Jon Reilly

For #CostModelMonday, here's a chart that started out focused on #HybridBonding, but really just came to be about lots of #chiplets versus a few. I ran a design that has the same total area of silicon coming in, but it's four 4x4mm chips versus sixteen 2x2mm chiplets. The bonding method is hybrid bonding, and the assumption is that placing lots of chiplets takes longer (a truth more than an assumption...), but there may be a yield benefit to handling and placing the smaller die. I found a crossover point by adjusting yields; in the chart below, the total cost is the same, but the two designs have different costs in different categories.

20

2 Comments

I recently sat down with Mike Vizard from TechStrong.ai to discuss how Phison's innovative aiDAPTIV+ technology is making generative AI accessible and affordable for small to medium-sized enterprises. We explored the barriers to entry and how our latest aiDAPTIV+ Pro Suite is simplifying the adoption of Gen AI. You can listen to our full conversation here: https://lnkd.in/ekBdzEk4 #Phison #AI #ML #LLM

27

Every single person on your team — no matter what their position is — should know and live by your company’s values. When I first joined Coherent Corp. (known as II-VI at the time) as a board member, we had 7 values. They were printed on the back of everyone’s business card but, still, I could never remember them all. In the wake of a large acquisition, I was walking around various facilities and chatting with everyone. I asked if anyone knew what our company’s values were — and no one knew them! That was a hint of an opportunity. I realized that the acquisition was a period of transformation for the company, and it was an opportunity to revisit our values. I remember studying the values of all the S&P 500 companies for inspiration, taking into consideration the founding values of II-VI as well as the values of the companies we had acquired. Eventually, I decided on: ICARE which stands for integrity, collaboration, accountability, respect and enthusiasm. Taken in that order, they have allowed our team to declare “I care” as the primary reason for working at Coherent and helping change the world. I firmly believe that values (and sticking to them) are integral to a company’s success. This might be a little bold to say but as far as I know, today, there’s no one in the company that doesn’t know what our values are.

354

24 Comments

In my new role at Sondrel, I wanted to offer my thoughts and views that we are seeing in the current semiconductor industry. I hope you find this article informative.

24

4 Comments

Cadence introduces the Janus NoC for SoC designs.

28

Debugging a software stack for complex System-on-Chip (SoC) designs is a highly iterative process, often requiring the execution of trillions of verification cycles even before end-user applications are considered. For instance, booting Android can take anywhere from 20 seconds to one minute at 1 GHz, depending on the version and end-application, which equates to at least 20 billion cycles. Rebooting this mobile OS ten times a day consumes at least 200 billion cycles. When end-user workloads are factored in, the challenge intensifies, adding more trillions of cycles. Read more in Lauro Rizzatti’s article here: https://lnkd.in/gmK6qwj7

31

1 Comment

Learn how the Cadence DDR controller IP is a key element in a new AI vision processor. https://ow.ly/UzsQ50RpGEf

30

That YUG meeting was awesome, a bunch of short talks covering a host of cool yosys FPGA use cases! Mine: 5 minute overview of how to learn and skip the wait for your digital ASIC design, dev and try everything with a drop in replacement FPGA carrier board.

10

Shedding light on 12 photonics and edge companies you need to know about If you're running large data centers or connecting dispersed locations with high-bandwidth connections, photonics are going to be increasingly important.   https://sdx.io/ECEB

2

🔍 AMD’s Strategic Acquisition of ZT Systems: A Game-Changer for AI and Data Centers AMD’s recent $4.9 billion acquisition of ZT Systems isn’t just about expanding our chip business—it's about transforming how we deliver AI and data center solutions. Here's why this move matters: 1️⃣ World-Class Expertise: By acquiring ZT Systems, AMD gains 1,000 top-tier design engineers with deep expertise in motherboard, power, thermal, networking, and rack design. This strengthens our ability to create comprehensive, high-performance AI systems that go beyond just chips. 2️⃣ Staying Focused: While we’re acquiring ZT Systems, we’re divesting their manufacturing business. This ensures we stay aligned with our core strengths and avoid competing with our valued customers like Dell Technologies and Supermicro. 3️⃣ Enhanced AI Capabilities: This acquisition boosts AMD’s ability to serve a wider range of AI use cases, helping us close the gap with industry leaders like Nvidia. Our AMD Instinct MI300 system is already driving strong growth in data centers, and with ZT Systems, we’re poised to do even more. 4️⃣ Sustaining Strong Margins: By focusing on design and leaving manufacturing behind, AMD can maintain our strong profit margins, even as we bring on new talent and expand our AI offerings. 💼 The Bottom Line: This acquisition is a strategic move that positions AMD as a stronger competitor in the AI space, enhancing our ability to deliver full-system solutions that meet the evolving needs of data centers. We’re not just thinking outside the box—we’re redesigning it. ▶️ Want to dive deeper? Read the full The Wall Street Journal article on the acquisition, "AMD Deal Shows AI Chip Business Isn’t Just About Chips" here: https://lnkd.in/gFjtSpAv #AMD #ZTSystems #AI #DataCenters #Innovation #Leadership

347

1 Comment

Great OFC coverage by LightCounting. TeraSignal’s Intelligent Re-Driver in CMOS is changing the power, reliability and performance calculus between DSP and LPO modules.

27

All in on #Transformer, it is beating all other architectures in end-to-end applications. Like #Etched, #Skymizer is betting on Transformer only accelerator with compiler-centric approach. Skymizer’s first IP product, EdgeThought, is a transformer based on-device LLM inferencing only accelerator.

11

Last week’s DAC showcased exciting announcements from startups that are revolutionizing semiconductor design however these rising stars overshadowed a major update from an industry stalwart - Alphawave Semi - whose news is poised to accelerate the proliferation of chiplets and give rise to a “chiplet marketplace”. Alphawave Semi announced a breakthrough with the tapeout of the industry's first off-the-shelf, multi-protocol I/O connectivity chiplet. This technology can seamlessly integrate into larger SoCs or multi-chip modules, giving designers increased flexibility and more connectivity options for different applications. Reflecting on the Chiplet Summit this past spring, I remember listening to Tony Chan Carusone, Alphawave’s CTO, as he passionately described the significance of chiplets and how Alphawave will foster a more open and democratic design ecosystem. Now, just one quarter later, his team has launched a technology that brings this vision much closer to becoming a reality. Adding to their momentum, Alphawave also announced an expanded partnership with Samsung Foundry to advance the industry's progression toward cutting-edge semiconductor nodes, crucial for the evolution of advanced compute. Samsung’s roadmap in sub-4nm technologies includes FinFET and Gate-All-Around, which are designed to maximize performance in the post-Moore era. Initially targeted for data centers and HPC, these innovations may trickle down to low-cost, AI-enabled edge devices, putting the transformative power of AI into the hands of billions of people worldwide. This news follows Alphawave’s early June announcement of their new partnership with Arm, aimed at unlocking the potential of processors. Alphawave will give Arm access to their robust portfolio of connectivity solutions and silicon IP, which Arm will integrate into their proven portfolio of compute cores. At a time when AI technology is accelerating at breakneck speed, many companies are searching for ways to boost compute performance and address the challenge of moving data between compute and memory more efficiently. While multiple startups offer unique solutions, many customers want battle-tested technology to avoid the risk of investing in solutions that may become obsolete by the time they reach tapeout. Alphawave has developed vital solutions and in doing so, has increased design flexibility, built bridges to new process technologies, and shortened the path to market. “Connectors” are often the catalysts of innovation and Alphwave’s is a testament to that word. I’m encouraged by what I’ve seen from Alphawave and strongly believe they stand at the forefront of an emerging chiplet marketplace. More broadly, I believe their technology serves to democratize design technology and will give emerging companies more options to bring their brilliant ideas to life. #semiconductorindustry #ai #chiplets

112

🚀 #SKHynix: Leading the Charge in #High #Bandwidth #Memory 💻 As the landscape of high bandwidth memory (#HBM) evolves, SK Hynix is firmly positioned at the forefront, emerging as the industry leader. Recent Q3-24 results showcase remarkable financial growth, with 8.2% QoQ and 88% YoYrevenue increases. This momentum is largely driven by the surging demand for #AI #technologies, where HBM plays a pivotal role. 📈✨ Key Highlights: 🔹 Dominance in HBM: Hynix has solidified its role as the primary supplier for #Nvidia’s #H100 #GPU, crucial for #AI #applications. The HBM segment is projected to account for 40% of revenue by Q4-24, emphasizing its strategic importance. 🔹 Impressive Financial Performance: Significant increases in gross profit and operating profit illustrate Hynix’s ability to capitalize on its HBM leadership. As the demand for high-speed memory continues to rise, this trend is expected to persist. 🔹 Challenges Ahead: Despite the strong performance, Hynix faces critical challenges related to underinvestment in capacity. With #Samsung expanding its memory production, Hynix must address potential capacity constraints to meet growing demand and maintain its competitive edge. 🏭⚠️ 🚀 #Strategic Expectations for #Sustaining #Leadership: 1) Increase Capital Investments: To keep pace with rising demand, Hynix needs to prioritize capital expenditures. Investing in new production facilities and upgrading existing ones will be essential for scaling HBM output and enhancing manufacturing efficiency. 2) Innovation in Technology: Continued innovation in memory technology and packaging methods is vital. As the industry evolves, Hynix must lead in developing next-generation HBM solutions that can deliver even higher speeds and efficiency. 3) Strengthen Partnerships: Collaborating with key players in AI and semiconductor ecosystems, such as TSMC and Nvidia, will be crucial for aligning product offerings with market needs and driving mutual growth. 4) Sustainability Initiatives: As environmental concerns grow, focusing on sustainable production practices will not only enhance brand reputation but also ensure compliance with global standards, positioning Hynix favorably in a competitive market. 5) Market Adaptability: Being agile in responding to market fluctuations and customer demands will be essential. Hynix should continuously analyze market trends and be prepared to pivot strategies to maintain leadership. In summary, while SK Hynix has a strong foundation and a promising trajectory, its ability to sustain leadership will depend on strategic investments, technological advancements, and adaptive market strategies. Let’s watch how they navigate these challenges ahead! #SKHynix #HBM #AI #Semiconductors #Innovation #TechnologyLeadership #BusinessGrowth #DataDriven #FutureOfMemory #Sustainability #Partnerships #clausaasholm https://lnkd.in/dyc-S4K3

2

1 Comment

Happy #CostModelMonday! I did a #fan-out panel-level versus wafer-level comparison at #IMAPS DPC this year, and I'm expanding on it for the IMAPS Symposium. I had a single #yield chart in my analysis previously, looking at yield crossover points in relation to how expensive the incoming chip was. I did a new analysis that looks at when the crossover point occurs for two different panel sizes. In this chart, the same package is built with fan-out packaging on a 300mm wafer and from two different panel sizes. The yield for the 300mm wafer example is held steady at 99%, while the yield is changed for the panel-level processes. For the package built with the 515x510mm panel, the crossover point occurs just past 95%. However, the crossover point is closer to 94% for the package from the 600x600mm panel. In other words, a little more yield loss can be sustained by the package from the larger panel before it no longer competes with a higher yielding package from the wafer.

15

Arm's taken over the NASDAQ building in New York City to share why the world's biggest technology companies are innovating on the Arm compute platform for the age of AI. By building on the industry's most power-efficient and highest-performing compute platform, Arm's empowering the technology ecosystem to unlock all that AI promises. See how: https://okt.to/wW3Q9X

18

You are on a journey to optimize your factory, or your entire factory network. You find you are making headway, but still missing that "special sauce" that puts you in the position you want to be. It's time to start looking at your product scheduling system. Unleash the power of high-speed full factory simulation, in-line and fully integrated into your scheduling process. Make production decisions based on the future state of your factory, not just the current state. Intel® Factory Pathfinder can enable this, giving you a new power and control over your production systems like you've never thought possible.

72

How breathtaking it is, we are now standing at the turning point of the industrial revolution: “If you run a factory, your throughput directly correlates to your revenue, the quality of service, the number of people who can use your service. We are now in a world where datacenter throughput utilization is vitally important. It was important in the past, but people don’t measure it. Today, every parameter is measured. When something is a factory, its operations directly correlate to the financial performance of the company,” Jensen Huang (COMPUTEX 2024) Watch the insightful keynote by Jensen Huang at COMPUTEX 2024: [Link to the video](https://lnkd.in/dB9aj2qr) #computex2024

2