监制：柯成韵 中国人民银行发布公告称，为保持银行体系流动性充裕，中国人民银行今天（25日） 以固定数量、利率招标、多重价位 中标方式开展6000亿元中期借贷便利MLF操作，期限为1年期。 据了解，由于本月有1000亿元MLF到期，此次操作实现 净投放5000亿元。 你会关注 特朗普：已设定最后期限！ 全线大涨！金价，又狂飙！ 央视财经评论丨滥用出口管制，美方做法损人不利己 你以为没用，实则超厉害的3个养生小妙招→ 想要膝盖"长寿"，少做这两个动作→ 来源： 央视财经（ID：cctvyscj） 主编：于小曼 编辑：张娅芳 转发，关注！ ...

如果您希望可以时常见面，欢迎标星收藏哦~ 来源：内容 编译自 cadence ，谢谢。 JEDEC 近期发布的 HBM4 规范对 AI 训练硬件开发者来说是个好消息。HBM4 是快速发展的高带 宽内存 (HBM) DRAM 标准的最新规范，据 JEDEC 称，HBM4 可提供 2TB/s 的内存性能和高达 64GB（32Gb 16 位高）的更高密度。JEDEC 的新闻稿指出："HBM4 带来的进步对于需要高效处 理大型数据集和复杂计算的应用至关重要，包括生成式人工智能 (AI)、高性能计算、高端显卡和 服务器。" 大型语言模型 (LLM) 数据集呈指数级增长，而当前的 CPU 和 GPU 性能通常受到可用内存带宽的 限制。由于存在这种"内存壁垒"，HBM 凭借其卓越的带宽、容量和内存效率，已成为生成式 AI 训练的首选内存。 HBM4 基于 HBM3（和 HBM2E）标准构建，后者广泛应用于数据中心的 AI 训练硬件。HBM4 的内存带宽比 HBM3 提升了 2 倍。带宽的提升是通过将频率提升至 8Gb/s（HBM3 为 6.4Gb/s） 并将数据位数翻倍至 2048 位（HBM3 为 1024 位）来实现的 ...

在竞争激烈且行业集中度较高的背景下，诺瓦星云(301589)坚持以"还原真实视界，让世界更美好"为使 命，致力于成为全球领先的视频综合性解决方案提供商，逐步实现由产品研发型企业向创新服务型企业 的转变，以先进的技术与优异的服务为客户发展赋能。 4月17日晚间，诺瓦星云一口气公布了2024年年报、2025年一季报以及关于"质量回报双提升"行动方案 的进展公告。整体来看，尽管面临着市场竞争加剧以及成本上升的不利因素，公司2024年公司财务表现 相对稳健，且多项指标取得优异成绩。 公司拟向全体股东每10股派发现金红利6.5元（含税），预计总派发金额为5942万元，分红率10%。公 司表示，未来将继续践行"质量回报双提升"行动方案，实现高质量、可持续发展，在保持可持续健康发 展的同时，积极回报投资者，为稳市场、稳信心和切实维护广大投资者合法权益贡献力量。 不惧竞争加剧，2024年营收创历史新高 财务数据显示，公司2024年营业收入达到32.8亿元，同比增长7.38%；归母净利润为5.93亿元，同比下 降2.31%；扣非归母净利润为5.76亿元，同比下降2.09%。 由于公司已提前布局MLED领域相关技术，推出了核心检 ...

Core Insights - Nova Star Cloud (诺瓦星云) reported a total revenue of 3.28 billion yuan for 2024, representing a year-on-year growth of 7.38% [1] - The company achieved a net profit attributable to shareholders of 593 million yuan, with a basic earnings per share of 6.57 yuan [1] - The company plans to distribute a cash dividend of 6.50 yuan per 10 shares, totaling 59.42 million yuan [1] Financial Performance - The domestic market contributed 2.65 billion yuan in revenue, a growth of 2.85% year-on-year, while the overseas market generated 626 million yuan, growing by 32.03% [2] - Research and development (R&D) investment reached 540 million yuan, an increase of 22.24%, with an R&D intensity of 16.47% [2] Business Overview - Nova Star Cloud focuses on algorithm-driven solutions for LED display and ultra-high-definition video applications, providing a full range of services from display control to video processing [1] - The company serves over 4,000 clients globally, including industry leaders like Leyard, Absen, and BOE [1][2] - The company has made significant advancements in the MLED sector, launching core detection equipment and integrated circuits, leading to rapid revenue growth in this area [2] Intellectual Property - As of the end of the reporting period, the company holds 1,177 domestic patents (including 656 invention patents), 27 foreign patents (including 22 invention patents), and 238 software copyrights [3] - The company aims to continue high-intensity R&D investments to support advancements in MLED and ultra-high-definition video technologies [3]

STI-Bench团队 投稿 量子位 | 公众号 QbitAI 多模态大语言模型（MLLM）在具身智能和自动驾驶"端到端"方案中的应用日益增多，但它们真的准备好理解复杂的物理世界了吗？ 上海交通大学联合中国地质大学、南洋理工大学、智源研究院以及斯坦福大学的研究团队推出首个多模态大模型（MLLM）时空智能评测基准 STI-Bench（Spatial-Temporal Intelligence Benchmark），向当前最先进的多模态大语言模型发起了关于精确空间时间理解的严峻挑战。 结果显示，即便是Gemini-2.5-Pro、GPT-4o、Claude-3.7-Sonnet、Qwen 2.5 VL等当前最强的多模态大模型，在需要定量分析真实世界空 间关系和动态变化的任务上，表现并不尽人意。 从语义理解到时空智能 MLLM在视觉语言理解上成就斐然，并被寄望于成为具身智能和自动驾驶的"端到端"解决方案。但这要求模型超越传统的语义理解，具备精准 的时空智能。 试想AI应用场景中的需求： 自动驾驶： 需知晓与前车的精确距离（米）、行人过马路的速度（米/秒）、安全过弯的车速限制等。 机器人操作： 需判断目标物体的尺寸位 ...

Omeros Corporation OMER on Thursday established the Omeros Oncology Clinical Steering Committee to advance Omeros’ OncotoX biologics program focused on acute myeloid leukemia (AML).AML is the most fatal form of leukemia. It accounts for approximately 80% of acute leukemias in adults and one-third of all cancers affecting the blood/bone marrow.Omeros’ OncotoX program for AML consists of proprietary targeted, engineered molecules (about half the size of an antibody) that deliver a toxic payload within the can ...

Company Overview - QuickLogic Corporation is a fabless semiconductor company specializing in embedded FPGA (eFPGA) Hard IP, discrete FPGAs, and endpoint AI solutions [3] - The company combines cutting-edge technology with open-source tools to deliver highly customizable, low-power solutions for various markets including industrial, aerospace, consumer, and computing [3] Product Offerings - QuickLogic's adaptable eFPGA technology allows developers to fine-tune and scale their designs, providing an optimal balance of power, performance, and area (PPA) [1] - The Australis™ eFPGA IP Generator enables QuickLogic to rapidly create custom eFPGA IP cores tailored to unique customer requirements [1] - Recent additions to the supported process portfolio include Intel® 18A, GlobalFoundries® (GF) 12LP and 22FDX®, TSMC® N12e™, and UMC 22nm [2] Industry Position - QuickLogic has over three decades of industry experience, making it a trusted partner in Aerospace and Defense [2] - The company delivers eFPGA IP, purpose-built FPGA devices, and FPGA design tools that meet Size, Weight, and Power (SWaP) constraints while ensuring security, reliability, and longevity in extreme environments [2] Event Participation - QuickLogic will exhibit at the HEART 2025 Conference in Monterey, California, showcasing its eFPGA technology [1]

Keysight Technologies, Inc. (KEYS) recently announced that it has collaborated with Coherent Corp., a leading provider of advanced illumination solutions for 3D sensing applications, to enable tuning, validation and 200G characterization of the latter's new vertical-cavity surface-emitting laser (VCSEL) technology. By enabling higher data rates, the state-of-the-art multimode technology promises to facilitate the deployment of Artificial Intelligence/Machine Learning (AI/ML) services while reducing both pow ...

DALLAS, March 31, 2025 (GLOBE NEWSWIRE) -- Applied Digital Corporation (Nasdaq: APLD) ("Applied Digital" or the "Company"), a designer, builder, and operator of next-generation digital infrastructure designed for High-Performance Computing (HPC) applications, will host a conference call on Monday, April 14, 2025, at 5:00 p.m. Eastern Time to discuss its operations and financial results for the fiscal third quarter ended February 28, 2025. A press release detailing these results will be issued after the mark ...

Core Insights - The Edge AI Foundation has rebranded from the TinyML Foundation and released the "2025 Edge AI Technology Report," highlighting the maturity and real-world applications of TinyML [1][3]. Group 1: Edge AI Technology Drivers - The report discusses advancements in hardware and software that support Edge AI deployment, focusing on innovations in dedicated processors and ultra-low power devices [3]. - Edge AI is transforming operational models across various industries by enabling real-time analysis and decision-making capabilities [3]. Group 2: Industry Applications of Edge AI - In the automotive sector, Edge AI enhances safety and response times, with examples like Waymo and NIO utilizing real-time data processing for improved performance [7][8]. - Manufacturing benefits from Edge AI through predictive maintenance, quality control, and process optimization, with reported reductions in maintenance costs by 30% and downtime by 45% [9][12]. - In healthcare, localized AI accelerates diagnostics and improves patient outcomes by analyzing medical data directly on devices [14]. - Retail operations are optimized through real-time behavior analysis and AI-driven systems, reducing checkout times by 30% [16]. - Logistics is enhanced by integrating Edge AI with IoT sensors, allowing for immediate analysis of data and optimization of supply chain operations [18]. - Smart agriculture utilizes Edge AI for precision farming, reducing water usage by 25% and pesticide use by 30% [21]. Group 3: Edge AI Ecosystem and Collaboration - The Edge AI ecosystem relies on collaboration among hardware vendors, software developers, cloud providers, and industry stakeholders to avoid fragmentation [24]. - A three-layer architecture is recognized for Edge AI, distributing workloads across edge devices, edge servers, and cloud platforms [24][25]. - Cross-industry partnerships are increasing, with companies like Intel and Qualcomm collaborating to enhance Edge AI deployment [26][27]. Group 4: Emerging Trends in Edge AI - Five emerging trends are reshaping Edge AI, including federated learning, quantum neural networks, and neuromorphic computing [30]. - Federated learning is expected to enhance model adaptability and collaboration across industries, with a projected market value of nearly $300 million by 2030 [31]. - Quantum computing is set to redefine Edge AI capabilities, enabling faster decision-making and real-time processing [34][36]. - AI-driven AR/VR applications are evolving with Edge AI, allowing for real-time responses and improved energy efficiency [39]. - Neuromorphic computing is gaining traction for its energy efficiency and ability to handle complex tasks without cloud connectivity [41].