top of page

RidgeRun Support to GStreamer CUDA Library

  • Writer: ridgerun
    ridgerun
  • 1 day ago
  • 2 min read


We’re excited to announce the release of the GstCUDA Wrapper, a new extension to the RidgeRun GstCUDA framework.

This wrapper is not a standalone product — it’s provided as an extension to RidgeRun’s GstCuda that simplifies migration and compatibility between RidgeRun workflows and the evolving GStreamer ecosystem.


Why Does This Matters?

Over the years, RidgeRun’s GstCUDA framework has enabled developers use RidgeRun CUDA-Accelerated GStreamer elements to their projects and to rapidly build custom CUDA-accelerated GStreamer elements for high-performance multimedia and vision pipelines, some examples are: RidgeRun stitcher, Birds Eye View, CUDA ISP. In general the GstCUDA wrapper will allow seamless integration between RidgeRun GstCUDA and GStreamer CUDA Library.


With recent improvements in the open-source GStreamer CUDA API (starting with GStreamer >= 1.24), native support for NVIDIA Jetson platforms and CUDA memory integration has become more standardized. The RidgeRun’s GstCUDA Wrapper ensures:

  • Seamless porting of existing RidgeRuns GstCUDA-based code to GStreamer Cuda Library.

  • Alignment with newer GStreamer CUDA APIs supported by NVIDIA and the community specially on newer Jetpacks which comes by default with GStreamer 1.24..


We introduced the GstCUDA Wrapper, which bridges RidgeRun’s GstCUDA interfaces to the open-source GStreamer CUDA API.

Specifically:

  • On Jetpack < 7.0, the wrapper uses RidgeRun GstCUDA as the backend, basically, RidgeRun GstCUDA can be used without the wrapper on Jetpack < 7.0.

  • On Jetpack >= 7.x, the wrapper uses the GStreamer CUDA library as the backend if NVMM support is required, but RidgeRun GstCUDA can be used regularly if NVMM is not required without compromising performance.

The wrapper can be seen as a tool to switch the RidgeRun GstCUDA GStreamer element backend from the RidgeRun GstCUDA backend or the GStreamer CUDA Framework Backend:  




What You Can Do with the GstCUDA Wrapper

  • Maintain compatibility of existing GstCUDA-based elements with newer GStreamer platforms

  • Abstract away boilerplate between RidgeRun GstCUDA and open-source GStreamer CUDA Library APIs

  • Support both legacy Jetpack environments and current/future GStreamer CUDA frameworks

  • Preserve your existing development workflows and APIs while gaining broader compatibility


For any questions or to discuss how RidgeRun can help with your specific use case, feel free to reach us at Contact us!

 
 
bottom of page