i can reproduce it and i'm fixing it.Amplitude wrote:Has anyone else noticed this?
version: beta 2.2 RC2
When you create a material , for instance a diffuse, and then switch it to something else all the little dots in the node graph disappear.
However if you have connected another function to the base shader first, it doesn't happen.
Here is a screen capture.
I think it's a new bug, tested with the beta 2.1 and everything works fine there.
As Radiance pointed, our argument parser supports quotes to delimit elements with spaces in them. however, when you type it from the shell, the shell interpretes the " and discard them. To prevent that, you should escape them with backslashes too i.e:livuxman wrote:I used the version of the benchmark set by acc24ex and seen that there is a problem (at least in the linux version)
If I try to load directly from the command line with "octane OctaneBenchmark/octane_benchmark\ for\ 1022\ beta\ 2.2rc.ocs" the program exits with:
ERROR: Program executed illegal instruction.. terminating
It seems something related to the treatment of spaces in the name of the files, without them it works perfectly. Loading from within octane works well with spaces.
the lambo has 4 million triangles and consumes 1100 MB video memory and runs fine on dual GTX480...ltchest wrote:I had found that, with the latest pre-multiGPU release, i no longer seemed to have the problem with Octane (or CUDA) crashing when a scene takes too long to start rendering. For example, now with the multi-gpu version (using 3 x GPU) and a really simple scene like the Benchmark scene, it is no problem. IF i load in a complex scene though, it crashes before it can start rendering. THis has always been a registy fix issue, but i had already done that and it was working OK. When i tested it yesterday it just crashed before it could start to render. Has anyone else tried a complex scene with multi-GPU ? The kind of scene where it takes several seconds before it starts to render ?
