There are two GPU-instances types:
g2.2xlarge - 1 x NVIDIA GRID GPU (Kepler GK104) == NVENC
cg1.4xlarge - 2 x NVIDIA Tesla M2050 GPUs == CUDA
In the Wowza Transcoder AddOn Performance Benchmark only two non GPU-instances are tested. Is that because the GPU-instances have not yet been tried or simply don’t work with Wowza?
Also, does the following quote mean that only NVENC and not CUDA is supported on Linux?
Wowza Transcoder supports Intel Quick Sync and NVIDIA NVENC accelerated encoding on Windows and Linux and NVIDIA CUDA accelerated encoding on Windows.
Hi,
It means that they haven’t been benchmarked. There’s no reason why they shouldn’t work though.
You are correct about the quote. NVENC is supported on Linux, but CUDA is not. Which would mean that
the cg1.4xlarge would need to be run under Windows only.
Daren
Hello,
GPU acceleration is supported on a couple AMI types mentioned above. The g2 instance type is the most commonly used AMI with GPU acceleration. Please see the following article for more information on how to use Wowza Streaming Engine with this AMI:
How to configure NVIDIA NVENC accelerated encoding on Amazon EC2 (G2, NVENC)
Best regards,
Andrew
None of the AMI’s use the gpu instances still. Why is this? Not viable, not supportable, just plain doesn’t work?