weixin
Author page description
September 9
- Guangzi Technology Launches Its Self-owned Intellectual Property Single-Wave 100G Driver and TIA Series with Support for LPO Architecture
Unknown
➀ Guangzi Technology introduces its single-wavelength 100G Driver and TIA series with support for LPO architecture, designed for 400G/800G high-speed communication with advantages of low power consumption, low latency, and low cost. These are suitable for data center and AI high-speed interconnection. ➁ The LPO architecture, a new type of optical communication module, reduces dependency on DSPs, lowering power consumption, latency, and cost, making it suitable for high-performance computing, 5G networks, and cloud computing. ➂ Guangzi's solutions include 4x100G MZM driver PHMD1124 and 4x100G TIA PHTA11245, as well as 4x100G VCSEL driver PHVD1124 and 4x100G TIA PHTA11242, offering high bandwidth, low power consumption, and excellent signal amplification. ➃ Guangzi's LPO solutions are adaptable to various boards, with reliable signal integrity, balance, and flexibility, making them ideal for next-generation data center large bandwidth optical connections, AI, and machine learning. - Samsung Unveils Industry's Thinnest LPDDR5X Packaging: Just 0.65 Millimeters Thick!
SSDFans
➀ Samsung has announced the mass production of the thinnest LPDDR5X modules in the industry, with a thickness of approximately 0.65 millimeters; ➁ The new packaging is 0.06 millimeters thinner than the standard LPDDR5X, improving thermal resistance by 21.2%; ➂ The thinner LPDDR5X封装 enhances internal airflow in smartphones, improving thermal management and battery life.
September 8
- Huawei Bets Big on 'High-End Bending Pass' with 257% Growth
Not specified in the provided content
➀ Huawei is launching a revolutionary product, a three-foldable phone, after a five-year development period; ➁ The smartphone industry is experiencing a resurgence, driven by the replacement cycle of phones and the integration of AI technology; ➂ The foldable phone market is growing rapidly, with a 125% year-on-year increase in sales in the second quarter of 2024; ➃ Huawei is heavily investing in foldable phone technology, seeing it as a key to competing in the high-end market. - Huawei's Most Expensive Phone Is a Hot Seller: Mate XT Master Edition with 2.38 Million Pre-Orders!
Unknown
➀ The Huawei Mate XT Master Edition has exceeded 2.38 million pre-orders since its reservation began on September 7th. The phone is expected to be officially launched on September 20th. ➁ Both online and offline channels have seen strong demand. The phone is available in two colors and two storage options. ➂ There are speculations that the phone's price could exceed 20,000 yuan, making it Huawei's most expensive phone to date. The high pre-order numbers have sparked discussions about potential scalping activities. - The Real Estate Boom in Taiwan: TSMC's Influence
微信公众号
➀ TSMC's presence in Taiwan's semiconductor industry is a key indicator of the real estate market's performance; ➁ Over the past five years, the ten places with the largest real estate price increases in Taiwan are all related to TSMC; ➂ Seven of them are near the Hsinchu Science Park, Southern Taiwan Science Park, and Central Taiwan Science Park; ➃ The price of houses in Zhubei City, Hsinchu County, which has a TSMC wafer factory, has increased by 118% in five years, propelling the city from a 'third-tier' to a 'new first-tier' city; ➄ The real estate prices in Taipei and New Taipei have lagged far behind, much lower than the island's average; ➅ TSMC's stock price has increased by over 500% since 2019, with a market value close to 90 billion USD, nearly equal to the combined market value of Tencent, Alibaba, Meituan, NetEase, JD.com, and Baidu; ➆ TSMC's revenue last year was 69 billion USD, almost 10% of Taiwan's GDP; ➇ Bloomberg predicts that with the introduction of a large number of EUV photolithography machines, TSMC's electricity consumption will reach 12.5% of Taiwan's total by 2025.
September 4
- Nvidia's $3 Trillion Chip Empire: The Power of 19 Strategic Acquisitions
微信公众号
➀ Nvidia has accelerated its acquisition strategy, buying one AI and data center-related company per month over the past four months. ➁ Key acquisitions include Run:ai, Deci, Shoreline, and Brev.dev, enhancing Nvidia's technical capabilities in these areas. ➂ Nvidia's acquisitions are part of a broader trend over the past five years, with 13 successful acquisitions since the 69 billion dollar Mellanox deal. ➃ The company focuses on end-to-end solutions, acquiring firms in communication network technology, object storage software, and data center management software to enhance its B2B offerings. ➄ Nvidia aims to become a platform-type enterprise by expanding into model miniaturization, deep learning development platforms, metaverse, and autonomous driving through acquisitions. - LG Display Sees 1400 Employees Opt for Voluntary Resignation
满天芯
➀ LG Display, a major Korean panel manufacturer, has implemented a voluntary resignation plan to reduce labor costs, with over 1400 employees applying. ➁ Employees who choose to resign receive compensation equivalent to 36 months of fixed salary. ➂ The company is focusing on OLED technology and phasing out its LCD business, leading to the closure of several factories. - Smartphone AP Market: UNISOC's Shipments Surge 42%, HiSilicon Captures 2.7% Share
芯智讯
➀ MediaTek maintains the top spot in the smartphone AP market with a 40% shipment share. ➁ Apple leads in revenue share with 39%, despite a slight decline. ➂ Qualcomm's shipments grow 6%, but revenue declines 6%, indicating challenges in the high-end market. ➃ Samsung's Exynos series sees a 9% increase in shipments and a 71% rise in sales, driven by flagship models. ➄ UNISOC's shipments surge 42%, though its market share remains low due to focusing on entry-level devices. ➅ HiSilicon sees a modest 2.7% market share in shipments, with stable revenue. - Affordable Underwater Communication Modems Built on Raspberry Pi
IEEE电气电子工程师学会
➀ Researchers at the University of Padova are developing low-cost underwater communication modems and transducers, aiming to make water acoustic communication more accessible. ➁ The SubSeaPulse modem, priced at a tenth of the market rate, is built on a Raspberry Pi base with additional sound card hardware and a front end. ➂ The team is also working on a cost-effective transducer, reducing the price from over $2000 to around $400, using a modified device typically used for listening to marine mammals. - Intel's Most Efficient x86 Chip: Revolutionizing AI PC Performance
量子位
➀ Introduces Intel's Core™ Ultra 200V series, claiming the title of the most efficient x86 processor with significant improvements in CPU, GPU, and AI performance. ➁ Features a new Xe2 graphics microarchitecture that enhances mobile graphics performance by 30% and supports advanced ray tracing. ➂ Boasts a total AI performance of 120TOPS, integrating CPU, NPU, and GPU capabilities, and claims superior battery life compared to competitors. - XGIMI Technology: Sharp Decline in First Half Net Profit
硬氪分析
➀ XGIMI Technology reported a 1.66% decline in revenue and a 95.58% drop in net profit for the first half of 2024. ➁ The company attributed the profit decline to inventory clearance of older products and weak domestic demand. ➂ XGIMI is adjusting its product strategy to include lower and higher price ranges to adapt to market trends. ➃ The domestic projector market is experiencing a price war due to increased competition and technological advancements, leading to lower average prices. ➄ XGIMI is expanding its overseas presence to mitigate domestic market challenges, with significant growth in international revenue. - NVIDIA Grace Hopper Superchip Architecture: Revolutionizing HPC and AI Workloads
微信公众号
➀ NVIDIA Grace Hopper Superchip integrates NVIDIA Hopper GPU and NVIDIA Grace CPU, offering a high-bandwidth NVLink-C2C interconnect. ➁ It provides up to 900 GB/s bandwidth, enhancing memory consistency and simplifying programming models. ➂ The architecture supports up to 150 TB of accessible memory across multiple GPUs, enabling efficient large-scale AI and HPC applications. - Large Model Training and Inference on Heterogeneous GPU Clusters
竹言见智
➀ HAP system optimizes SPMD model training on heterogeneous clusters by automating tensor sharding and communication methods. ➁ The system uses A* search algorithm to generate distributed programs and optimize sharding ratios. ➂ OEF framework enhances resource efficiency and fairness in heterogeneous GPU clusters, improving training throughput by up to 32%. - First Investor of 'Black Myth: WuKong': Will Never Invest in Web3 Games, These People Don't Love Games
微信公众号
➀ Daniel, the first investor of 'Black Myth: WuKong', expressed strong opposition to investing in Web3 games, stating that those who shift to Web3 do not truly love games. ➁ He emphasized the importance of focusing on the team rather than being swayed by new trends. ➂ Daniel's company, Hero Games, was the sole investor and largest external shareholder of Game Science until Tencent's entry in 2021. - HotChip2024 Postscript: Why RDMA is Inappropriate for ScaleUP Networks
算力猩
➀ Discusses the limitations of using RDMA in ScaleUP networks, highlighting issues with latency, CPU overload, and chip area constraints. ➁ Explores alternative protocols suitable for ScaleUP, emphasizing the benefits of Ethernet and direct memory access. ➂ Analyzes the challenges and solutions in implementing efficient interconnects for AI accelerators, focusing on network convergence and memory semantics. - Winning in the Humanoid Robot Industry: Entrepreneurship Like Operating a Large Corporation with Peng Zhihui
晚点 LatePost
➀ Peng Zhihui, known as 'Zhihuijun' on Bilibili, co-founded Zhiyuan Robotics as CTO, aiming to commercialize humanoid robots. ➁ The company adopts a 'high-stakes, high-play' approach, developing multiple generations of robots and diversifying product lines within a short period. ➂ Zhiyuan Robotics has raised over 1.5 billion RMB in funding, valuing the company at 7 billion RMB, with investments from top-tier firms like Hillhouse, Sequoia, and Gaorong. - From Rich Brat to AI Pioneer: The Remarkable Journey of Herbert Simon
超模君
➀ Herbert Simon, born into a wealthy family, was initially a rebellious 'rich kid' who excelled in nothing but partying and fighting. ➁ His life took a dramatic turn when he met Dorothya Pye, who encouraged him to pursue academic excellence. ➂ Simon became a renowned expert in management and later delved into computer science, becoming one of the pioneers in artificial intelligence. ➃ He developed the first AI program and received numerous prestigious awards, including the Turing Award and the Nobel Prize in Economics. - Accelerating MEMS Speaker and Active Cooling Chip Development with Series C1 Funding
微信公众号
➀ Cloudview Capital leads C1 round investment in xMEMS Labs to support the commercialization of their innovative MEMS speakers and ultra-thin active cooling chips. ➁ xMEMS introduces a 1mm-thick active cooling product in August 2024, designed for high-performance smartphones and AI chips. ➂ The company's Cypress speaker, launched in late 2023, significantly enhances low-frequency sound pressure levels, marking a breakthrough in MEMS speaker technology. - Lumentum's Financial Performance and Market Dynamics in 2024 Fiscal Year
微信公众号
➀ Lumentum's revenue for the 2024 fiscal year is projected to decline to $1.3592 billion from $1.767 billion in the previous year. ➁ The company has shifted its focus from traditional products to core technologies like chips and coherent optical modules. ➂ Lumentum has expanded its customer base, notably adding Google as a significant client, reflecting its strategic move into the AI market through acquisitions.
September 3
- Breakthrough in Key Technology: Trench-Type SiC MOSFET Chips
半导体技术天地
➀ The National Third-Generation Semiconductor Technology Innovation Center (Nanjing) has successfully developed trench-type SiC MOSFET chips, breaking through the technical barriers of planar SiC MOSFET chips. ➁ These new chips offer a 30% improvement in conduction performance compared to planar types. ➂ The trench-type SiC MOSFET chips are expected to be applied in areas such as new energy vehicle electric drives, smart grids, and photovoltaic energy storage within a year, potentially increasing the range of electric vehicles by about 5% and reducing chip costs. - Huawei's Showtime: A Whale Falls, All Things Rise
君临财富
➀ Huawei's performance dominates the stock market from late August to the end of the year, with significant impacts on related industries. ➁ In 2024, Huawei reported a 34.3% increase in sales revenue to 417.5 billion yuan, driven by strong smartphone sales and other tech products. ➂ Huawei's upcoming events include the Huawei HiSilicon Connect conference, featuring new technologies like StarFlash, and a series of product launches, including a revolutionary tri-fold smartphone. - Huawei Announces Two Major Developments: Mate XT Master and 2024 H1 Financial Results
旭日大数据
➀ Huawei announces the upcoming launch of the Mate XT Master, a new foldable phone. ➁ Huawei reports strong financial results for the first half of 2024, with a 34.3% increase in sales revenue. ➂ Huawei continues to innovate in satellite communication and foldable display technologies. - The Future of Semiconductors: Industry Predictions and Market Trends
半导体行业观察
➀ The semiconductor market is dominated by PC and communication sectors, accounting for 57% of total sales. ➁ SEMI predicts a 20% growth in global semiconductor revenue in 2024, driven by AI chips and memory. ➂ Automotive semiconductors are expected to exceed $88 billion by 2027, driven by ADAS, EV, and IoV technologies. ➃ AI chip market is projected to reach $138 billion by 2028, with Nvidia leading the market. - The Future of Semiconductors: Industry Predictions and Market Trends
半导体行业观察
➀ The semiconductor market is dominated by PC and communication sectors, accounting for 57% of total sales. ➁ SEMI predicts a 20% growth in global semiconductor revenue in 2024, driven by AI chips and memory. ➂ Automotive semiconductors are expected to exceed $88 billion by 2027, driven by ADAS, EV, and IoV technologies. ➃ AI chip market is projected to reach $138 billion by 2028, with Nvidia leading the market. - HotChip2024-Day1: AI Chip Highlights
芯东西
➀ AMD MI300X features Infinity Fabric Advanced Package for high-speed interconnects and supports FP8 operations. ➁ Intel Gaudi 3 introduces a 256x256 matrix multiplication engine and enhanced interconnectivity. ➂ SambaNova SN40L offers large on-chip memory and a hybrid Mesh/Ring interconnect structure. ➃ FuriosaAI introduces a Tensor Contraction Processor based on Einstein summation notation. ➄ Tenstorrent's architecture emphasizes asynchronous memory access and a scalable 2D/3D Torus network. ➅ BRCM focuses on CPO technology for high-density optical interconnects. ➆ IBM Telum 2 integrates a DPU and AI accelerator in its next-gen mainframe processor. - Broadcom's AI Chip Revenue Surges, Nearing Nvidia's Dominance
微信公众号名称或转载的出处
➀ Broadcom's AI-related revenue is expected to reach $32.5 billion in the third quarter, tripling from the previous year. ➁ The company's total revenue for the second quarter was $124.87 billion, a 43% increase year-over-year. ➂ Analysts predict Broadcom's AI revenue will exceed $110 billion this year, with ASIC and networking products making up 65% and 35% respectively. ➃ Broadcom is seen as the second largest AI chip supplier globally, trailing only Nvidia, and leads in custom chip market share for 7nm, 5nm, and 3nm technologies. ➄ Citibank views Broadcom as the next hot AI stock, citing growth in new customers and the acquisition of VMware. - OpenAI's Custom Chip Revealed: TSMC's 1.6nm for Sora
量子位
➀ OpenAI's first custom chip will utilize TSMC's cutting-edge A16 process, specifically designed for the Sora video application. ➁ OpenAI has reserved TSMC's A16 production capacity alongside Apple, aiming to reduce external AI chip dependency. ➂ TSMC's A16 process, entering the angstrom-level node for the first time, promises significant enhancements in speed, power efficiency, and chip density. - How to 3D Print Glass?
IEEE电气电子工程师学会
➀ Researchers at the University of Notre Dame are developing a robotic platform that uses lasers and computer-controlled stages to replicate the traditional glassblowing technique, aiming to create transparent solid structures for optical, microfluidic, and photonic devices. ➁ Ed Kinzel, an associate professor at Notre Dame, aims to capture the geometric freedom of hand-blown glass methods with the precision of modern 3D printing, addressing the challenges of rapid prototyping in glass. ➂ The team has developed a digital glass forming technique using a CO2 laser to locally melt small diameter glass filaments, allowing for the creation of 3D shapes through controlled deformation and interaction with a substrate. - Huawei Mate 70 Series: Four Major Camera Upgrades Unveiled, Poised to Become a 'Black Myth' in Mobile Phones
Industry Weatherman
➀ Pixel Upgrade: Mate 70 Pro features a 6000MP main camera, 4800MP ultra-wide, and 4800MP telephoto, supporting 5x optical and 100x digital zoom. ➁ Diamond Module: The camera layout adopts a diamond pattern for enhanced aesthetics and grip. ➂ One-Inch Large Sensor: Standard and Pro versions use advanced sensors like OV50H and OV50K, with LOFIC technology and a large variable aperture. ➃ Telescopic Lens: Mate 70 series may introduce the ultra-convergent telescopic lens from Pura70 Ultra, enhanced by XMAGE imaging technology. - The Challenges and Evolution of MOSFET Modeling
微信公众号
➀ The article discusses the challenges in creating accurate models for MOSFET devices due to the complexity of quantum mechanics and multi-scale interactions. ➁ It highlights the role of BSIM models in semiconductor design, which incorporate both theoretical physics and empirical data to describe device behavior. ➂ The limitations of current models and the potential for future advancements through computational methods are explored.