OTOY Forums

Hey folks,

is it me or is the performance of octane while working really slow.
I'm not talking about the render performance, it's about how the operating system
is acting.
When i open Firefox or try to do something else while rendering in octane
everything is in slomo! But also checking some nodes in octane is really a pain in
the ass while rendering.

Sidenote:

Im curious to know that Nvidia settings other users are using.
I'm talking about the nvidia display settings.

Are their any hints to know, in order to boost performance?

cheers

Kian

Recommended is a second Graphic Card for the Display.
While rendering you should have a better performance in your workflow.

I have found that the larger the render size, the slower your O/S will respond. So I will commonly work with a smaller image (800x600) until I get everything the way that I want it, and then I will increase the render size for a final render.

@GeoPappas

Actually that is what i'm always doing.
Actually it is cumbersome to build up the scene with a huge render size.

Yeah a secon card would be the way to go i think.
Cause to be honest, when octane is rendering it is a pain
in the ass to work further with my workstation.

Besides, is there any PAUSE button in order to temporarily stop the
rendering?

Kian

naik3d wrote: Besides, is there any PAUSE button in order to temporarily stop the
rendering?

Yes, it's the two vertical bars at the very top.

Just for your information, this is what happens when the rendering runs:

The rendering runs in a loop. Each iteration Octane does a render pass and then displays the new result as OpenGL image in the GUI. Each render/ pass is done for the whole image and the more pixels it has, the longer it takes (exactly proportional to the number of pixels (height x width)). During that time the CPU still can do other stuff, but the GPU is locked and with that the framebuffer for the display doesn't get changed and although "in the background" the GUI changes you don't SEE it. There is nothing we can do, without reducing the render performance.

The only practical way around this problem, is to use a second graphics card, that is not rendering and therefore doesn't get locked and still can update its framebuffer.

Cheers,
Marcus

abstrax wrote:Just for your information, this is what happens when the rendering runs:

The rendering runs in a loop. Each iteration Octane does a render pass and then displays the new result as OpenGL image in the GUI.

Cheers,
Marcus

presumably this is part you are rewriting.Dumping the open GL code and using Cuda to display the image part.I assume if you do the rewrite and as all the code will be in Cuda then there may be a small speed up in execution as there is no translation from the cuda code to open GL code.

Cuda is a mathematical library that uses the GPU to make calculations, like the mathlib on C++ for the CPU.
Cuda is nothing to display anything.
If you want to display lines, points or whatever, you need OGL or DX.
Because Linux don´t have DX, OGL is the best choice to meet both worlds.

face

face

thanks for the explination.

I did not know, not being a C++ programmer.

I know that sometime in the past it was mentioned that they were going to write the open GL code out,I assumed it was this part of it.I must be wrong.As you say Linux & Apple OSX would not be using Direct X so i guess it's got to stay.