OTOY Forums

HI there!
today my 7 GPU Rig is finished.
But i have a problem.
the export takes longer as the rendering.

Is there a option that one Frame will be calulate from one GPU?
It will speeed up my rig....

Greets NICK

could You give some details of Your rig? like what CPU You're using, what is cooling solution? Could You OC that piece?..

glimpse wrote:could You give some details of Your rig? like what CPU You're using, what is cooling solution? Could You OC that piece?..

no the problem is not my rig.
Export time is about 2 Seconds...

the problem is Octane Render.
If u have a little Scene (maybe one object to illustrate) the Rendering with 3 Cards tooks 10 Seconds/Frame.
Now i have a Slave Machine with 7x 980ti.......witch takes 2 Seconds to export and when they start the rendering takes only 2 Seconds to finish.

So it will be better/faster when octane handle every frame with one GPU.

I hope u understand it?

exporting 7 times to be rendered on each GPU..
or
exporting 7 times to render on all GPUs seems the same for me.. as You do render seven times & export seven times..

I'm asking what CPU You have, how fast it runds & whether You use Your 7GPU as render node..

if..You have slow CPU - change it,, overclock it..whateverr, but make 4cores to run as fast as You can.
if..You use this as a slave.. better save out Your file, open on that machine & run it on that machine alone (to eliminate lan influence).

I gues, some fancy thigns could be done on Octane sides from developers as wel,, but i'm trying to help You with hardware side here..

Sometimes it really is faster to let different gpus or workstations render different frame ranges instead of to let all GPUs render the same frame over the network.

Octane should have a way to deal with those scenarios and be able to assign brackets to machines or gpus. This would greatly speed up rendering in some cases. I often just start different frame brackets manually on each workstations instead of to use network render.

glimpse wrote:exporting 7 times to be rendered on each GPU..
or
exporting 7 times to render on all GPUs seems the same for me.. as You do render seven times & export seven times...

sry this is not right.

NVN wrote:

glimpse wrote:exporting 7 times to be rendered on each GPU..
or
exporting 7 times to render on all GPUs seems the same for me.. as You do render seven times & export seven times...

sry this is not right.

mind explaining why, NVM? what You say is that having let's say a lot of GPUs, You waste more time on export? (that's on single machine, with fast interface or You noticed this with networkd rendering, that might explain the issue).

Sorry for botherign, jsut tryign to to understand what's causing the problem.

glimpse wrote:

NVN wrote:

glimpse wrote:exporting 7 times to be rendered on each GPU..
or
exporting 7 times to render on all GPUs seems the same for me.. as You do render seven times & export seven times...

sry this is not right.

mind explaining why, NVM? what You say is that having let's say a lot of GPUs, You waste more time on export? (that's on single machine, with fast interface or You noticed this with networkd rendering, that might explain the issue).

Sorry for botherign, jsut tryign to to understand what's causing the problem.

ok i hope i can explain you with my poor english

The problem is when u have maybe 2-3 GPU on your Main-PC and a Scene with low Spp (because it needs not more) your Slave-PC will start only on the End of the Frame (in the last 1-2 Seconds)
And the Slave-PC will not start with 100 Percent....it takes maybe 0.5-1Second....and the export takes 1Sec too.
So...the full power of my Slave-PC is maybe (if any) only for 1 Sec.



So in my Eyes its better when u can handle one Frame/GPU.

Here is a Screenshot (its a extreme Example) but one Case....
Dont forget my 2 GPU's are 780, not 980TI like my Renderslave.

Sooo...one GPU needs ~8 Seconds.
With 10 GPU i have 10 Frames in 8 Seconds.
As u can see, my Renderslave needs with 9 GPU's (~2 Seconds/Frame) 20 Seconds......

open 7 copies of Maxon C4D. Select one gpu for each one. Divide sequence to 7 and render. This is most effective way to distribute sources.