Jeffrey Welser’s Post

#GenAI and #ChipDesign are the two hottest topics today. Ruchir Puri Thanks for your leadership in organizing this conference.

#day19 Researchers have leveraged machine learning, specifically neural networks, to revolutionize the sizing of operational amplifiers (opamps). Traditionally, designing opamps is a complex, time-consuming task requiring expert knowledge. This new approach uses neural networks to predict performance metrics based on design parameters, significantly streamlining the process. - Efficiency Boost: Reduces design time and manual iterations. - High Accuracy: Neural networks accurately predict performance, optimizing the design process. - Scalability: Applicable to various opamp designs, enhancing versatility. This innovation marks a significant step towards automating and optimizing analog circuit design, making it faster and more efficient! https://lnkd.in/eSMcfZJp

This article "Handling the Challenges of Building HPC Systems We Need" by Arteris's CEO K. Charles Janac in EE Times | Electronic Engineering Times charts an interesting vision towards our AI-enabled future and its compute requirements. Enjoy! "Artificial intelligence (AI) is a key driver behind accelerating the knowledge doubling curve. As AI systems become more advanced, they can analyze data and make predictions more accurately, leading to faster progress in various fields. All this information translates into new capabilities that will eventually impact human development. [...] The increasing prevalence of AI, virtual and augmented reality, and other emerging technologies will require robust communication networks that can handle the vast amounts of data generated by these systems. #Arteris #Semiconductor #ArtificialintelligenceAI  #NetworkOnChip #SystemOnChip #PhysicallyAwareNoC https://bit.ly/45JQg2t

Here is a great (detailed but accessible) read for this long Easter weekend: IEEE explaining the advances in #semiconductor chip design that underlie the recent advances in #artificialintelligence. A very interesting section includes mention of #energyefficiency advances: “Over the past 15 years, the semiconductor industry has increased energy-efficient performance about threefold every two years.” This trend is expected to continue. Great news for the modeling that will be required to develop #industrialAI applications further.

How has AI been used to Innovate Chipmaking in 2023? Case Studies: 1. Google's AI-Powered Chip Design: Google has successfully used AI, particularly Reinforcement Learning, to design its next-generation Tensor Processing Unit (TPU) chips. The AI-designed floorplans outperformed those created by human experts, highlighting AI's potential in optimizing complex designs. 2. DARPA's POSH Program: The Defense Advanced Research Projects Agency (DARPA) initiated the POSH (Posh Open Source Hardware) program to leverage AI in chip design. This program focuses on creating open-source EDA tools that utilize ML to automate chip design, paving the way for more innovative and efficient design processes. 3. Synopsys' AI Innovations: Synopsys launched Synopsys.ai™, the first full-stack AI-driven suite of EDA solutions. This suite, including Synopsys Design Space Optimization AI (DSO.ai™), has been instrumental in developing advanced-node chips and optimizing resource consumption for companies like STMicroelectronics and SK hynix. 4. NYU Tandon's AI-Enabled Chip Fabrication: Researchers at NYU Tandon School of Engineering have used LLMs (Large Language Models) in chip design, transforming English instructions into Verilog code for functional and verification purposes. This novel approach could potentially reduce human error and shorten design times. These developments signal a transformative era in semiconductor design, where AI's integration is not just enhancing efficiency but also driving innovation at an unprecedented scale. As we continue to harness the power of AI, the future of chipmaking looks more dynamic and promising than ever before. I’d love to hear your thoughts on these advancements. How do you see AI shaping the future of semiconductor design? Let's discuss in the comments below!

GPT-5 will be a multimodal model with textual and visual inputs and outputs. AGI is anticipated to merge the functionalities of natural language processing, image recognition, data from sensors like cameras and microphones, pattern recognition, predictive analysis, abstraction, and logical reasoning

Researchers have developed a mechanical computer inspired by the Japanese art of kirigami, using 64 interconnected polymer cubes that can be rearranged to store, retrieve, and erase data without any electronic components. The cubes represent binary data that can be manipulated by physically pushing them up or down, allowing the computer's configuration to be changed to correspond to different computational states. This concept opens doors for physical encryption-decryption systems and haptic displays in three-dimensional contexts. #Innovation #Technology #ComputerScience

NVIDIA and HP Inc. have joined forces to integrate NVIDIA CUDA-X™ data processing libraries into HP AI workstation solutions. This collaboration aims to accelerate data preparation Jensen Huang, founder and CEO of NVIDIA, highlights the significance of accelerating pandas, a fundamental tool for data scientists, with zero code changes. This enhancement enables data scientists to process data swiftly, significantly reducing processing time and enabling handling of larger datasets for training generative AI models. Enrique Lores, president and CEO of HP Inc., emphasizes the importance of providing fast access to software and systems for data science, which forms the foundation of AI development. The integration of NVIDIA AI software and GPU acceleration into HP AI workstations offers a robust solution to meet customer needs effectively. Read more - https://lnkd.in/eZgPY2vJ

Alberto Sangiovanni-Vincentelli, professor and Edgar L. and Harold H. Buttner Chair of Electrical Engineering and Computer Science Department at the University of California, Berkeley, sat down to an exclusive interview with DIGITIMES Asia at the sidelines of SEMICON Taiwan 2023 on September 8. As a legend who created two of the world's most prominent electronic design automation (#EDA) companies and who has accumulated decades of knowledge and experience in the #semiconductor industry, Professor Sangiovanni-Vincentelli shed light on how far #generativeAI can go as a tool to advance the semiconductor industry. He also shared his thoughts on the solution to geopolitical tensions surrounding the semiconductor #supplychain. #China

Princeton professor of electrical and computer engineering Naveen Verma is leading a revolution of AI hardware. Backed by the US government, Naveen Verma developed a new kind of “super” chip with 10 times more computational power per square millimeter and 10 times more energy efficiency. Hardware, which at one point was a limiting factor for the expansion of artificial intelligence is soon becoming just as advanced as its software counterpart. With great advancements in both traditional and quantum computing hardware/software, who knows what the future will hold. https://lnkd.in/efFDm53Z