Performance & Benchmark

From CUVI Wiki
Revision as of 16:47, 5 July 2018 by Jawad (Talk | contribs) (Benchmark)

Jump to: navigation, search

If one thing CUVI gives you, it's performance boost over competitive libraries and solutions. Using GPGPU as the underlying hardware, Imaging and Vision modules get maximum benefit due to their inherent parallel algorithms. In addition to cost cutting on CPU-based clusters, CUVI gives up to 15x speedup over Intel IPP.

CUVI Speedup.jpg

Applications using CUVI are generally ten times faster than their CPU counterpart. CUVI framework also gives the ease to scale the application on more than one GPU making it as fast as you want.

CUVI Bench.jpg


The following benchmark is performed on NVIDIA GTX 1080 via Nsight for Performance tool on Windows 10 (64-bit) and CUDA toolkit version 9.1. Timing figure represents frames per second (fps) based on only the processing time on the single GPU. The benchmarks are performed on 8-bit images except if mentioned otherwise. The benchmarks for 16-bit demosaicDFPD on 1080p, 4k and 8k image are 1550fps, 412fps and 94fps.

1080p Full HD 4k Ultra HD 8k Ultra HD
Auto Color 7088.83 fps 1850.02 fps 461.36 fps
Demosaic (DFPD) 1707.94 fps 412.72 fps 101.86 fps
Demosaic (Linear) 4258.88 fps 1025.64 fps 234.66 fps
Low Light Enhancement 2143.02 fps 525.16 fps 145.52 fps
Resize (2x - Nearest Neighbor) 4169.51 fps 1048.44 fps 260.164 fps
Resize (2x - Linear) 2494.80 fps 613.65 fps 151.53 fps
Resize (2x - Cubic) 1778.42 fps 456.68 fps 108.44 fps
Resize (0.5x - Nearest Neighbor) 47,265.68 fps 12,396.48 fps 3145.28 fps
Resize (0.5x - Linear) 26,365.05 fps 6793.71 fps 1703.32 fps
Resize (0.5x - Cubic) 11,232.92 fps 3143.94 fps 799.00 fps