OpenImageDenoise needs a lot of time on Cycles rendering, and while the GPU renders blistering fast 14 of my 16 Threadripper cores are not denoising. Is there somewhere a parameter to tweak to get all Open Image Denoise to use all cores? To clarify: if usinf low sample number GPU Cycles is ready long before the denoising is finished and I see 16 buckets (32 because of Hyperthreading) denoising, but their accumulated CPU usage is 2-4 cores.
$\begingroup$
$\endgroup$
