Associate
- Joined
- 8 Oct 2004
- Posts
- 2,283
Here is the first (as I can see it), benchmark comparing CUDA performance within VRay RT using GPU-enhancement: VERY promising!
Notice that OpenCL is totally broken with the current drivers, so ignore that result. But that time saving is HUGE. Very significant. This is what we were hoping for!
EDIT: That was from a chap called Brian on the VRay forums, so not publicly accessible unless you own Vray.
I quote myself for reference, but I did realise that this is comparing to a 680 not a 580. 680s are notoriously poor compared to 580 for CUDA rendering. Something to do with the way the architecture changed and not making use of all the cores etc... so the REAL test will be Titan V 580. Hopefully someone will post some results soon... but meanwhile a 47% Speed Boost of Titan over a 680 is GREAT. Very promising.
Last edited: