Indigo Renderer Forum

Oscar J wrote:I'd recommend you get the 6GB 780 card if you want to do any somewhat large renders in Indigo. 3 GB will give you 768 MB in Indigo, which you don't get very far with.

The whole RAM situation is something you really have to consider karklinskarlis1993!

How much RAM do your renderings use in general?
For my projects using LightLayers, SS3, big ass HDRI and many textures and models I often end up with at least 8GB and often more.
And to be honest the trend is rather to get even more fancy and consume even more RAM!
Imagine Indigo will support render channels like direct light, reflections, normals etc. on the go at some point. All in 32bit and full SS and LightLayer dependent!
This will cost some extra RAM for sure!

Some crippled NVIDIA RAM is no option here (for me), I hope at somepoint they'll "fix" this...

thats ok, mostly i am rendering my huge scenes on vray. octane/indigo for small purposes, closeups, models in studio etc.
sadly but when i read about this gtx 780ti windforce oc there are loads of complains, very high failure risks, flickering, artifacting.. looks like i will just stick to 980 4gb.

just a quick question- does overclocking gpu improves workflow for cudas? or overclocking makes no changes during gpu rendering?

Zom-B wrote:

Oscar J wrote:I'd recommend you get the 6GB 780 card if you want to do any somewhat large renders in Indigo. 3 GB will give you 768 MB in Indigo, which you don't get very far with.

The whole RAM situation is something you really have to consider karklinskarlis1993!

How much RAM do your renderings use in general?
For my projects using LightLayers, SS3, big ass HDRI and many textures and models I often end up with at least 8GB and often more.
And to be honest the trend is rather to get even more fancy and consume even more RAM!
Imagine Indigo will support render channels like direct light, reflections, normals etc. on the go at some point. All in 32bit and full SS and LightLayer dependent!
This will cost some extra RAM for sure!

Some crippled NVIDIA RAM is no option here (for me), I hope at somepoint they'll "fix" this...

Nick mentioned that tile rendering to save memory could be something to look at if necessary.

Is there big difference in rendering speeds?

I need to upgrade my mobo. I found one, It supports PCI-E x16 3.0, but my GPU supports PCI-E x16 2.1.

Maybe it is silly question, so don't judge

Thank you

Tramis wrote:Does the PCI-Express speed affects the rendering speed?

Why I am asking? When people are making Bitcoin, Litecoin minning PCs, it doesn't matter which PCI-E slot it uses. Main calculations are calculated by GPU.

Maybe it is silly question, so don't judge

Thank you

The reason for this smiley was the picture he posted of his rig before.

The PCI Express bus should be relatively unused (as you say, the GPUs are doing the work), however for low-latency / realtime mode it could make a significant difference I guess. So if you're rendering for longer than 10 seconds or so then it probably won't make any difference.

Oscar J, it wasn't my rig. I wish it was mine.

Maybe in the future releases GPU accelerated rendering will have more Rendering Modes? That would be really awesome!

Nvidia Titan X cards are just landing in stores now. 12GB DDR5 Ram and 3000 cuda cores.

not sure how these compare to the ATI Open CL cards - anyone hazard a guess?

Price is $1700 USD or similar. Sounds expensive but actually could save needing a 2nd PC.

I think it's safe to say that for that price, two 8GB AMD 295X2's (would add up to more than 11000 shader cores) for example would give you a lot more performance. If 8GB of memory is enough for you, that is.