GPUltra combines multiple GPUs as if they were a single one. Data is rendered on multiple GPUs on multiple computers across a cluster. Each computer renders a part of the data. The parts are sent and combined into a single animation frame on a dedicated computer. The combined frame is presented to the user.
GPUltra is optimized in many aspects. Utilization of GPU, CPU and network is as parallel as possible. Time of inactivity is thus very low. Moreover, the key parts of the platform are optimized at the algorithm and code level to get the most out of the CPU and GPU. The hardware usage is maximized in result. The architecture of the platform allows it to adapt to requirements and needs. It can be easily customized and extended.
GPUltra provides a web console to get insights into the system, control it and receive notifications when particular events happen. The same functionality is exposed via an API that can be used out of the box. The already existing API has possibility to extend nearly all components of the platform and make GPUltra easy to integrate with.