[=;201185]
I just did a similar test using FPS template:
://i.imgur/J20YrPT.jpg
So I did with 2 maps: one with manually placed foliage and another one using foliage tool, without changing any options. Both maps have ~4k instances of SM_Bush
Using exactly same point of view, here are results on 3 different PCs:
GTX 770:
manual 4.6: ~22fps
manual 4.7: ~20fps
foliage tool 4.6: ~66fps
foliage tool 4.7: ~62fps
GTX 770 mobile (laptop):
manual 4.6: ~15fps
manual 4.7: ~16fps
foliage tool 4.6: ~47fps
foliage tool 4.7: ~43fps
GTX 970:
manual 4.6: ~35fps
manual 4.7: ~34fps
foliage tool 4.6: ~100fps
foliage tool 4.7: ~90fps
It’s really weird we got such different results. And for looks of it, in my tests 4.6 version performed better.
[/]
You placed ~4k instances, Terrian had ~25k of them, so over 6 times more. Perhaps new foliage rendering method has slightly bigger overhead, but performs much better with significant number of foliage instances. Also, there may be other reasons for performance differences between engine versions considering that in most tests FPS difference was insignificant and could be just measuring error/badly prepared test case.