如果您希望可以时常见面，欢迎标星收藏哦~ GPU 是加速高性能计算（HPC）工作负载的热门平台，广泛应用于人工智能和科学模拟等领 域。然而，学术界的大多数微架构研究仍基于 15 年前的 GPU 核心流水线设计。 本文通过逆向工程剖析了现代英伟达 GPU 核心，揭示了其设计的关键方面，并解释了 GPU 如何利用硬件-编译器协同技术，在执行过程中由编译器引导硬件工作。具体而言，研究揭示 了指令发射逻辑的工作机制，包括发射调度器的策略、寄存器文件及其相关缓存的结构，以 及内存流水线的多个特性。此外，分析了基于流缓冲区的简单指令预取器如何与现代英伟达 GPU 相适配并可能被采用。进一步研究了寄存器文件缓存和寄存器文件读端口数量对模拟准 确性和性能的影响。 通过对这些新发现的微架构细节进行建模，与之前的先进模拟器相比，我们的模型在执行周期的平 均绝对百分比误差（MAPE）降低了18.24%，与实际硬件（英伟达RTX A6000）相比，平均绝对 百分比误差为13.98%。此外，我们证明了该新模型适用于其他英伟达架构，如图灵架构。 最后，研究表明，现代英伟达 GPU 中基于软件的依赖管理机制在性能和面积方面优于传统的基于 ...

Core Viewpoint - The "12th Automotive Electronics Innovation Conference and Automotive Chip Industry Ecosystem Development Forum (AEIF 2025)" will be held in Shanghai on May 14-15, focusing on key technological breakthroughs in the automotive sector and inviting industry experts to share insights on trends, technology hotspots, and innovative applications [1]. Event Overview - The conference will feature a summit forum, supply-demand matching sessions, specialized forums, and product exhibitions, expecting over a thousand attendees [1]. - The organizing committee aims to enhance connections between upstream and downstream players in the automotive industry, providing more diverse showcasing opportunities for capable automotive chip companies [1]. Agenda Highlights - The event includes various sessions such as: - VIP and exhibitor registration on May 14 [4]. - A closed-door meeting for automotive electronics experts [5]. - A supply-demand matching meeting for automotive chips [5]. - A high-level forum and specialized discussions on automotive electronic ecosystems, smart connected vehicles, and AI in autonomous driving on May 15 [5][30][34]. Key Speakers and Topics - Notable speakers include: - Chen Dawei, discussing the current state and development of domestic automotive chips [28]. - Dai Weimin, presenting on Chiplet technology and its applications in high-end smart driving chips [29]. - Lu Wancheng, addressing the challenges and opportunities in the automotive electronics industry under the vision of "zero casualties" and "carbon neutrality" [22]. - Topics will cover the integration of domestic chips, advancements in automotive electronics, and the role of AI in driving innovation [12][23][34]. Location and Logistics - The conference will take place at the Shanghai Zhongxing Pullman Hotel, with detailed transportation options provided for attendees [6][8][11].

Core Viewpoint - The semiconductor industry is shifting from a singular focus on process miniaturization to diversified innovation, with advanced packaging and specialty processes becoming key drivers for performance optimization and differentiation [1][2]. Group 1: Specialty Processes and Market Growth - The global specialty process market has surpassed $50 billion, with a compound annual growth rate (CAGR) of 15%, significantly outpacing the average growth rate of the semiconductor industry [1]. - Specialty processes focus on customized and diverse process optimizations, achieving a precise balance of performance, power consumption, and cost, particularly in demanding fields like automotive electronics and IoT [1]. Group 2: TSMC's Leadership in Specialty Processes - TSMC is establishing itself as a global benchmark in specialty processes through a combination of technological breadth and ecosystem depth, expanding its capabilities across various domains including automotive and RF technologies [2][4]. - TSMC's advanced logic technologies, such as N7A, N5A, and N3A, are specifically designed for automotive applications, ensuring high reliability and long lifecycle [4]. Group 3: Innovations in Embedded Non-Volatile Memory (eNVM) - TSMC is addressing the limitations of traditional eFlash memory by advancing RRAM and MRAM technologies, which are expected to replace eFlash in automotive and IoT applications [6][8]. - The introduction of RRAM and MRAM technologies allows for significant improvements in performance, reliability, and power efficiency, with TSMC's RRAM already in mass production at 40, 28, and 22 nm nodes [7][8]. Group 4: Competitive Landscape and Future Trends - Major MCU manufacturers are collaborating with foundries to leverage specialty processes, with companies like Infineon and NXP adopting eNVM technologies to enhance their product offerings [9][16]. - The market for embedded NVM is projected to grow rapidly, with wafer production expected to increase from approximately 3 KWPM in 2023 to about 110 KWPM by 2029, indicating a strong shift towards new storage technologies [26]. Group 5: Diverse Storage Technologies - Various new storage technologies, including eRRAM, eMRAM, and ePCM, are being explored by different manufacturers, each offering unique advantages in terms of speed, power consumption, and integration capabilities [30][32]. - The trend indicates a move towards a multi-storage technology ecosystem rather than a single dominant solution, reshaping the MCU landscape in the post-eFlash era [32].

如果您希望可以时常见面，欢迎标星收藏哦~ 来源：本文 综合自电子时报等 ，谢谢。 据digitimes引述供应链消息透露，NVIDIA遭遇多方危机，甚至因H20降规芯片禁销中国，而不得 不在季报中认列55亿美元损失后，首席执行官黄仁勋不仅于美中双边奔走，希望将冲击降至最 低，期能让美国政府回心转意放宽AI禁令限制。 同时，市场传出，为维持稳定获利表现，NVIDIA近期几乎「全线产品」都调涨官方报价，并放行 合作伙伴同步调升价格。 如华硕RTX 5090高阶显示卡年初上市至今，已拉升至新台币9万元，近日通路价格再大涨至10万 元，涨幅逾1成，同时也调涨H200/B200等GPU模组与伺服器价格。 以此来看，加上中国以外AI芯片需求依旧强劲，CSP大厂仍持续扩大资本支出，NVIDIA月底财报 表现应可在财测内，缴出优异获利成绩。 4月以来美国总统川普（Donald Trump）发动对等关税，以及进一步收紧AI芯片禁令等政策，直 接封杀NVIDIA的H20降规芯片输往中国后，NVIDIA首当其冲成为超大的箭靶。 黄仁勋倾尽全力拆弹，力图将冲击降至最低。 除迅速回应川普的「美国制造回流」大计，宣布4年内与台积电、鸿海 ...

如果您希望可以时常见面，欢迎标星收藏哦~ 来源：本文 编译自 nytimes ，谢谢。 三位知情人士透露，特朗普政府正在考虑一项协议，该协议可能会向阿联酋人工智能公司 G42 运 送数十万块美国设计的人工智能芯片，美国政府过去曾因其与中国的关系而对其进行审查。 正在进行的谈判凸显了美国科技政策在特朗普总统本周访问波斯湾国家前夕的重大转变。谈判也加 剧了特朗普政府内部的紧张局势，一方面，科技和商业头脑的领导人希望在特朗普访问前达成协 议，另一方面，国家安全官员担心这些技术可能被阿联酋滥用。 知情人士表示，特朗普政府已接受与中东官员直接达成人工智能芯片交易，以加强美国在该地区的 关系。由于谈判仍在进行中，这些知情人士要求匿名。这一做法与拜登政府的做法截然不同。拜登 政府曾拒绝类似的人工智能芯片销售，担心这些芯片可能会让与中国关系密切的专制政府在未来几 年在开发最尖端人工智能模型方面超越美国。 在与 G42 和阿联酋官员的谈判中，白宫人工智能主管戴维·萨克斯 (David Sacks) 一直在努力达成 一项协议，允许这家阿联酋公司在有限的监管下使用芯片。一位知情人士表示，部分芯片将提供给 G42 与美国 OpenA ...

如果您希望可以时常见面，欢迎标星收藏哦~ 来源：本文 编译自 blocksandfiles ，谢谢。 目前有两家公司高度专注于 DRAM 和 NAND 的生产——美光科技和SK 海力士。两家公司都在企 业级 SSD 和高带宽内存领域展开激烈竞争，但都通过早期庞大的业务扩张，以复杂而间接的方 式，在两个市场中保持着各自的定位，期间既有失误，也有灵光乍现。 一个是英特尔的恩赐，另一个是被诅咒的。美光与英特尔结盟，推出了命运多舛的傲腾技术，最终 一蹶不振；而SK海力士则收购了陷入困境的英特尔SSD和NAND晶圆厂业务，并迅速进军高容量 SSD市场，该市场迅速腾飞，目前正蓬勃发展。SK海力士还阻止了西部数据与铠侠的合并，并早 早进军高带宽内存（HBM）业务，如今正借助英伟达的GPU内存优势一路飙升。 美光科技 美光科技于2005年通过与英特尔的合资企业进入闪存业务。2010年，该公司以12.7亿美元收购了 闪存芯片制造商Numonyx。随后，该公司于2013年收购了尔必达存储器(Elpida Memory)，从而 拓展了其内存业务，获得了苹果iPhone和iPad的内存供应业务。此外，该公司还于2016年收购了 PC ...

Core Viewpoint - Entegris, Inc. announces the retirement of CEO Bertrand Loy in August 2025, with David Reeder set to succeed him, highlighting a leadership transition aimed at maintaining stability and growth in the semiconductor industry [2][4][5]. Leadership Transition - Bertrand Loy has served as CEO for 13 years and will become Executive Chairman of the Board until mid-2026 to ensure a smooth transition [2][5]. - David Reeder, currently a board member, brings extensive financial and operational leadership experience from companies like Chewy Inc., GlobalFoundries, Texas Instruments, and Broadcom [4][6]. - The board emphasizes the importance of succession planning and believes Reeder is the ideal leader for Entegris' future [6]. Company Performance - Entegris reported Q1 earnings of $0.67 per share, slightly below the market expectation of $0.68, with revenues of $773.2 million, which also fell short of the $791.6 million forecast but showed a year-over-year growth of 0.3% [7]. - The company provided a weak guidance for Q2, expecting earnings between $0.60 and $0.67 per share and revenues between $735 million and $775 million, both below analyst estimates [7]. - CEO Loy noted strong demand for CMP consumables and contamination control solutions but acknowledged increased uncertainty due to new tariff regulations affecting visibility [7]. Future Outlook - Despite short-term challenges, Loy expressed optimism about long-term prospects, citing customer demand for new materials and higher purity to enhance equipment performance [7]. - Entegris aims to leverage its expertise in materials science and purity to drive growth and outperform the market in the coming years [7].

在前道工艺控制中，关键薄膜量测 (Critical Thin Film Thickness)和光学关键尺寸量测（Optical Critical Dimension）具有极高的系统性技术门槛，不断迭代的工艺平台对量测机台性能提出更高 要求，高端量测机台的国产化支撑迫在眉睫。 匠岭科技长期布局半导体前道高端量测领域，通过持续的技术积累和扎实的客户实践， 核心产品 关键薄膜量测机台和光学关键尺寸量测机台，已能够支撑本土晶圆厂稀缺工艺的稳定量产，切实补 齐了本土先进逻辑与高端存储芯片制造上的供应链短板，显著提升了本土半导体产业链在关键量测 环节的自主可控能力。 近日，国产量测设备厂商匠岭科技连续完成B轮与B+轮战略融资，其中 B轮融资由石溪资本 领投，联合投资方为合肥产投国正与俱成资本；B+轮融资由启明创投领投，联合投资方为元 禾璞华、混沌投资、超越创投、临港数科基金、横琴瀚瑞、凯风甬翔，并由老股东冯源资本 多轮次追加投资。 新融资资金将主要用于新产品研发和规模化产能布局。 多重突破， 满足本土稀缺工艺对高端量测的紧迫需求 经过在高技术壁垒领域的长期深耕与持续创新突破，匠岭科技已成功实现多款高端量检测设备的国 产化 ...

如果您希望可以时常见面，欢迎标星收藏哦~ 来源：本文 编译自 omdia ，谢谢。 英伟达 (Nvidia) 攀升至 Omdia 2024 年全球营收排名中位居榜首的芯片公司。与此同时，英飞凌 (Infineon) 和意法半导体 (STMicroelectronics) 均跌出前十。 据 市 场 研 究 公 司 Omdia 称 ， 2024 年 全 球 芯 片 市 场 价 值 将 达 到 6830 亿 美 元 ， 比 2023 年 增 长 25%。 芯片市场的飙升源于对AI相关芯片（包括高带宽内存（HBM）DRAM）的强劲需求。这足以弥补 汽车、消费和工业市场领域芯片销售额的下降。Omdia表示，这些领域在2024年都经历了收入下 降。 | 2023 | 2024 | | 2023 | 2024 | Revenue Percent | | --- | --- | --- | --- | --- | --- | | Rank | Rank | Company Name | Revenue ($) | Revenue($) | Change | | 2 | 1 | NVIDIA | 49,161 | 107,47 ...

Core Insights - The article discusses the significant acquisition of Annapurna Labs by Amazon for approximately $350 million, which has become crucial to Amazon's success, particularly in its AWS division and AI strategy [1][2][4] - Amazon's investment in AI infrastructure is projected to exceed $100 billion this year, highlighting its commitment to developing custom chips to reduce reliance on Nvidia [2][4] - The article emphasizes the strategic foresight of Amazon's leadership in recognizing the importance of chip design for cloud computing and AI [2][5][9] Group 1: Acquisition and Impact - Annapurna Labs was acquired by Amazon in a pivotal moment for AWS, contributing significantly to its profitability and revenue, which surpassed $100 billion last year [2][4] - The acquisition has allowed Amazon to develop custom chips like Trainium and Graviton, enhancing its AI capabilities and operational efficiency [10] Group 2: Strategic Vision and Development - Amazon's leadership, particularly Andy Jassy, recognized the need for in-house chip design to better control supply chains and offer competitive pricing to AWS customers [5][9] - The article highlights the role of James Hamilton, an Amazon executive, in advocating for chip design, which was initially met with skepticism [8][9] Group 3: Market Context and Competition - The current AI boom has led to a multi-trillion dollar "arms race" among cloud computing giants, with Amazon leading in capital expenditure for AI infrastructure [2][4] - The article notes that other tech companies, including Microsoft and Google, are also investing heavily in custom hardware to support their AI initiatives [2]