NVIDIA GDN: Transforming Cloud Gaming and Graphics Rendering

In a groundbreaking move, NVIDIA has unveiled its Graphics Delivery Network (GDN), a revolutionary platform aimed at transforming the landscape of cloud gaming and graphics rendering. According to the NVIDIA Technical Blog, GDN aims to eliminate the remaining barriers in cloud gaming, allowing for seamless one-click play across various devices.

Seamless Gaming Experience

Traditionally, cloud gaming required users to own the game and maintain multiple accounts, but GDN simplifies this by enabling publishers to stream games built with any 3D engine directly to users. Game publishers can choose a suitable virtual machine, upload their content to a developer portal, and let GDN handle the streaming process, making high-quality gaming accessible to a broader audience.

Users can access the content through SDK integration points, such as deep-link URLs or embedded webpages, ensuring a smooth and user-friendly experience. This development follows over a decade of research and innovation in cloud graphics infrastructure by NVIDIA, starting with the Kepler generation of GPUs in 2013 and culminating in the flagship cloud gaming service, GeForce NOW.

High-Performance Streaming Events

GDN's capabilities extend beyond gaming to hosting large-scale, multi-platform events. For instance, it recently streamed Epic Games' 'Metallica: Fuel. Fire. Fury' Fortnite concert to mobile devices, allowing users to join with a single click. This event showcased GDN's ability to deliver a low-latency, near-native experience to users globally.

Another example is ZENOS, a virtual stadium offering immersive VR and 3D spectating experiences for esports. In 2023, ZENOS utilized GDN to stream the League of Legends' Worlds Championship, enabling users to join without any downloads or installations, thus demonstrating GDN's potential for enhancing user engagement in virtual events.

Infrastructure Behind GDN

At the core of GDN are NVIDIA L40 series GPUs, based on the NVIDIA Ada Lovelace architecture. These GPUs are designed for heavy rendering workloads and high-definition streaming, making them ideal for GDN's requirements. The platform offers various instance types, ranging from full to fractional GPUs, ensuring efficient resource sharing and low-latency performance.

GDN leverages NVIDIA virtual GPU (vGPU) technology to partition the GPU frame buffer, dedicating memory to each virtual machine. With more than 30 data centers strategically located worldwide, GDN ensures low-latency performance for users in over 130 countries.

Supported Devices

GDN supports a wide range of client devices, including PCs, laptops, tablets, smartphones, smart TVs, and soon, Apple Vision Pro headsets. It can stream in low-bandwidth environments as low as 5 Mbps/user, making it accessible to a broad audience.

GDN Billing

GDN's billing operates on a deposit system, with costs varying based on instance type, region, and consumption model. Users can choose between on-demand and reserved instances, with the latter guaranteeing access and reducing loading times.

To explore and compare GDN instance types, visit the NVIDIA Graphics Delivery Network page.