Image Processing SDK for Tegra K1 GPU
Tegra K1 is NVIDIA’s first mobile processor which includes high-performance Kepler GPU architecture. Tegra K1 offers very high CPU and GPU performance to mobile computing. PC class imaging applications that require high performance, low energy consumption and large amounts of memory can now be developed for mobile devices with Tegra K1.
We have ported our high performance image processing SDK to Tegra and now we can offer high performance imaging solutions for Nvidia’s quad-core Tegra K1 GPU.
Tegra K1 benchmarks for image processing on GPU
We have done measurements of kernel times for the most important components of GPU image processing SDK from Fastvideo. We utilized images with Full HD resolution (1920×1080) and got the following averaged results:
Results show that on Tegra K1 GPU one can do realtime and high quality image processing for Full HD cameras at 30 fps, including full image processing pipeline, starting from camera raw data to compressed MJPEG video stream.
The fastest realtime solution for camera application could be achieved in the case if we do direct JPEG compression of RAW data. We can get 100 fps performance for Full HD resolution (quality 90%) which corresponds to data rate 200 MPix/s for 8-bit data. For viewing compressed data one can do JPEG decompression and debayering on desktop GPU in off-line which is straightforward with our image and video processing SDK.