Part of the Technology photoes in this website are created by rawpixel.com - www.freepik.com

H3 Platform × NVIDIA Webinar: Breaking the GNN Memory Wall with NVIDIA® SCADA™

H3 Platform and NVIDIA Technical Webinar: Breaking the GNN Memory Wall with NVIDIA® SCADA™. US Session December 9, 2026, 4:00 PM PST. EMEA Session December 10, 2026, 5:00 PM CET / 8:00 AM PST.
  • US Session

    December 9, 2026

    4:00 PM PST

  • EMEA Session

    December 10, 2026

    5:00 PM CET / 8:00 AM PST

Language: English | Platform: Google Meet

45-minute pre-recorded presentation + 15-minute live Q&A

Both sessions feature the same presentation and Q&A experts.

Accelerate Graph AI with a New Memory Hierarchy Built for Scale

Graph Neural Networks (GNNs) are deep learning models that learn from the relationships between data, such as users and products or accounts and transactions. They power applications like recommendation systems and fraud detection.

As graphs grow to billions of nodes and edges, the graph structure and feature data can quickly exceed the memory of a single GPU. Join H3 Platform and NVIDIA to learn how NVMe storage can extend GPU memory for large-scale GNN workloads.

What you'll learn

  • Why large GNN workloads hit a memory wall
  • How NVIDIA® SCADA™ enables GPU-initiated access to NVMe storage
  • Storage Lender, a key function of NVIDIA SCADA: what it is and how it works, presented by NVIDIA
  • Falcon 6048 configuration and benchmark results
  • A GNN demonstration, followed by live Q&A with H3 Platform and NVIDIA

Who should attend

AI/ML engineers, data scientists, and infrastructure architects working with large-scale graph or recommendation workloads.

Speakers

  • Brian Pan

    CEO

    H3 Platform

  • Jeff Chang

    Senior Software Engineer

    H3 Platform

  • Rax Liang

    Senior Software Engineer

    H3 Platform

  • CJ Newburn

    Distinguished Engineer

    NVIDIA

  • Joe Eaton

    Distinguished Engineer

    NVIDIA

About Falcon 6048

H3 Platform Falcon 6048 is a 3U PCIe 6.0 GPU-accelerated memory and storage server with up to 48 E1.S NVMe SSD bays. It is built to serve as a second-tier memory layer beyond GPU HBM for large AI datasets.

Learn more about Falcon 6048 →

Register

Your joining link will be sent in your confirmation email.

Your Information

Session

Which session would you like to attend?*

GNN Proof of Concept

Interest in a potential GNN PoC*