Render stopping/stalling

[rajm]

A terrain I'm currently working on seems to have the habit of sitting there for ages soaking up cpu but the picture doesn't update. Sometimes it appears to get going again and at others after an 8 hour wait it hasn't done anything.
The stdout - in one example render reports as follows
Rendering final pass... 10:43:30s, 50% of final pass, 1801124 micro-triangles
....
Rendering final pass... 14:34:00s, 50% of final pass, 1801124 micro-triangles
Rendering final pass... 14:34:30s, 52% of final pass, 1949925 micro-triangles (a bit of an advance here)
....
Rendering final pass... 19:13:00s, 52% of final pass, 1949925 micro-triangles

It seems to stop around where the low level clouds start though it's not always exactly the same point.

I'm attaching the tgd and a hung render image I don't think there's much complicated in the terrain? Render 1 is the one that hangs

[Dune]

I would at least suggest to take out water shader 03.

[rajm]

Well spotted - thank you, that appears to be an improvement but will see how the full render goes - hopefully less than 18 hours...

[Oshyan]

It does indeed look like it's slowing down when it hits the Water Shader 03 area, which makes sense. You've got a rough surface (terrain) with raytraced reflections which are interacting with a fairly high number of cloud layers as well. And there are some real discontinuities in the terrain from what I can see when I move up closer, which I think come from an added Displacement effect with high amplitude that is masked by altitude and with a fractal breakup. Strongly discontinuous surfaces cause lots of interreflection which slows things down in the Water Shader.

The bottom line is I would say if you want a reflective effect here, you will probably be just fine with a Reflective Shader *without* raytraced reflections (uncheck Raytraced Reflections). You will still get the specular/shiny effect that tells you a surface is "reflective", but it will take much less time to render. The terrain is so rough and the reflective areas so small, patchy, and hidden in fog that you are not likely to even be able to tell whether real reflection is happening or not. In fact in many cases non-raytraced reflectivity can be quite effective since you need a rather mirror-like surface to really tell if there is real reflection.

- Oshyan