Video Core Next is AMD's brand for its dedicated video encoding and decoding hardware core. It is a family of hardware accelerator designs for encoding and decoding video, and is built into AMD's GPUs and APUs since AMD Raven Ridge, released January 2018.
Video Core Next is AMD's successor to both the Unified Video Decoder and Video Coding Engine designs, [1] which are hardware accelerators for video decoding and encoding, respectively. It can be used to decode, encode and transcode ("sync") video streams, for example, a DVD or Blu-ray Disc to a format appropriate to, for example, a smartphone. Unlike video encoding on a CPU or a general-purpose GPU, Video Core Next is a dedicated hardware core on the processor die. This application-specific integrated circuit (ASIC) allows for more power-efficient video processing. [2] [3]
All versions of VCN support: MPEG-2 Decode, MPEG-4 Decode, H.264/MPEG-4 AVC Encode/Decode, HEVC (H.265) Encode/Decode, and VP9 Decode. 10-bit color depth in the P010 format is supported. VCN 1.0 supports up to 4K resolution. VCN 2.0 and beyond supports up to 8K. [4] Support for H.264 and H.265 Encode methods differ among generations (see below). VC-1 Decode is supported since VCN 3.0, excepting 3.0.33. [4]
VCN 2.0 is implemented with Navi products and the Renoir APU. The feature set remains the same as VCN 1.0. [4]
VCN 3.0 is implemented with Navi 2 products. [5] VCN 3.0 implements H.264 B-frames, which was present in Video Coding Engine 2.0 but taken out with VCE 3.0. [6]
VCN 4.0 adds AV1 encode. [7] H.264 quality is higher with VCN 4.0 (as part of RDNA 3) compared to previous generations, but still lags behind Intel and Nvidia hardware codecs. [8]
There is no support for encoding or decoding in YUV422 and YUV444 in H.264 and H.265.
VCN Generation | GPU code name | JPEG | H.262 (MPEG-2) | VC-1/WMV 9 | H.264 (MPEG-4 AVC) | H.265 (HEVC) | VP9 | AV1 | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Decode | Decode | Decode | Decode | Encode | Decode | Encode | Decode | Decode | Encode | |||||
B-frame [a] | Pre-analysis [b] | Resolution, color depth | Chroma | Resolution, color depth | ||||||||||
VCN 1.0 | Raven, Picasso | ![]() | ![]() | ![]() | ![]() | ![]() | ? | 4K @ 10b | ![]() | 4K @ 10b | ![]() | ![]() | ![]() | |
VCN 2.0 | Navi 1x | ![]() | 8K @ 10b | 8K @ 10b | ||||||||||
VCN 2.2 | Renoir, Lucienne, Cezanne, Barcelo | |||||||||||||
VCN 2.5 | Arcturus | |||||||||||||
VCN 2.6 | Aldebaran [12] | |||||||||||||
VCN 3.0 [13] | Navi 21, Navi 22, Navi 23 | ![]() | ![]() | |||||||||||
VCN 3.0.33 | Navi 24 | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ||||
VCN 3.1.0 | Van Gogh [14] | ![]() | ![]() | ![]() | ![]() | 8K @ 10b | 8K @ 10b | ![]() | ||||||
VCN 3.1.1 | Rembrandt, [15] Mendocino | ![]() | ||||||||||||
VCN 3.1.2 [16] | Raphael, Dragon Range | |||||||||||||
VCN 4.0 [17] [18] | Navi 3x, Phoenix | ![]() | ||||||||||||
VCN 5.0 | Navi 4x |
Early versions of AMD VCN had lower overall quality (VMAF) compared to offerings from Intel and Nvidia. B-frame narrowed the gap, but did not eliminate it; [8] further improvements, including pre-analysis, resulted in nearly equivalent performance to competitors at the release of VCN 4.0. [9] [19]
Despite a lack of B-frame support, H.265 provides better quality (VMAF) and near-identical speed for the same bitrate compared to H.264 on VCN 2.0, 3.0, and 4.0. [20]
Qualcomm
The AMF PA feature is supported by Radeon RX 5000 Series or newer GPUs as well as Ryzen 2000 U/H series or newer APUs.