CUDA Toolkit Interest...

A few hours ago, I installed nVIDIA's CUDA Toolkit 10.2 (SDK, IDE tools/extensions, etc.). However, I installed it so I could start playing around with CUDA-accelerated video processing and encoding. The issue is, nVIDIA's CUDA Video Encoding support was discontinued after CUDA Toolkit version 6.0, according to nVIDIA's documentation. But I need it for legacy hardware where NVENC either sucks to the point that it's not worth using, or is non-existent. I want to use a compute card (Tesla K10) as a dedicated video encoding device. The issue is, Kepler's NVENC encoder is trash when compared to everything on the market (1st gen), and makes NVCUVENC look amazing in comparison. NVCUVENCis the same tech I used to make a GTX 550 Ti capable of accelerated video encoding a few years ago, and it did just fine with realitime 720p60fps while playing modern games. So, imagine what an entire Tesla K10 with NVCUVENC can do for video encoding, since the encoding doesn't have to be done in realtime...


Here are my sources thus far:

• https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/TB-Quadro_VideoCaptureStreaming_v01.pdf
• https://docs.nvidia.com/cuda/cuda-samples/index.html#new-features-in-cuda-toolkit-6-5
• https://developer.nvidia.com/video-processing-software-solutions-nvidia-developers

If anyone can help with this (and not try to advertise NVENC to me instead), it would be greatly appreciated. I'll be primarily working on Windows 10 for now, due to lack of a GPU-accelerated Linux environment. However, I fully intend to bring it over to Linux if anything ever comes from it - assuming the idea doesn't get shot down first...

Some Things Die Hard...

https://linustechtips.com/main/profile/511347-tophatproductions115/?status=254848&type=status