CN
JiWei General-Purpose Chip Industry Application Summit explores new opportunities in the era of massive data
Chinese article by 陈炳欣
English Editor 张未名
06-07 18:08

By Greg Gao

(JW Insights) Jun 6 -- JW Insights held the JiWei General-Purpose Chip Industry Application Summit on June 2, bringing together renowned experts from leading semiconductor firms to share insights and analysis and explore the potential of general-purpose chips in the era of massive data.

The demand for high-performance computing power is growing rapidly as data volumes increase and algorithms evolve. High-end general-purpose chips, serving as the foundation for computational power, play a decisive role in shaping the future direction of the digital industry. Whether it’s automotive, HPC, cloud computing, 5G, or other applications, high-end general-purpose chips are crucial to enabling new applications. It is the background for the presentations at the summit.

Speakers at the summit came from leading Chinese semiconductor firms, including Black Sesame Technologies(黑芝麻智能), GigaDevice(兆易创新), Iluvatar CoreX(天数智芯), Flagchip(旗芯微), Kiwimoore(奇异摩尔), ePowerComm(鸿智电通) and TRONLY-ERAY(镭昱光电).

Yang Yuxin, chief marketing officer of Black Sesame Technologies, talked on “High-performance computing chips empower the rapid development of the smart car industry.” Currently, the trend of evolution from traditional cars to smart electric cars is unstoppable. Yang Yuxin said that under this change, the global automotive industry chain is being restructured, including new emerging players and supply relationships, which undoubtedly provide development opportunities for chip suppliers.

The rapid growth of intelligent driving is inseparable from the evolution of automotive electronic and electrical architecture.

Yang Yuxin believes the auto architecture shift is evolving from distributed to domain-controlled architecture and will even enter the central computing platform architecture. The change requires higher computing power and higher integration of core chips to support it, Yang said.

Xue Ting, marketing director of GigaDevice’s memory business, introduced GigaDevice called on “Establishing China’s Niche Memory Chip Department Store”, and shared his company's success in the storage business sector. As a domestic chip pioneer, GigaDevice started with storage products. After more than ten years of development, it has become a leading semiconductor solution provider integrating memory, microcontrollers, sensors, and analog chip products.

He added that the demand for data storage is increasing, as applications multiply for 5G, the Internet of Things, big data and AI. Smart homes, wearable devices, new energy, and automobiles are also the hot markets that are rapidly growing. 

Iluvatar CoreX's product line chief, Zou Wei explained how general-purpose GPUs can help the development of large-scale language models with the theme of “Practical Implementation of Domestic GPUs Empowering Large-scale Language Models.”

Iluvatar CoreX’s general large-scale language model training and promotion solution supports mainstream large-language models and Finetune framework. The effect is consistent with mainstream GPU and supports CUDA-based custom API developed by the company's customers, Zou added.

In his speech, Wang Chunlong, marketing director of Flagchip, emphasized the increasing importance of driving safety with more intelligent new energy vehicles on the road. The development necessitates the support of functional safety MCUs.

Zhu Jundong, vice president of Kiwimoore, discussed the Chiplet technology with the theme “Chiplet-Based Ultra-Large-Scale Heterogeneous Computing Platform Driven by Large Language Models.” He emphasized the need for more versatile and powerful capabilities in ultra-large-scale computing platforms. 

With the exponential growth in computational demands, the performance improvement of individual transistor areas has slowed down. The vast number of computing scenarios has decreased efficiency in general-purpose processors. Fragmented applications and rapid algorithm iterations have extended the chip development cycle. All these factors require computing platforms to possess stronger general-purpose computing capabilities, Zhu added.

Qian Yu, senior analyst at JW Insights, delivered a presentation titled “Data Center Chip Demands in the Era of Big Data and High Computing Power,” providing an in-depth analysis of the development of core chips in data centers.

linkedin twitter facebook line
Copy succeeded
link