News Articles

CUDA 4.1 released

NVIDIA has posted a new release of the CUDA Toolkit. Of note is a significant enhancement to the NVIDIA Performance Primitives (NPP) library, a collection of GPU-accelerated image, video and signal processing functions that deliver 5x to 10x faster performance than comparable CPU-only implementations.

Using NPP, developers can take advantage of over 2000 image processing and signal processing primitives to achieve significant improvements in application performance. Whether you are simply replacing CPU primitives with GPU-accelerated versions or integrating NPP primitives with your existing GPU-accelerated pipeline, NPP delivers high performance while reducing development time. We have already seen speedups in PCL trunk by compiling and using our Kinect Fusion implementation as well as other GPU code with CUDA 4.1.

Other new features in CUDA 4.1 include:

  • 10% performance improvement with new LLVM-based CUDA compiler.
  • Re-designed Visual Profiler to give you step-by-step performance optimizations.

Find out more at

CUDA 4.1

OpenCV team joins PCL development

Due to a generous grant from nVidia, our friends and collaborators from ITSeez, the maintainers of OpenCV, are joining forces with the PCL development team, in order to provide GPGPU optimizations for PCL.

ITSeez has been proudly developing and maintaining OpenCV for the past few years, and are bringing in a large baggage of Computer Vision algorithmic knowledge and optimizations.

Part of PCL 2.x (development will start soon), we plan to completely integrate GPU accelerations for all our 3D algorithms. Together with the OpenCV team, we will try to concentrate and coagulate our efforts to better support the 2D/3D perception community. Our collaboration started after last year's GTC (see slides below), but will intensify from now onwards.

Latest Posts

Latest Comments