That's encouraging, good to see BVH8 and Embree are faster on all these scenes, with Embree 15-25% faster than our BVH4. Koro with Embree shows a quite different render, probably due to lacking support for minimum pixel width, so that one is not really comparable.
Regards, Brecht. On Tue, Nov 14, 2017 at 12:04 PM, Jaros Milan <[email protected]> wrote: > Hi, > > I made some new tests on Intel Xeon Haswell and Intel Xeon Phi Knights > Landing (KNL) using BVH2, BH4, BVH8 and EMBREE-BVH8: > > http://blender.it4i.cz/research/bounding-volume-hierarchy-bvh/ > > Please let me know, what do you think? > > Special thanks to Stefan Werner for Embree implementation. > > Best regards > > Milan Jaros > > _______________________________________________ > Bf-committers mailing list > [email protected] > https://lists.blender.org/mailman/listinfo/bf-committers > _______________________________________________ Bf-committers mailing list [email protected] https://lists.blender.org/mailman/listinfo/bf-committers
