Giga Computing, a subsidiary of GIGABYTE and an industry leader in generative AI servers and advanced cooling technologies, today will exhibit at the OCP Global Summit 2024 to share AI solutions that are in demand for all scales of data centers and workloads. The GIGABYTE booth (A26) will showcase a scalable rack-level solution for NVIDIA GPUs and networking, a direct liquid cooling (DLC) server designed for NVIDIA HGX™ H200 systems, and the NVIDIA Spectrum-X™ networking platform.

Like the theme of the OCP Summit “From Ideas to Impact”, GIGABYTE will exhibit the cutting-edge solutions that cloud service providers are deploying today for generative AI applications that impact areas such as drug discovery, language learning, and coding assistance. Besides by providing the synergy of compute resources in a well-designed server, another way GIGABYTE is making a significant impact is in green computing. Cloud computing at scale requires a significant amount of energy consumption, and to be more energy efficient, servers must move to liquid cooling. So, at the GIGABYTE booth is the G4L3 series server that uses DLC loops to cool dual CPUs and NVIDIA HGX™ H200 systems. And for data centers that are not ready yet for liquid cooling, at the booth is the G593-SD1 that has a dedicated chamber for air cooling of the NVIDIA H200 Tensor Core GPU. Both AI solutions offer great benefits that can impact various industries today.

Once data centers go down the route of a GPU cluster seeking greater performance than what a single server can offer, the choice of networking switches plays a crucial role because nodes require high-speed data transfer and low latency to exchange data quickly and efficiently. For this, GIGABYTE is exhibiting the NVIDIA Spectrum™ SN5600 switch, which has 64 ports of 800GbE for enabling cloud-scale infrastructures.

Additionally, visitors at the GIGABYTE booth can engage in discussions about the NVIDIA GB200 NVL72 platform, which can integrate 36x GIGABYTE GPU servers, each equipped with an NVIDIA GB200 Grace Blackwell Superchip connected via NVIDIA NVLink™ technology to two NVIDIA Blackwell GPUs. The NVIDIA Blackwell architecture is also seen in the NVIDIA MGX™ platform with NVIDIA GB200 NVL2. The modularized architecture of NVIDIA MGX™ has been shown by GIGABYTE in the past with the XH23-VG0 and new solutions will support GB200 NVL2 as it is used for fine-tuning training and AI inference.

All these products and more are at the GIGABYTE booth at OCP.
Register for the OCP Global Summit: https://www.opencompute.org/summit/global-summit/registration