Hybrid Parallel Volume Rendering with Distributed Graphics Hardware
Yi Cao
Abstract
Yi Cao
Abstract
For lack of graphics module,general shared memory Multiprocessors can not be used to visualize the three-dimensional scientific computational datasets generated on themselves using hardware-accelerated volume rendering techniques.An algorithm of the hybrid parallel rendering based on multiprocessors and distributed graphics workstations is presented.The datasets are processed on both local multiprocessors and distributed graphics workstations.The following image composition is performed by multiprocessors to utilize efficiently communicating capabilities.Through load balancing optimization,parallel rendering pipeline can overlap rendering,compositing and display process.Experimental results show that the interactive rendering abilities can be achieved when handling large datasets resided in multiprocessors with 1 024×1 024 image resolution.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
For lack of graphics module,general shared memory Multiprocessors can not be used to visualize the three-dimensional scientific computational datasets generated on themselves using hardware-accelerated volume rendering techniques.An algorithm of the hybrid parallel rendering based on multiprocessors and distributed graphics workstations is presented.The datasets are processed on both local multiprocessors and distributed graphics workstations.The following image composition is performed by multiprocessors to utilize efficiently communicating capabilities.Through load balancing optimization,parallel rendering pipeline can overlap rendering,compositing and display process.Experimental results show that the interactive rendering abilities can be achieved when handling large datasets resided in multiprocessors with 1 024×1 024 image resolution.
Key concepts: Computer science, Parallel rendering, Rendering (computer graphics), Texture memory, Software rendering, Tiled rendering, Workstation, Real-time rendering