I’ve been working on a rather heavily displaced scene and I was struck by the differences in render time I was getting on our M1 systems vs the PC with GPU.
Ordinarily, we do most of our design on the Intel Macs, then transfer to PCs for rendering. This has held true for the M1’s as without a GPU, the PC’s generally still spanked the M1’s for render time… though the M1’s were talking back from time to time.
For this micropoly displacement scene, the M1’s blazed through where the PC’s couldn’t. This seems to underscore how heavily the PC’s are dependent on the GPU’s. For this scene, the PC’s were not only slower to build the render, but the GPU failed entirely with an ‘out of memory’ error after 37 minutes. Switching Blender to CPU compute allowed the PC to complete the render, but much slower that the Apple Silicon.
Not trying to start a flame-war… we use both platforms, and if I could get my hands on a 3090… I’d happily build a new PC, but for these types of heavy scenes, the MAC is now the only viable design and render channel in our studio.
Note: The image is tightly cropped from the original 2080x2080 resolution full image. Just wanted to show the type of details we’re dealing with.