung, CEO of The Six Semiconductor, for example. I had a conversation with Richard and asked him to share his approach to motivating his team. His response was like a mini-masterclass on the topic of talent attraction and retention. Richard emphasized that a culture that encourages engineers to dive deep into the intricacies of their craft, understand the systems they’re building, and connect with the end customer is a successful culture. Richard has been using this approach and it's been paying dividends for The Six Semiconductor so in his honor, here are his six key concepts for creating this type of organization: 1. Foster a Learning Culture: Support ongoing learning and skill development beyond employees' core responsibilities. 2. Prioritize Customer Focus: Embed a customer-centric mindset, and encourage employees to enhance user experiences through their work. 3. Champion Problem-Solving: Empower employees to creatively solve challenges with a focus on customer needs. 4. Promote System-Level Understanding: Encourage employees to grasp the bigger picture of the company's offerings and their role within it. 5. Break Down Silos: Promote collaboration to ensure a comprehensive understanding of projects across teams. 6. Offer Meaningful Challenges: Provide intellectually stimulating projects that make a tangible impact on the company's success. What Richard highlighted is that success goes much deeper than getting great people to join. It's about unleashing the full potential of every individual on your team to drive innovation and help shape the company, and ensuring your team heads in the right direction together. I’m thrilled I got to learn Richard’s perspective and believe other leaders can benefit from hearing it, so I shared the video below. My "challenge question" to other startup leaders is this: what are you doing to foster this type of culture in your organization? Either share below or, feel free to DM me. #semiconductorindustry #startup #leadership

67

2 Comments

The autonomous vehicle revolution is here, with the world anticipating Elon Musk's October 10th Tesla robo-taxi announcement and Waymo's recent milestone of 100,000 weekly rides. Lumotive's Liquid Crystal Metasurface (LCM) technology is ready to lead the way—delivering scalable 3D sensing solutions to transform mobility from thousands to millions of autonomous vehicles. How LCM Transforms Autonomous Driving: Superior Reliability: Our solid-state design eliminates mechanical components, increasing reliability and system longevity—essential for autonomous vehicles. Adaptive Flexibility: Software-defined beam steering adjusts scanning patterns and resolution on the fly, optimizing performance for diverse conditions. Scalable for Adoption: Built with semiconductor manufacturing, our LCM technology is scalable and cost-effective, meeting the industry's needs for large-scale deployment. As the demand for advanced 3D sensing grows, Lumotive is ready to shape the future of mobility. Join us as we drive forward into this exciting era. #AutonomousDriving #Lumotive #robotaxi

66

2 Comments

I worry that Abingdon-based ‘fabless’ ASIC (Application Specific Integrated Circuit) designer EnSilica could become a victim of its own success. Today’s FY results came with a guarded warning that EnSilica could run out of cash within the next 12 months without additional financing. Cofounder and CEO Ian Lankshear attributed the problem in part to customer order and payment slippage, and in part to ‘complications’ in the global semiconductor supply chain exacerbated by the deteriorating relationship between the US and China. The latter was already evident and beyond EnSilica’s control. The former leans towards overoptimistic forecasting (see my recent post on Substack at https://lnkd.in/echyDK-y.) In fact, EnSilica now has a new CFO. In truth I think the problems are of a more fundamental nature. First, EnSilica is both a semiconductor design and a supply business. The design is done in-house, the manufacturing by partner foundries. While this business model has the advantage of long-term, recurring revenues (8-10 years is not unusual), it is also hostage to changing demand in customer delivery schedules, as has already happened. Second, as I have said many times before, management chose to IPO far too early, in my opinion. EnSilica listed on AIM in May 2022 when its revenues were under £10m and was losing money. Since then, management has had to go back to investors and lenders multiple times – all, of course, in the unforgiving spotlight of a public market. EnSilica’s shares currently trade around 4p below the 50p IPO price. EnSilica is growing fast on the back of an impressive order book. Revenues last FY (to 31st May 2024) exceeded £25m, though the company is now back to a (small) net loss. However, this rapid growth – along with the aforementioned challenges – means that the business spends more cash on capex than its operations generate. And this gap risks widening the faster EnSilica grows. I would not want EnSilica to suffer the same fate as its ‘neighbour’, semiconductor design startup, Sondrel, which listed on AIM a few months after EnSilica. Never having made a profit, Sondrel quickly ran out of cash, its shares crashed and Sondrel delisted from AIM in August at a fraction of its IPO valuation (see https://lnkd.in/eiaWY5iH). For now, EnSilica is in a much better position than Sondrel ever was. But Lankshear and new CFO Kristoff Rademan need to take a long, hard look at the best capital structure that will help the business achieve its proven potential in the long term. EnSilica should be cherished as a jewel in the UK semiconductor industry crown. Personally, I think it will be in the best interests of its customers, employees and backers for EnSilica to go private on its own terms while it still can. 

6

https://lnkd.in/gKdPEBuK Good Luck to Qualcomm in Getting Intel Inside. The semiconductor giant’s tilt at Intel has opportunistic attractions. But getting a deal completed will be hard, as will making a success of any takeover. #intel #qualcomm #ai #technology #innovation #business #Semiconductorindustry #electronics #semiconductor #supplychain #computerchips #semiconductors #chips #chipmaker #foundry credit: Bloomberg

8

Here's what our VP of Products Bob Beachler highlights in the EE Times article about the Arm & Untether AI collaboration... “What we’re seeing is that a lot of our automotive customers are trying to figure out ways to differentiate their hardware and make it better than an off the shelf system from Mobileye, Qualcomm or Nvidia,” “With the growing amount of AI in automotive infrastructure, they are looking at how to solve this problem with scalable solutions that go from level 2+ all the way to full autonomy.” This type of system would require a scalable CPU architecture with the option for multiple AI accelerators, which will be provided by the new Arm-Untether partnership. #autonomous #driving #powerefficient #atmemory #ai #compute #acceleration

29

2 Comments

Jim Keller, former Intel, AMD, and Tesla engineer, aims to challenge Nvidia's dominance in AI with Tenstorrent's energy-efficient chip design. Tenstorrent's upcoming second-generation AI chip promises superior energy and cost efficiency, presenting a viable alternative to Nvidia's offerings. The innovative design of Tenstorrent's chip features independent decision-making cores, potentially reshaping AI chip technology across diverse applications. #AI #Nvidia #semiconductor #semiconductors #chip #technology #innovation #research #computing Tenstorrent NVIDIA https://lnkd.in/g6a3vAu4

31

Some tough love here for Intel by Steve Rosenbush -- just getting to it. Steve makes some good points, and I agree with most of it. The foundry business should be a great business to be in for as far out as we can see, it needs competition, and the critical elements should be manufactured in the U.S. and other secure locations, but that doesn't mean Intel has the financial, operational, and cultural mix needed to compete in both. It became brutally clear over the past few months that Intel bit off a lot more than it could chew. As I've said repeatedly, national security and supply chain security issues aside, Intel itself has a competitiveness problem, and like or not that comes down to innovation -- people, culture and leadership. Steve's closing statement is well deserved -- Intel should get its own house in order before attempting to save anyone else, including U.S. competitiveness. I haven't personally done an operational audit on Intel, which is what's needed to make recommendations on structural and leadership changes, but the financial picture is sufficiently challenging that it would almost certainly include large spin-offs (if not a sale). The foundry business is where most of the hemorrhaging is taking place. If you want to save the patient, that's typically a good place to start... "Rebuilding domestic semiconductor manufacturing capacity in the U.S. is critical. In an ideal world, Intel would play a major role in that project. But the reality is that the greater good may be best served by allowing Intel to get its own house in order before it takes on the broader interests of U.S. security."

7

2 Comments

Before I worked in the #semiconductorindustry, I worked as a civil engineer. I found this article about CalTrans using #AI to answer complex traffic-related questions fascinating. "Caltrans says the data generated from traffic sensors and cameras — coupled with the continuous velocity of data generation and the variety of videos, images, log files, third-party data streams and guidance — poses challenges for humans to digest and utilize." Sounds a bit like using #AI and #ML for smarter #scheduling in a #waferfab, doesn't it? More #industrialengineering than civil engineering.

17

The new Snapdragon Dev Kit is a machine designed to aid developers in coding and recompiling their Windows applications natively for the new Arm-based Snapdragon CPUs. — PC Gamer gives a clear breakdown of how the #Snapdragon Dev Kit for Microsoft Windows is accelerating development for #CopilotPCs.

360

3 Comments

New SiFive Performance P870-D Complements Intelligence Chipset for AI TL;DR - SiFive announced the SiFive Performance P870-D, a high-performance RISC-V datacenter processor designed for demanding AI workloads, offering up to 256 cores, enhanced scalability, and advanced Reliability, Availability, and Serviceability (RAS) features. This processor positions SiFive as a competitive player in the data center market, providing data center architects with a flexible, power-efficient solution for AI-driven applications. SiFive aims to leverage RISC-V's open-source flexibility to challenge established industry giants like Intel and AMD. Read the Research Note in the comments below.

1

1 Comment

We use a combination of Triton and native vendor libs (cuBLAS, hipBLAS, etc) to max out performance. I love triton because we can reaaally quickly put together a unique fused kernel, and love the BLAS libs because the triton perf just doesnt match up for stuff like GEMMs, which is to be expected :) Shoutout to Sanjeev Rampal for a neat overview. https://lnkd.in/gVR4X_aT

50

1 Comment