Why AMD cards are so slow in render software? And for so long why i didn't see any move from AMD to close this gap?
The only software that i know, that still suporting professional gpu render in open Cl, is vrayRT, now in version 3.4.
In this thread: Chaos Group Forums , using version 3.3 a guy with a firepro W8000 was able to render the test scene in 41m 51,3s
where Titan X in cuda achieve the same results in only 10m 07s.
Here is a link to get a picture of the actual situation: V-Ray RT GPU CUDA v3.10.02 - Render time in sec (smaller is better) - Google Sheets
Probably OpenCL isn't so good for this kind of job, but i really see a future in GPU render, and will be nice to see AMD doing great in this front too.
if anyone with a more capable card W9100 as example using this software could benchmark the scene with latest AMD driver?
Thanks in advance.
Several websites I visit are reporting that the current AMD opencl 2.0 drivers are botched for GPU renders. (At least for Luxrender users, V-Ray users I assume would have the same issue). Once the drivers are working as intended perhaps you'll see better benchmarks?