Activity News

Publish Date : 2024/05/30

Leadtek Unveils COMPUTEX 2024 Highlights:WinFast High-End Mini AI Workstation Makes Its Debut

Under the theme of "Reshaping GPU-driven AI applications with Scalability and Agility" Leadtek will showcase a range of innovative products at COMPUTEX 2024 to meet the diverse AI needs of various scales. These products include the first-ever high-end WinFast Mini AI workstation, NVIDIA-Certified systems equipped with NVIDIA RTX Ada Lovelace architecture professional GPUs and PCIe Gen5 support, and NVIDIA HGX H100 servers for large-scale AI. All new products are fully equipped with liquid-cooling technology, providing unparalleled performance and reliability for large language models, generative AI, deep learning applications, and digital twins. In terms of software and AI applications, a more secure digital twin solution and an AIDMS-assisted smart fishery AI landing case study will also be on display from June 4 to 7 on the 4th Floor of Nangang Exhibition Hall 1.

Mini AI Workstations: Empowering AI for Everyone with Agile, Compact Liquid Cooling Innovation

Ditch the sprawling server rooms and noisy workstations. Now, anyone can develop digital transformation or generative AI applications right in their office. Unveiling a new era of compact AI workstations, the WinFast Mini AI Workstation is powered by Intel's latest 5th Gen Scalable Processors, delivering exceptional performance in a compact chassis. Leading the industry, it supports two high-end NVIDIA GPUs and provides a more stable high-wattage power supply, ready for next-generation NVIDIA GPUs. Leveraging liquid-cooling technology, the WinFast Mini AI Workstation effectively lowers GPU temperatures and offers a quieter operating environment, making it an ideal solution for those seeking both computational power and spatial flexibility. It's perfect for small- and medium-sized businesses without large server rooms looking to embrace digital transformation, schools and departments conducting academic research, or individual users venturing into the AI realm. For those with more AI training needs, opt for the WinFast WS2050L Workstation, which supports four liquid-cooled GPUs and delivers high performance and quiet operation without the need for external liquid-cooling lines. This makes it ideal for offices with AI training requirements.

Liquid-Cooled AI Servers for Large LLMs

Leadtek offers a wide range of servers in various form factors, including 8U, 5U, 4U, 2U, and 1U. The 8U NVIDIA HGX AI server, showcased at COMPUTEX, is specifically designed for large AI deep learning workloads, including LLMs, generative AI, and foundation models. It is equipped with two 5th generation Intel scalable processors and eight NVIDIA H100 SXM5 Tensor Core GPUs. It incorporates industry-leading NVIDIA NVLink 4.0 technology and the NVIDIA NVLink Switch interconnect, significantly enhancing GPU-to-GPU communication speed with breakthrough high-speed transmission technology. This provides exceptional acceleration capabilities, overcoming AI performance bottlenecks.

The 8U NVIDIA HGX AI server employs innovative direct-to-chip (D2C) liquid-cooling technology, which directly transfers coolant to each GPU chip, effectively reducing GPU temperature and maintaining system stability.

Additionally, to cater to various scales of AI training workloads, Leadtek has also launched AI workstations and servers with configurations ranging from 1 to 10 GPUs. Notably, understanding the importance of time in AI development, Leadtek provides Leadtek RTX AI Software Pack, which offers the highest compatibility, along with the GPU Docker Management System (GDMS) for GPU resource allocation and management. This ensures that customers can immediately start AI application development and GPU resource management without the need for additional time to configure related NVIDIA libraries, frameworks, drivers, and Docker environments.

Spark the Success of AI Applications, Bringing Real Benefits to Society

With the assistance of AI, industries have seen significant improvements in quality assurance, error reduction, and cost control, leading many companies to eagerly adopt AI. However, most enterprises, lacking the necessary infrastructure and application capabilities, have been hesitant to fully embrace AI transformation. This has dramatically changed with the introduction of the WinFast AI Workstation and AIDMS.

During COMPUTEX, the LumiGood development team, Superluminal Pirates, will share the challenges and solutions of AI transformation at the Leadtek booth. Traditionally, counting fish and shrimp fry relied on the human eye, but these fry are often less than 1 cm long, with some being extremely tiny and nearly transparent, posing a significant challenge. Now, with the introduction of Leadtek's AIDMS (AI Development Management System), the aquaculture industry can shorten AI model development time and quickly establish an AI fry recognition system. Users can simply take photos with their smartphones, completing a task that originally required three people and three days in just two hours, with an accuracy rate of over 97%. This breakthrough not only provides convenience to fishermen but also establishes a basis for transactional integrity, aiding in the transformative development of the entire aquaculture industry.

Leadtek's AIDMS (AI Development Management System) is a foundational MLOps management platform designed for team collaboration. Its significance lies in offering a no-code, easy-to-learn, and easy-to-use platform, enabling professionals from various backgrounds to participate in AI application development. For example, professionals familiar with aquaculture can collaborate with AI engineers through AIDMS to jointly develop AI recognition systems, greatly reducing communication errors between different professions, while quickly enjoying the convenience and benefits brought by AI technology.

Building a Safe and Reliable Industrial Digital Twin: The Powerful Combination of NVIDIA Omniverse Enterprise and HP Anyware

NVIDIA Omniverse Enterprise enables world-class digital twin performance, significantly advancing industrial digitalization and simulation. Plus, the integration of NVIDIA Omniverse Enterprise with HP Anyware not only offers high-level cybersecurity measures under a zero-trust architecture but also achieves photorealistic 3D design collaboration and seamless interactivity in simulations. The new generation of Trusted Zero Clients and Anyware Trust Center establishes optimal remote connection security and PCoIP Ultra performance from the endpoints, helping IT personnel integrate zero-trust architecture into their digital workspace environments. This enables engineers, architects, and designers to achieve the full frame rate rendering capability required for creating complex designs and visuals.

Join us at COMPUTEX 2024 for a four-day exhibition at Leadtek's booth #M1001a on the 4th floor of Nangang Exhibition Hall 1. Experience the exciting demonstrations of innovative technology and explore the industrial advancements brought by our collaborations on AI workstations, AI servers, AIDMS, and security.