Gigabyte has announced four new servers based on the NVIDIA HGX B200 platform. This expansion of the server portfolio offers greater thermal design flexibility and compatibility with the most recent processors, including the new CPUS Intel Xeon 6 optimized for AI, which offers customers more options to adapt their systems to real workloads.
The new Gigabyte servers are divided into those who use liquid cooling and air cooling. They are based on the NVIDIA HGX B200 platform, solutions that promise to boost data centers towards a new era of accelerated and generative computing, since with Blackwell chips they can offer real -time inference 15 times faster in large models of billion parameters.
The Taiwanese manufacturer distributes two servers of the G4L3 series with liquid cooling system. The series is designed for a high computing density with dedicated cameras for CPU and GPU. Designed with advanced direct liquid cooling (DLC), Gigabyte 4U servers with liquid cooling offer great thermal efficiency and focus on high performance high performance training and inference. Compatible with the HGX B200 platform, they maximize the use of the GPU while reducing energy and cooling costs to create a platform prepared for the future.
The other two models are from the G894 series air -air -air. The G894-AD1 is fully compatible with the Intel Xeon processors of the 6900 series, while the G894-SD1 is compatible with the Intel Xeon processors of the 6700 and 6500 series.
Servidial Compassion for GIGABYTE
Gigapod is a solution of the Rack scale supercomputing specifically designed to address the enormous computer demands of generative training and large language models with billions of parameters. Integrate high performance GPUS, high -speed networks and scalable architecture, to offer performance and flexibility. With the new Gigabyte Nvidia HGX B200 platforms, Gigapod presents an updated design that optimizes energy performance and efficiency and offers advanced cooling options to adapt to the changing workloads of AI.