Fabric Engine’s Creation Platform Helps Software Developers Tap AMD GPUs
Mocap Processing, Character Solvers, Muscle Simulation Among Potential Applications
Fabric Engine is aiming to make the development of proprietary, high-performance software viable for a broader range of VFX studios. The company's new Creation Platform, shown in the AMD booth at SIGGRAPH in Los Angeles this week, is a framework allowing users to make tools that ensure high performance without reinventing the wheel.
"Graphics software is not using AMD hardware effectively, and most companies don't have C++ experts on staff," said CTO Phil Taylor, noting that the speedy Fabric Core Execution Engine is integrated with the Python programming language, simplifying the development of tools that effectively address AMD graphics hardware. Qt is used for user interface, window management, and widgets. "We allow film and television companies to build custom applications to take advantage of graphics hardware and speed up the process of high-performance software development."
Taylor said the company plans to offer site licenses for VFX studios that want to create their own tools as well as ISV licenses for ISVs looking to build new apps and resell them to users. Fabric Engine says that TDs and developers who are comfortable with Python shouldn't have any trouble adapting to a custom high-level kernel language, KL, that addresses the most critical performance elements (such as multi-threading and memory management) of an application.
The Creation Platform is targeting AMD hardware for now, but Taylor told StudioDaily that Fabric Engine hopes to support Nvidia's CUDA architecture in the future. The platform is currently in beta-test mode at studios including Zoic Studios, Hybride, Halon Entertainment, Ingenuity Engine, and The Third Floor.
Here's a video from Fabric Engine showing some of the potential application sthat can be built using the Creation Platform. You can watch more videos at the platform's official Vimeo page — check out the one that shows KL being dynamically compiled for either CPU or dramatically accelerated GPU rendering without any changes to the code.
For more information: fabricengine.com