AMD contributes ONNX Runtime backend to FFmpeg DNN filter
Summary
An AMD engineer contributed an ONNX Runtime backend to FFmpeg's DNN filter, enabling AI model inference on GPU and NPU platforms for tasks like upscaling and object detection, notably making Ryzen AI NPU useful for FFmpeg.
View Cached Full Text
Cached at: 06/25/26, 01:21 PM
AMD Contributes ONNX Runtime Backend To FFmpeg DNN Filter
Source: https://www.phoronix.com/news/FFmpeg-DNN-ONNX-Runtime
An AMD engineer has contributed to the upstream FFmpeg library an ONNX Runtime back-end for its DNN filter. The FFmpeg Deep Neural Network (DNN) filters allow for running AI models natively inside the video processing pipeline for upscaling, object detection, background segmentation, and more. This ONNX Runntime back-end support is notable in that it expands the GPU and NPU capabilities with FFmpeg.
AMD engineer Steven Xiao authored the ONNX Runtime back-end for FFmpeg’s DNN processing filter. This addition opens up FFmpeg DNN to supporting inferencing on multiple GPU and NPU platforms. Beyond the CPU execution, there is NVIDIA CUDA support, Windows DirectML execution across all the major GPU vendors, and also AMD Ryzen AI NPU support across platforms using the ONNX Runtime VitisAI execution provider. The main takeaway is AMD beginning to make the Ryzen AI NPU useful for FFmpeg.
Those interested can learn more about this ONNX runtime back-end for the FFmpeg DNN filter viathis commit.
Similar Articles
AMD's tiny AI PC points to a more local future for model inference
AMD's Ryzen AI Max platform with 128GB unified memory enables local inference of large models up to 200 billion parameters, aiming to shift AI workloads from cloud to compact personal hardware.
You can now run Qwen 3.8 27B on AMD NPUs via FastFlowLM (at a killer 1 tps decode)
AMD's ROCm/FastFlowLM now supports running Qwen 3.8 27B (and other MoE, vision, audio, and embedding models) on Ryzen AI XDNA2 NPUs with no GPU required, offering up to 256k context in a lightweight 17 MB runtime — though decode speed on the 27B model is noted as a sluggish 1 token/s.
AMD's Lemonade SDK for local AI adds NVIDIA CUDA support
AMD's Lemonade SDK for local AI adds NVIDIA CUDA support in version 10.7, enabling the same local AI server experience on competitor GPUs. The release also introduces lemonade bench for cross-backend LLM benchmarking and broader Vulkan support.
Intel & AMD now collaborating on software optimizations with PyTorch/oneDNN
Intel and AMD jointly announced they will collaborate on x86 CPU software optimizations for PyTorch via oneDNN, with AMD contributing its ZenDNN improvements upstream and joining the UXL Foundation as a steering member.
@AMD: From bring-up to tuning, AMD and @OpenAI engineers are sharing insights to push performance further. Go behind the coll…
AMD and OpenAI engineers collaborate to share insights on performance optimization, featuring a behind-the-scenes look with OpenAI's VP of Compute Strategy, Sachin Katti.