Core Viewpoint - The article emphasizes the resilience and potential of China's semiconductor industry, highlighting the importance of foundational research and the development of artificial intelligence technologies. Group 1: Chip Development - Ren Zhengfei stated that there is no need to worry about chip development, asserting that many Chinese companies are making significant progress in this area, including Huawei [3] - He acknowledged that Huawei is currently one generation behind the U.S. in single-chip technology but emphasized the use of mathematical methods to compensate for this gap [3] - The potential for growth in mid-to-low-end chips and compound semiconductors in China was highlighted, along with the assertion that software development will not be hindered by external pressures [3] Group 2: Fundamental Research - Ren Zhengfei called for greater societal support for fundamental theoretical research, stressing that without it, China cannot catch up with the U.S. [4] - He noted that Huawei invests approximately 180 billion yuan annually in R&D, with around 60 billion yuan dedicated to fundamental research, which is not subject to performance evaluation [4] - The importance of establishing a strong foundation in technology to ensure long-term growth and stability in the industry was emphasized [4] Group 3: Artificial Intelligence - Ren Zhengfei described artificial intelligence as potentially the last major technological revolution for humanity, with a development timeline spanning decades [6] - He pointed out China's advantages in AI, including a large youth population and robust infrastructure for power and communication networks [6] - The article mentioned the potential for a unified market in China, which could facilitate breakthroughs in technology despite external pressures [6] Group 4: Product Development - The article noted the launch of the Huawei Mate60 series featuring "Chinese chips" and the upcoming release of the HarmonyOS NEXT, which aims for full self-reliance in operating system technology [5] - Huawei's development of various hardware and software products, including Kunpeng processors and Ascend processors, was highlighted as part of its strategy to build a comprehensive computing ecosystem [5]

Core Viewpoint - The dialogue between Ren Zhengfei and People's Daily highlights Huawei's resilience in the face of external pressures, emphasizing a focus on progress rather than obstacles [1]. Group 1: Discussion on Chips - Ren Zhengfei downplays concerns regarding chip technology, stating that many Chinese companies are making significant strides in this area, and Huawei is one of them [3]. - He acknowledges that Huawei is currently one generation behind the U.S. in single-chip technology but believes that through innovative methods, practical results can be achieved [3]. - Ren emphasizes the potential for Chinese companies in mid-to-low-end chips and highlights the importance of compound semiconductors [3]. Group 2: Importance of Basic Research - Ren stresses the necessity of foundational theoretical research for technological breakthroughs, indicating that without it, China will lag behind the U.S. [5]. - He mentions that Huawei invests approximately 180 billion yuan annually in R&D, with around 60 billion yuan dedicated to basic theoretical research [5]. - The company has made significant advancements in "root technology" and operating system innovation, contributing to the foundation of China's information industry [5]. Group 3: Digital Infrastructure and AI - Huawei has launched a series of products, including Kunpeng processors and Ascend processors, to build a computing ecosystem for general and AI applications [6]. - Ren views artificial intelligence as potentially the last major technological revolution, with China having advantages such as a robust power supply and advanced communication networks [8]. - He believes that the future of the country is promising, with increasing openness leading to progress and the potential for a unified market [8].

在人工智能快速演进的浪潮下，大模型正加速重构各行业的技术底座，而 推理性能优化 正成为应对算力挑战、内存瓶颈与通信压力的关键突破口。 当前，大模型推理性能优化主要围绕 模型优化、推理加速与工程优化 在即将于 5 月 23 日 -24 日举办的 AICon 全球人工智能开发与应用大会·上海站 中，我们特别策划了《大模型推理性能优化策略》专题论坛，由阿里云公共云大模型技术服务负责人 王德山 担任专题出品人，现已确认多位业内实践者参与分享。以下为嘉宾阵容及即将带来的精彩议题简介～ 向乾彪 – 腾讯推理架构师 向乾彪在 GPU 推理加速拥有丰富经验。他的技术专长覆盖高性能异构计算及深度性能优化，并在实 践中不断突破前沿技术瓶颈。目前，向乾彪带领团队负责混元大语言模型的推理加速框架 【AngelHCF】 三大方向展开：通过模型量化、剪枝与蒸馏等手段降低计算复杂度、提升推理效率，例如 DeepSeek-R1-Distill-Qwen-32B 采用蒸馏策略，在保持高性能的同时显著压缩资源开销；依托 SGLang、vLLM 等高效推理引擎提升生成速度与系统吞吐能力；同时结合实际业务场景，合理规划 并发策略、优化 GPU 配置 ...