浙江芯片独角兽，拿下"中国边缘AI芯片第一股"！ 成立于2019年的爱芯元智，刚刚在港交所敲钟。开盘价28.2港元，开盘市值约166亿港元。 公司创始人是仇肖莘——一位在半导体行业内，有超20年技术研发经验的清华系海归女博士。 从AI视觉芯片起家至今，爱芯元智已完成终端计算、智能汽车、边缘AI推理等多场景布局，位居全球中高端视觉端AI芯片出货量第一。 其智能汽车SoC累计出货51.8万颗，也让公司成为第二大国产智能驾驶SoC供应商，吉利、零跑、福特等等都是爱芯的客户。 而在IPO之前，爱芯元智也早早就被资本市场注意到，韦豪创芯、启明创投、旷沄管理、美团、腾讯、联想之星等等明星投资者都是股东，还有十多位基 石投资者站队，已共同认购14亿港元。 "中国边缘AI芯片第一股"上市 爱芯元智（以下简称"爱芯"），刚刚在港交所上市，成为"中国边缘AI芯片第一股"。 本次上市全球发售1.05亿股，发行价为每股28.2港元，募资总额约29.6亿港元。 敲钟前，豪威集团、雅戈尔、德赛西威等十多家基石投资者，共同认购了约1.85亿美元(约14.4亿港元)股份，占总发行股份的48.76%。 爱芯元智今日开盘价28.2港元，和发售价 ...

Solana Ecosystem Development - Solana Foundation 发布 ConnectorKit，一个新的 SDK，用于向 Solana 应用程序添加钱包连接和管理功能 [1] - ConnectorKit 兼容 web3js 和 kit [1]

6/ Tongo provides everything as a single SDK: https://t.co/iDcgielGnJ> Simple API + high-level methods for every operation> Type-safe, full TypeScript support> Auto ZK proof generation> Encrypted balance management under the hood> Wallet/provider-ready for Starknet ...

You can build confidential payment flows today on Starknet, with @tongoxyzHidden amounts, no trusted setups, efficient ZK proof verification, and flexible compliance, all in one SDKTongo (@tongoxyz):🎉 Confidential Payments are live on Starknet 🎉The Tongo SDK is ready to be integrated TODAYOur frens at @Starknet & @StarkWareLtd built Kage with it to prove how comfy confidentiality actually isWhat are you waiting for anon?Come build the confidential future with us https://t.co/cEzXjv9rL7 ...

记者12月9日从工业和信息化部获悉，依据相关法律法规，工业和信息化部对APP、SDK（第三方 软件开发工具包）违法违规收集使用个人信息等问题开展治理。近期，经组织第三方检测机构进行抽 查，共发现24款APP及SDK存在侵害用户权益行为。 据悉，上述APP及SDK应按有关规定进行整改，整改落实不到位的，工业和信息化部将依法依规组 织开展相关处置工作。（记者周圆） ...

最近， 英伟达 CEO 黄仁勋 在 一次 内部会议 上 罕见发火 了 。 因为他发现 ： 在这家市值 4 万亿美元、 全球 最热 的 AI 公司 中 ， 居然 还 有 管理者 要求团队少用 AI —— " 我听说英伟达内部居然有人让员工 少用 AI？你疯了吗？ " "任何能被 AI 自动化的任务，都必须自动化" 据 外媒 Business Insider 报道，这场全员会发生在英伟达发布历史最高季度营收（570 亿美元，同比暴涨 62%）后的第二天。 当时 ， 公司上下沉浸在"大丰收"氛围中， 但 听闻 有人 要求 团队 少用 AI 后 ， 黄仁勋 直接当场发飙 ， 并 强调 ： "我希望所有能被 AI 自动化 的任务， 全部由 AI 自动化。 " 他 指出 ， 就算 AI 暂时做不好，也要继续用，"用到它 好 用 为止"，并亲手把它变得更好。 或许是 考虑到 员工 会紧张 "用太多 AI 是否 会被替代" ， 黄仁勋 还补充道 ， AI 是员工在公司持续扩张中保持竞争力的基础能力，而不是威胁 ： "我保证，你们永远会有事情做。" 与微软、Meta、谷歌、亚马逊等推动员工使用内部 AI 工具、甚至将其纳入绩效 ...

Core Insights - NVIDIA's CEO Jensen Huang expressed frustration over internal management urging teams to use less AI, emphasizing that all tasks that can be automated by AI should be automated [1][3] - Following a record quarterly revenue of $57 billion, Huang reinforced the importance of AI in maintaining employee competitiveness and job security, stating that AI should be integrated into daily workflows [3][6] Group 1: Company Performance and AI Integration - NVIDIA reported a record quarterly revenue of $57 billion, a 62% year-over-year increase [3] - The company has expanded its workforce from approximately 29,600 to around 36,000 in the past year, with an additional 10,000 positions still needed to meet demand [3][4] - Huang highlighted that the software engineering team is utilizing AI coding assistants like Cursor, indicating a shift towards an "AI-native" engineering culture [4] Group 2: CEO's Vision and Employee Assurance - Huang envisions NVIDIA growing to a workforce of 50,000 with 100 million AI assistants, demonstrating a commitment to AI integration [4] - He reassured employees that AI is not a threat to their jobs but a tool to enhance productivity, stating that the real competition comes from those who can leverage AI effectively [6] - Huang's message aims to alleviate employee anxiety regarding job security in the face of AI advancements, emphasizing that job roles will evolve rather than disappear [6]

RT Umbra (@UmbraPrivacy)An SDK for builders. A wallet for everyone.Starting now: cohort applications for dev teams, and waitlist signups for our new wallet. Privacy on Solana, any way you need it.More info: https://t.co/RaSexZItED https://t.co/T12FrT9hJt ...

Development & Deployment - AMD Embedded Development Framework (EDF) supports SDK-based development flow for AMD Versal adaptive SoCs [1] - EDF SDK can be installed from a shell script and potentially speeds up build tasks compared to on-target building [1] - The SDK allows cross-compilation of applications for Arm cores using an x86 host PC [4] - On-target development allows packages to be installed from the package feed, offering flexibility compared to the fixed content of the SDK [1] Workflow & Tools - The process involves installing the AMD EDF SDK, developing and compiling applications/kernels on an x86 host PC, and deploying them to a Versal AI Edge Series Gen 2 (VEK385) evaluation board [1] - QEMU flow is explored for application deployment, offering system emulation capabilities [1][2] - The "scp" command is used to copy executable files from the SDK to the evaluation board via the PS Ethernet [1] Kernel Module Development - Kernel modules can be developed and compiled on a host PC and deployed to a Versal evaluation board [1][4] - Preparing the kernel involves setting the kernel source directory environment variable and the local version variable [1] - The generated kernel module file can be copied into the kernel module directory of the active kernel on the evaluation board [2]

Software Development on AMD Versal SoCs - AMD Embedded Development Framework (EDF) SDK facilitates software application development and deployment targeting AMD Versal adaptive SoCs [1] - The SDK supports cross-development on an x86 host, enabling application and kernel development without direct board access [1] - The SDK can accelerate build tasks compared to on-target building, but its content is fixed, requiring recreation from EDF for updates [1] - On-target development allows package installation from the package feed, offering flexibility [1] - The process involves installing the SDK, cross-compiling applications for Arm architecture, and deploying them to the Versal AI Edge Series Gen 2 (VEK385) evaluation board [4] Kernel Module Development - Kernel modules can be developed and compiled on a host PC and deployed to a Versal evaluation board [1] - Preparing the kernel involves setting the kernel source directory environment variable and creating a unique identifier for the kernel in use [1] - The generated kernel module file can be copied into the kernel module directory of the active kernel on the evaluation board [2] Emulation with QEMU - The SDK provides QEMU scripts for full system emulation of the VEK385 board [2] - Booting QEMU requires the same boot artifacts and disk image used to boot on hardware [2] - Applications can be deployed to the emulated Versal system using QEMU, mirroring on-hardware execution [4]