Compiling, libraries, modules, coding guidelines and porting
1 post • Page 1 of 1
I can imagine things like rotations, resizing, and sculpting being very straightforward to program as a SIMD operation in Blender. Each vertex has it's own independent data in the form of its location, and when you select a certain number of points and press "R", they're all given the same command to rotate a certain distance around a central point. Same goes for sculpting- just bring all the vertexes within this circle "up" when I click the mouse and drag. In each of these examples, separate threads could handle the movement of each point. Problem is, when I DO sculpt or rotate or whatever, the task manager only shows one core maxed out, while the others are sitting idle. Is there a feature that I don't know about that can be turned on, or am I just doomed to single-threaded performance until a future version of Blender add this?