We are excited to announce that CloudConvert now offers GPU-based video encoding for H264, HEVC, and AV1 formats. This new feature brings a significant performance boost, especially for HEVC and AV1 encoding.
GPU video encoding uses the power of a graphics processing unit (GPU) instead of a central processing unit (CPU) to handle the task of video compression. The main advantage of GPU encoding is speed. Since GPUs are highly parallel processors, they can handle many operations simultaneously, leading to faster encoding times compared to CPUs.
However, GPU encoding has a few trade-offs. While it's faster, it can sometimes result in slightly lower quality at the same file size when compared to CPU encoding. Depending on the use-case, though, the speed advantage may outweigh any minor quality differences, especially for larger files or longer videos.
Our internal tests showed significant improvements when using GPU encoding over CPU encoding:
These performance boosts are particularly important for HEVC and AV1, which are newer, more efficient formats but are typically slow to encode on a CPU. With GPU encoding, you can convert these formats at a fraction of the time it usually takes.
The new GPU encoding feature on CloudConvert uses the latest generation NVIDIA Ada Lovelace GPUs, ensuring the best performance and efficiency for your video conversions.
GPU video encoding is an opt-in feature. If you're using our web interface, you can now switch between CPU and GPU encoding in the advanced options section.
For API users, you can enable GPU encoding by setting the engine parameter to ffmpeg_nvidia. Also, the available parameters for codec, target quality and presets do slightly differ from the ones for CPU encoding. You can show the available options by opening the Job Builder and setting the engine to ffmpeg_nvidia.
Given the power and speed of GPUs, they are expensive and sometimes difficult to source. As a result, GPU encoding on CloudConvert comes at a slightly higher base cost. The pricing for GPU encoding is 5 base credits + 1 credit per minute per conversion. In contrast, CPU encoding remains priced at 1 base credit + 1 credit per minute.
That said, GPU encoding is significantly faster, meaning you'll use fewer credits overall due to the shorter conversion times.
We're looking forward to hearing from our customers about this new feature. Try it out and let us know how it’s working for you. Your feedback will help us continue to improve and refine CloudConvert.