This is an open source project, so obviously no one here can answer that definitively.
The project plans are here: Dev:Doc/Projects - BlenderWiki
It looks to me like perhaps the 2.75 build, which has listed as one of the targets "
- Cycles kernel split for OpenCL patch
And the schedule for 2.75 looks like 2nd half of May. That assumes that the patch is stable and passes review, I'm sure.
Thanks, and what does this patch do?
When I run OpenCL now, there is one tile on blender render screen for one device(Cpu core or GPU), so at all on my laptop with AMD A6-5200+Radeon HD8400+Radeon HD8570M(A) i see 6 tiles. Does this patch make every GPU unit like 1 compute core in blener, and makes 2 tiles for HD8400 and 5 tiles for HD8570 during rendering in blrender? And can it to increase speed of rendering?
1: check out the open source discussion.
2: I'll check with the senior engineer working on this. He happens to sit, literally, right next to me. But I never see him, he's usually working remote. Probably gets a lot more done.
1 of 1 people found this helpful
There is a bug in the Blender build that prevents mixed device rendering from working at all.
Even if that's fixed, "mixed mode" type rendering is not suggested. CPUs and GPUs have different floating point characteristics. (Blender uses the -cl-fast-relaxed-math flag, meaning that the GPU is not IEEE compliant, but faster.) As a result one may see artifacts at tile boundaries. Even worse, when you render animations, one could see similar artifacts that are frame-to-frame (temporal), which could be more annoying. Your mileage may vary. And I hope this helps.