I am sure I read that people can do this. Is this a 2.8 feature as I can only render gpu or cpu in 2.79, not both.
Cheers
And after that just make sure to put the tile size to 16x16 or 32x32, you can find some benshmark on the forum i think, and it only works for f12 and not in the viewport
Don’t bother right now if you are on an AMD card.The low tile sizes makes it even slower. nVidia is the only one that can use this 100%.
The tile size thing is hindering also Nvidia rendering. There should be option to set tile size for gpu and cpu separately. Then if used together, the bigger is first used, and if the second is smaller, then one big tile is divided into smaller ones. That way cpu-threads could work inside one big tile, acting as one compute unit, such as gpu is.
Thanks for the info on how to do this. I have a 1080 Ti and a 4 core i7 CPU so will give it a test on my current project later on. I am looking at ways to speed up rendering without going to an external render farm. I currently do GPU rendering in cycles and just let my machine churn out renders over night for 3 days so this may speed that up a tad.