Dinesh Maheshwari

The combination of software-defined networking (SDN) and photonics is emerging as a potential solution for network optimization. This blend of technologies aims to offer dynamic traffic routing and resource allocation, responding to real-time network conditions and user demands. It's a development that could shape the future of telecom and datacom, introducing programmability, adaptability, and flexibility. Authored by iPronics Programmable Photonics: https://lnkd.in/dSkM3aP7 #Photonics

55

3 Comments

Got a chance to discuss TCG Key Per I/O capabilities and its possible use cases with confidential compute at the Confidential Compute Consoritium's TAC call yesterday. Please see the link below. The topic is covered from around 20 mins into the video and runs till about the 1-hour mark. https://lnkd.in/gvupy8iP

20

This letter proposes a scalable channel estimation and reflection optimization framework for #reconfigurable #intelligent #surface (#RIS)-enhanced #orthogonal #frequency #division #multiplexing (#OFDM) systems. Specifically, the proposed scheme firstly generates a training set of RIS reflection coefficient vectors offline. For each RIS reflection coefficient vector in the training set, the proposed scheme estimates only the end-to-end composite channel and then performs the transmit power allocation. As a result, the RIS reflection optimization is simplified by searching for the optimal reflection coefficient vector maximizing the achievable rate from the pre-designed training set. The proposed scheme is capable of flexibly adjusting the training overhead according to the given channel coherence time, which is in sharp contrast to the conventional counterparts. Moreover, the authors discuss the computational complexity of the proposed scheme and analyze the theoretical scaling law of the achievable rate versus the number of training slots. ---- Jiancheng An, Qingqing Wu, Chau Yuen More details can be found at this link: https://lnkd.in/eK6dgdST

7

📣 The IEEE Transactions on Very Large Scale Integration (VLSI) Systems is proud to recognize one of its Best Associate Editors of 2023, CHIRN CHYE BOON. Our Associate Editors' contributions are instrumental in maintaining the journal's industry-leading standards for publishing impactful research. What makes a research paper stand out in the field of VLSI? Associate Editor Chirn Chye Boon shares his insights! 👇 Q: As an Associate Editor, you play a vital role in maintaining the high quality of the journal. What advice would you give to researchers who are submitting papers to the IEEE Transactions on VLSI Systems for the first time? A: Your paper could make a difference in the world’s technical advancement, thus making our world a better and more comfortable place for all. Do explain your work as best as you can and verify your work so that it leaves no doubt on its merits. Q: The field of VLSI is constantly evolving. In your opinion, how has the role of the Associate Editor changed over the years to keep pace with these advancements? A: Associate Editor works most efficiently only when there are many expert reviewers willing to help to perform the review. The success on performing the role of Associate Editor owes deeply to these capable and selfless reviewers, I want to say a “Big Thank You! “, for their great help and support. Q: Beyond technical merit, what are some of the key qualities you look for when evaluating a paper for publication in the IEEE Transactions on VLSI Systems? A: To put it simply, a clear, cohesive discussion of the work done, complete with proper verifications and detail comparisons with the state-of-the-art works is rudimental for a successful paper. Thank you for your contribution to the IEEE Transactions on Very Large Scale Integration (VLSI) Systems! #TVLSI #BestAssociateEditor #VLSI

17

1 Comment

total power will need to include power of digital layers also but PHY power is impressive

15

There was HUGE news in AI hardware this week - chip startup Etched broke a year-long silence and announced they closed a $120M Series A, a story that has taken the semiconductor community by storm. Co-founders Gavin Uberti and Chris Zhu first made headlines in May 2023 when they closed a seed round of $5M and shared their mission to create an ASIC to optimize transformer model training and challenge NVIDIA head-on. Many in the industry then referred to Uberti and Zhou as “two Harvard dropouts trying to challenge Goliath.”. The term “dropout” was used in a perjorative sense. Just a year later, they’ve demonstrated there is merit to their thesis, and visionary investors have taken notice. The founders believe scale is what the industry needs to harness superintelligence; which means unprecedented compute power that can bring AI to everyone, everywhere. They believe GPU performance is hitting a wall, and the existing hardware has a huge moat around it that’s keeping newer and better solutions out of the competitive arena. Uberti and Zhou decided that was unacceptable and started Etched to tackle the problem head-on. They’ve created the world’s first transformer ASIC, called SOHU. They claim one SOHU can replace 160 NVIDIA H100 GPUs, which would result in phenomenal savings in cost and power consumption, enabling companies around the world to make this AI transformation. Furthermore, the Etched team is doing something that would give most startup founders crippling anxiety - they’re going all-in on this application and pouring EVERYTHING they have into it. Uberti’s comment about this put a smile on my face. He said, “We’re making the biggest bet in AI. If transformers go away, we’ll die. But if they stick around, we’re the biggest company of all time.” As a recruiting leader for over 18 years, I’ve met hundreds of leaders and founders but RARELY have I encountered such a precocious team that is willing to commit to a single cause. Startups are hard enough, but building a startup to compete against one of the largest companies in the world is an order of magnitude more challenging. So as Etched builds their team, passion will a “nice to have” but commitment and perseverance will be mandatory. Robert W. Etched’s Operating Chief commented, “The reason we dropped out of school and we’ve convinced so many people to leave these chip projects — [because] this is the most important thing to be working on.” I've been saying for over 18 years that engineers are inherently curious and compelled to act when presented with meaningful challenges and opportunities to change the world. Etched is a prime example of this and while details about the team are sparse, they did attract luminary former CTO of Cypress, Mark Ross, to join them, which gives credence to Wachen’s comment. While the Etched team is in for a long, difficult journey, their future could be remarkable. What’s your take? #semiconductorindustry #artificialintelligence #transformermodels

179

6 Comments

Our game changing inference performance will enable new applications

31

By successfully implementing a Variational Quantum Eigensolver (VQE) in a 16-dimensional photonic system, the research highlights the potential of qudit-based quantum computing for practical applications, such as drug development, battery improvement and solving complex challenges like climate modeling. https://lnkd.in/eZMB4XHX https://lnkd.in/e-WMbGgH #technology #tech #innovation #business #engineering #technews #science #software #huawei #gadgets #design #electronics #apple #programming #android #coding #security #samsung #ai #smartphone #cybersecurity #education #computer #mathematics #compositivelogic #pro #marketing #technologynews #artificialintelligence #math #uni #student #college #phd #research #quantumcomputing #quantum #quantumphysics #quantummechanics #physics #quantumentanglement #technology #quantumcomputer #artificialintelligence #ai #China #machinelearning #computerscience #tesla #nasa #dwave #cern #maths #innovation #settheory #college #it #quantumtechnology #datascience #cryptocurrency #mit #vision #blockchain #sustainability #it #computerscience #logic #critical #thinking #theoreticalphysics #stanfordphysics #data #finance

1

#Resistive #memory #devices exhibit rich dynamics and enable efficient computations like stateful logic, temporal processing, and analog matrix computing. Their #nonvolatile #switching and #crosspoint #architecture can overcome the von Neumann bottleneck and provide parallelism, notably accelerating matrix-vector multiplication in #neuralnetworks. Prof. Zhong Sun, Prof. Yuchao Yang and colleagues from Peking University demonstrate that a passive #resistive #switching #circuit, represented as an #attractor #network model, provides high-capacity, compact and efficient solution for #associative #memory. Now out 👉 #naturecommunications Nature Portfolio

49

More advances in #Optical #neuralnetworks to implement #AI in low-power hardware

3

MODULAR SCALABLE Silicon Photonics GUARANTEED: INSIDE OpenLight WITH Adam Carter. We get many questions from young photonics companies trying to scale up production to serve new markets and increase their turnover. And that’s where Optica can help. We have access to the leading CEOs and CTOs in over 600 photonics businesses who have a wealth of experience and are willing to share their advice with others. NEW INTERVIEW SERIES In this long-form interview series hosted on our YouTube channel, we explore the business strategies being developed by industry leaders. I’m joined at the questioner's table by Michael Hochberg, an authority in the Silicon Photonics industry. https://lnkd.in/d9TPJexs For this edition, we sat down with Dr Adam Carter, the CEO at OpenLight. Formed as a joint venture between Juniper Networks and Synopsys Inc, the company has a proven ability to integrate lasers, optical amplifiers, DFBs, EA-modulators, and photodetectors on a single chip. They grant their customers access to a comprehensive library of photonics components that are heterogeneously integrated on the PASIC. OpenLight’s approach removes the need to source and package separate lasers on silicon photonics chips and improves device performance. In our interview, Adam explains that their approach will be key to producing next-generation, pluggable transceivers for datacenters and AI/ML applications. We discuss their partnership with Optica corporate member Jabil. This move is to meet the needs of the global Silicon Photonics market, which, according to some reports, is expected to grow in value to around US$8.3 billion by 2030. Earlier this year, they also announced strategic partnerships with Optica members like VLC Photonics (Hitachi Group), an end-to-end integrated photonics services firm, now part of the Hitachi group. VLC Photonics now offers design and test services using the OpenLight process design kit (PDK). The collaboration increases the number of designs that OpenLight can support at any one time on the Tower Semiconductor PH18DA process. We also spotted a similar agreement with Optica corporate member Spark Photonics in the US. Who is Adam? Dr Adam Carter has more than 25 years of experience at large and small companies. He is perhaps best known for being chief commercial officer at Foxconn Interconnect Technology (FIT), Cisco, and, earlier, at Oclaro at the time of its acquisition by Lumentum. Originally from the UK, we started by asking Adam to sketch what it was like when the Telecom Bubble burst at the start of this Millennium. I hope you enjoy this conversation. Feedback encouraged.

86

6 Comments

DML Talk by prof. Kaushik Sengupta: "Pushing the limits of Chip-scale mmWave/THz Systems: Can AI-enabled Inverse Methods help?" Silicon-based Terahertz systems is a field that is only about a decade and a half old. In this time, we have seen a phenomenal growth of silicon systems operating at THz frequencies for a wide range of applications in sensing, imaging and communication. It can be argued that both the 'THz gap' and the 'technology and applications gap' is closing in meaningful ways in the THz range.  In the first half of the talk, I will highlight approaches that cut across electromagnetics, circuits, systems and signal processing to enable THz beamforming arrays, programmable THz metasurfaces with CMOS tiling (Nat Elec'20) and physically secure sub-THz links (ISSCC'20, ISSCC'24, Nature Elec'20, Nature Elec'21). In the second part of the talk, I will discuss some of our recent work in data-driven techniques for such high-frequency ICs. Traditionally, RF/mmWave/THz IC design has been in the domain of the expert, dominated by thumb rules and trial and error techniques, guided by human intuition and aided by time-consuming parameter sweeps, or ad-hoc population-based metaheuristic optimization methods. Designing these ICs is complex, time-consuming, requires years of expertise, and therefore, can be very expensive. There is also no reason to believe that this approach is optimal in any sense. This talk will discuss how inverse design with AI-based approaches can open a new design space and allow rapid designs on demand. It will discuss deep-learning based modeling and generative AI approaches, that are transferrable across process design technologies, for inverse design and automated synthesis of mmWave/sub-THz circuits and antennas. Date and Time Date: 18 November 2024 Time: 01:45 PM to 03:30 PM All times are (UTC+01:00) Amsterdam https://lnkd.in/e2kqrBbH

12

Mivium's Game-Changing Gallium Nitride Tech Seeks to Outperform Silicon, Reducing Toxicity and Enhancing Efficiency for Data Centers and AI Applications http://s4g.biz/14nov24 tl;dr: Eric Tsai’s Mivium focuses on gallium nitride (GaN) to replace silicon, improving efficiency and sustainability. GaN's fast-switching properties make devices smaller, more powerful, and less energy-intensive. Mivium’s unique, clean GaN production process avoids the toxic materials typical in semiconductors. Nevada’s tech-friendly policies support Mivium’s plans to establish a new “gallium valley” in Reno. By crowdfunding, Eric brings everyday investors into Mivium’s journey, making early-stage investing accessible. https://lnkd.in/e2nunP4g #ImpactCrowdfunding #DiverseFounders #SocialEntrepreneurs #CommunityCapital #ImpactInvestors #RIC #InvestmentCrowdfunding #SuperCrowd #gallium

3

Fortaegis Technologies emerges from stealth -- unveils the company's 5nm Secure Processing Unit (SPU). The SPU is designed to secure and scale collaborative AI while addressing critical challenges in data security and scalability. #AI #CollaborativeAI #Stealth https://lnkd.in/edx8mEaW

22

Researchers from NVIDIA, CMU and the University of Washington Released ‘FlashInfer’: A Kernel Library that Provides State-of-the-Art Kernel Implementations for LLM Inference and Serving FlashInfer incorporates a block-sparse format to handle heterogeneous KV-cache storage efficiently and employs dynamic, load-balanced scheduling to optimize GPU usage. With integration into popular LLM serving frameworks like SGLang, vLLM, and MLC-Engine, FlashInfer offers a practical and adaptable approach to improving inference performance. FlashInfer's unique features include: ✅ Comprehensive Attention Kernels: covering prefill/decode/append attention for various KV-Cache formats (Page Table, Ragged Tensor, etc.) for both single-request and batch-serving scenarios. ✅ Optimized Shared-Prefix Batch Decoding: 31x faster than vLLM's Page Attention implementation for long prompt large batch decoding. ✅ Efficient Attention for Compressed KV-Cache: optimized grouped-query attention with Tensor Cores (3x faster than vLLM's GQA), fused-RoPE attention, and high-performance quantized attention...... Read the full article here: https://lnkd.in/g69DqsgM Paper: https://lnkd.in/g4KK_9x4 GitHub: https://lnkd.in/gY5MeJYC NVIDIA NVIDIA AI Machine Learning Department at CMU Zihao Ye Lequn Chen Ruihang Lai Wuwei Lin Yineng Zhang Stephanie Wang Tianqi Chen Baris Kasikci Vinod Grover Arvind Krishnamurthy Luis Ceze

135

5 Comments

Cerebras Systems, under the leadership of CEO Andrew Feldman, made a significant leap in AI performance with its breakthrough in fast inference. By achieving lightning-speed processing of the world’s largest AI models, the company unlocked new possibilities for real-time, complex applications like multi-agent collaboration and advanced reasoning. This rapid inference technology is set to transform industries, enabling real-time responses and dramatically enhancing the capabilities of emerging AI systems. Feldman emphasized that this progress opens doors to use cases, driving the future of AI beyond simple query-based tasks. Read more here: https://lnkd.in/g__sm6Be

1

Z-Wave Alliance Principal member Silicon Labs wrote a blog discussing #Matter vs #ZWave and it's important to share with the industry. The blog highlights key differentiators between the two #standards, as well as how they coexist and complement each other in the smart home industry. Learn more about these standards, use cases, the next generation of Z-Wave with #LongRange (which supports up to 4,000 nodes per network) and more. https://lnkd.in/e4rdeniY

6

AI will be ubiquitous. With AI’s demanding computation, memory and power requirements, there is a big benefit in shifting to leading fabrication nodes. There was a report saying that the poor demand for Samsung Semiconductor’s trailing-edge nodes is forcing them to reduce production but for their leading-edge node, the future looks promising. Here are some numbers on the impact of nodes (from wiki) – Samsung’s optimized 3nm node achieves 45% reduced power usage, 23% improved performance and 16% smaller surface area compared to 5nm node; TSMC‘s 3nm FinFET chips reduce power by 25–30% at the same speed, increase speed by 10–15% at the same power and increase transistor density by about 33% compared to its previous 5 nm. Chip costs for equivalent capability come down with leading nodes. There’s a good reason why fabs invest heavily to switch to new nodes!! #AI #node #Semiconductor #SemiconductorFab #Foundry #Semi #MakrMicrosystems

16

1 Comment

The coordination of semiconductor subsidies between the European Union and the United States is crucial to avoid duplicating efforts and maximize industry innovation. Through the European Chips Act and U.S. Chips Act, significant public funds are being invested to enhance semiconductor capabilities and reshore manufacturing activities. Imec, a leading microchips research center, highlights the necessity for synchronized and coordinated research and development efforts across continents to ensure efficient use of resources and technological advancements. #imec #chip #cooperation #coordination #technology #semiconductor #innovation #research #microchip #semiconductors #semiconductormanufacturing #EU #USA #ChipsAct https://lnkd.in/dZ3SZvMd

15