⁠

June 4th, 2024 – GIGABYTE TECHNOLOGY Co. Ltd, a leading manufacturer of motherboards, graphics cards, and hardware solutions, presented the first flagship motherboard in the AI TOP series, the TRX50 AI TOP. With its exceptional versatility and connectivity, the TRX50 AI TOP is specially designed to unleash the ultimate computational power for your local AI training and fine-tuning.

⁠

In the age of local AI, TRX50 AI TOP is built to maximize the configuration for AI model training across CPU, GPU, memory, and SSD. Based on the GIGABYTE recommended GPU/DDR/SSD configuration with the AI TOP Utility and the AI TOP Tutor, the TRX50 AI TOP can support up to 236B-parameter large language models of AI training and fine-tuning while maintaining privacy and security.

⁠

Reinventing AI Training

The TRX50 AI TOP boasts the groundbreaking AI TOP Utility with reinvented workflows, user-friendly interface and experience, and real-time progress monitoring. It supports various top-ranking open-source AI models for beginners to easily start up their own local AI training and fine-tuning without any skills of AI programming. The AI TOP Tutor provides comprehensive consultation for AI TOP solutions, intuitive set-up guidance, and technical support.

⁠

Ultimate Scalability

The TRX50 AI TOP is equipped with four PCIe Gen5 x16 slots and supports four dual-card wide GPUs. This design can provide ample bandwidth, allowing each GPU to operate at full speed.

⁠

The TRX50 AI TOP also features four PCIe 5.0 x4 M.2 slots directly connected to the CPU. This allows data stored in high-speed SSDs to reach the computing center via the shortest path and with the largest bandwidth, enabling local computations to be completed in the shortest possible time.

⁠

The 8-channel ECC R-DIMMs design can provide the highest bandwidth for computation throughput. Additionally, with a maximum capacity of up to 2TB, the TRX50 AI TOP offers ample temporary storage space for model training or machine learning, significantly enhancing the efficiency of large-scale computing.

⁠

Future Connectivity

Built with dual 10GbE LAN and Wi-Fi 7 network, the TRX50 AI TOP delivers blisteringly fast transfer speeds on both wired and wireless networks. It ensures the most stable transmission environment when the system is performing cloud computing or providing external computing services.

⁠

DIY-friendly Innovation

TRX50 AI TOP introduces a number of unique, DIY-friendly advancements in both hardware and firmware. The newly conceived EZ-Debug Zone brings together debug LEDs and control buttons for a streamlined and efficient troubleshooting process. The WIFI EZ-Plug design consolidates Wi-Fi antenna plugs into a single adapter, saving users from the hassle of screwing during setup. Furthermore, existing designs such as EZ-Latch and UC BIOS simplify the process of building your own system.

⁠

Featuring a variety of groundbreaking technologies that can be easily adapted by beginners or experts for most common open-source LLMs, GIGABYTE AI TOP is the all-round solution to win advantages ahead of traditional AI training methods.

⁠

The TRX50 AI TOP will soon be launched in early July. Please visit GIGABYTE at COMPUTEX or GIGABYTE official site for more information: https://www.gigabyte.com/Motherboard/TRX50-AI-TOP#kf