Reposted from AMD China

On December 21, 2025, at the MoDa Community Developer Center in Hangzhou, AMD, together with DataWhale and MoDa Community, hosted the "Igniting Intelligence, Building the Developer Ecosystem Together" event. AI developers and enthusiasts from across China gathered in Hangzhou to witness the official launch of the jointly built ROCm Developer Zone and the inauguration of the ROCm Developer Learning Center.

AI has been one of the most transformative technologies of the past 50 years, and developers are the key driving force behind this transformation.

Dr. Lisa Su, Chairwoman and CEO of AMD, sent a signed message:
"Open source is the key to accelerating AI innovation. We are committed to building the most powerful, open, and developer-friendly AI ecosystem in the industry. An open, developer-first ecosystem brings the entire industry together under open standards, enabling everyone to participate in AI innovation."

In his opening video speech, Mr. Pan Xiaoming, Senior Vice President and President of Greater China at AMD, emphasized that openness and open source are core values that AMD consistently upholds. AMD is committed to supporting and promoting a thriving open-source ecosystem. He expressed his hope that AMD’s innovative technologies and developers’ creativity could come together through open source in leading communities such as MoDa and DataWhale, unlocking potential and continuously driving AI development and adoption.

Unleashing AI Potential Requires Ubiquitous AI Engines

During the keynote, Mr. Ji Zhaohui, AMD Vice President of Marketing for Greater China, introduced AMD’s leading and comprehensive product capabilities, as well as ecosystem partner innovations in edge applications using Mini AI workstations. AMD empowers AI developers through its ROCm open-source software stack, which was honored with the "Leading Technology Award" at this year’s World Internet Conference. This recognition not only affirms the ROCm open-source software stack itself but also validates the spirit of open-source collaboration and shared innovation. AMD, together with DataWhale and MoDa Community, aims to provide AI developers with an open and innovative learning platform, as well as space for communication and growth.

Simultaneously, representatives from AMD, DataWhale, MoDa Community, and other open-source contributor organizations jointly launched the ROCm Developer Zone.

Mr. Wang Junyang, Senior Algorithm Expert from Alibaba Tongyi Lab, shared insights and practices on large model development at the edge.

As China’s largest AI open-source learning community, DataWhale helps AI learners connect with knowledge, people, and real-world scenarios, promoting AI talent development. Ms. Fan Jingjing, Founder of DataWhale, stated that through collaboration with AMD and MoDa Community, they aim to contribute innovation to the AI developer ecosystem based on openness and open-source principles.

Finally, representatives from AMD, MoDa Community, Mianbi Intelligence, and startup companies discussed opportunities, challenges, and the future of the ecosystem for developers in the context of open-source, open-access, and democratized computing power.

"Unity is strength; collective intelligence achieves success." The ROCm Developer Zone co-created by AMD, MoDa Community, and DataWhale aims to gather AI enthusiasts, learners, developers, and industry pioneers under the banner of open source, jointly contributing to AI innovation and the integration of industry applications. Together, they strive toward a new AI-driven world.

GMKtec EVO-X2 Overview

The GMKtec EVO-X2, powered by the AMD Ryzen AI Max+ 395 processor, features the advanced Zen 5 architecture with 16 cores and 32 threads, reaching a maximum frequency of 5.1GHz and 80MB of cache. It integrates a Radeon 8060S GPU with 40 RDNA 3.5 compute units, significantly enhancing graphics performance. In addition, the processor includes a 50TOPS XDNA 2 NPU, delivering a combined CPU+GPU+NPU performance of 126 TOPS, making it the world’s first Windows 11 AI+ PC APU capable of supporting 70B large language models (LLMs).

This heterogeneous computing power ensures exceptional AI performance, with the EVO-X2 outperforming the NVIDIA GeForce RTX 4090 by 2.2× in LM Studio AI benchmarks, while consuming less power, providing users with an efficient and energy-saving AI computing experience.

Built-in Local Large Models – Ready-to-Use, Privacy-Friendly

The EVO-X2 comes pre-installed with local large models, allowing instant use without internet connectivity. Its local knowledge base supports personalized content storage and retrieval, keeping private data securely on-device. Its compact form factor fits multiple scenarios, enabling offline Q&A, document processing, and delivering a high-efficiency, private AI experience.

• Includes LLaMA 109B Large Model (128GB version)

• Supports Qwen3-235B and DeepSeek 70B models (128GB version)

Equipped with the AMD Ryzen AI Max+ 395 processor, featuring a 50TOPS XDNA 2 NPU and up to 128GB unified memory (with up to 96GB assignable as VRAM), the EVO-X2 can smoothly run full-scale Qwen3-235B models locally, and successfully deploy and execute 70B models such as DeepSeek-V3 and DeepSeek-R1-Distill-Llama-70B through custom memory scheduling and deep quantization optimization.

Endorsements

• Signed by Dr. Lisa Su, AMD CEO

• Recommended by Ji Zhaohui, AMD Vice President of Marketing, Greater China

 

About GMKtec

Shenzhen GMKtec Technology Co., Ltd. specializes in desktop AI supercomputing centers, mini-PCs, and other intelligent terminals, focusing on R&D, production, and sales. Recognized as a national high-tech and specialized innovative enterprise, GMKtec’s brand has earned the title of “Famous Shenzhen Brand”, alongside Huawei and DJI, solidifying its tech innovation identity in Shenzhen.

In 2025, GMKtec received the American Design Award (Platinum) and French Design Award (Gold) – a first for Chinese desktop AI supercomputers. The company has established a self-developed, self-produced, and globally distributed brand, leading the global mini-PC market, and pioneering the concept of a desktop AI supercomputing center. Its team comprises industry experts with 20 years of experience in electronics, design, production, and brand marketing.

GMKtec collaborates strategically with industry leaders such as Intel, Microsoft, and AMD to drive technological innovation and industrial upgrades, offering high-quality products and services. The Longhua industrial base, operational since 2019, covers R&D, operations, labs, and warehousing logistics for intelligent terminals. GMKtec products are sold in over 70 countries, with distribution and service networks in 35 countries, offering desktop AI supercomputers, mini-PCs, portable GPUs, and home & business NAS systems.