Puget Systems' AI Server: AMD Meets NVIDIA

Puget Systems is set to demonstrate its cutting-edge on-premises Generative AI and Machine Learning solution at SIGGRAPH 2024. The custom server, designed for GPU-intensive workflows, combines the power of AMD EPYC processors and NVIDIA GPUs to deliver high-quality imagery with smooth frame rates.

Behind the Scenes

• The server supports up to eight NVIDIA RTX Ada generation GPUs, NVIDIA L40S GPUs, or NVIDIA H100 Tensor Core GPUs, along with NVIDIA Networking Platforms.

• AMD EPYC processors, designed for demanding server workloads, offer up to 128 cores, support for 1.5TB DDR5 ECC RAM, and 128 PCIe Gen 5 lanes.

• The demonstration will showcase a StreamDiffusion pipeline, running locally on the Puget Systems server, highlighting the capabilities of the NVIDIA RTX 6000 Ada Generation GPU.

• Attendees can interact with a GUI to control StreamDiffusion parameters and select various input methods, such as webcams, visual noise, Adobe Photoshop, and a mouse pointer for physical fluid simulation.

Last May, Puget Systems was at the NAB Show to showcase their powerful laptops and custom AI training machines.

Final Take

Puget Systems' AI Training and Inference Server represents a significant advancement in on-premises AI computing. By combining the strengths of AMD and NVIDIA hardware, the company offers a powerful solution for businesses and organizations looking to harness the potential of generative AI and machine learning without relying on cloud-based services.

As AI continues to transform various industries, Puget Systems' innovative approach to hardware design positions them as a key player in the evolving landscape of AI infrastructure.