Page Comparison

NVIDIA GPU transcoding uses NVIDIA GPUs for hardware accelerated video transcoding and will operate on any hardware configuration supported by the Zixi Broadcaster, including on-premise hardware and cloud-based hardware and on X86-64 and ARM64 processors.

Hardware Requirements

NVIDIA GPUs are available on consumer gaming graphics PCIe cards, professional CAD acceleration PCIe cards, custom motherboards, and on cloud services such as AWS, Azure, and Google Cloud. 

Below are

the recommended modern NVIDIA GPUs

, although older and new GPUs should work as well:

• NVIDIA Quadro P2000 and & RTX 4000
• AWS g3.4xlarge and p3.2xlarge
• Azure NC6, ND6, and NV6
Kontron MSP8041 blade computerPCIe Cards
• AWS VM - g4dn (T4)
• GCP VM – G2 (L4), T4
• Azure VM – NCas (T4)
• Tata Cloud VM – A16, T4

NVIDIA Pascal architectures or later are required for HEVC/H.265 decoding.

NVIDIA Tesla architectures cannot decode HEVC/H.265, but can decode MPEG2 and H.264

.  As of this writing, NVIDIA does not support decode or encode of 4:2:2 content.  More details on specific features supported by different architectures can be found in the encode/decode support matrix from NVIDIA at:

https://developer.nvidia.com/video-encode-decode-gpu-support-matrix

Key information to look for when selecting a GPU card and architecture is the “Max # of concurrent sessions”.  This value determines how many

encodes can be

performed in parallel

, which is relevant for ABR streaming

where you will typically need between three to nine bitrate profiles.  The best choice for transcoding for ABR streaming is a GPU card with unrestricted sessions.

Operating System Requirements

CentOS 7.2 through CentOS 7.6 64-bit with glibc 2.14 or greater has been validated by Zixi and is recommended.The operating system requirements are listed in the Zixi Installation Guide.