NVIDIA GeForce GRID processors
NVIDIA's new GeForce® GRID processors are specifically designed to deliver the best platform for cloud gaming services. GeForce GRID processors solve three main issues that prevented earlier GPUs from delivering a great gaming experience from cloud servers: GPU virtualization, low-latency remote display, and power efficiency. Together, these technologies create a significant shift in the performance, efficiency, and end-user experience for cloud-rendered games.
Low-latency remote display
Traditional remoting technologies have long latencies associated with the network, rendering, capturing, and encoding in the server. The new GeForce GRID architecture addresses these delays with built-in, low-latency remote display capabilities. This performance advantage greatly benefits cloud gaming services by enabling super low latency delivery of a game. The GeForce GRID architecture also includes a high-performance H.264 encoder engine that's capable of encoding up to four simultaneous 720p 30fps streams with superior quality. This provides a giant leap forward in cloud server efficiency by offloading the CPU from encoding functions and allowing the encode function to scale with the number of GPUs in a server. Plus, the on-chip encoder can be fed in a single pass from the low-latency capture that creates the most efficient and lowest-latency architecture.
Optimizing performance per watt demanded an entirely new approach to power management. Every hardware unit in GeForce GRID has been designed and scrubbed to provide outstanding results, including the new streaming multiprocessor called "VGX". The result was an innovative solution that delivers revolutionary performance per watt for the cloud gaming services.
GeForce User-Selectable Machine
The GeForce USM loads a complete server-edition NVIDIA GeForce GTX experience with industry-leading game compatibility and performance onto a real or virtual GPU. This machine image includes support for the latest game APIs, including DX-9, DX-10, DX-11, and OpenGL. It's also optimized for cloud gaming, with low-latency capture APIs for individual applications or the entire desktop, single-pass, low-latency encoding, and network-encoder optimizations.
Find other user-selectable machines