System Information
Archlinux, Sandy Bridge I5.
Blender Version
Broken: 2.74
Short description of error
Baking the same texture twice leads to different results.
Exact steps for others to reproduce the error
Can confirm the issue, but not sure it is expected to be deterministic? Sergey, maybe you know? To me, there is no real reason to make it so here, and it would probably be harder to do too, deterministic reproducible randomness in multi-threaded context is rather hairy afaik…
Also, note that this is not really related to baking imho, this is valid for any render. This is just highly noticeable here because of very low res.
No, the normal render produces deterministic results, for every test I did (including this example). So I think this is baking specific.
Ack, yes, you are right, regular renders seems to produce same result always (at least with same amount of threads - changing that changes result)… So guess there is something wrong in handling of randomness in baking.