Return to

1080Ti in x 8 Gen 4.0 PCIe slot for rendering

Hi guys. Just created an account with a question I have. Seems like best place to ask it.

How it will -possibly- affect the performance of a card.

Planning to get TRX40 Designare for TB3 (but I can yet switch to one of these Asus’ triple x16 boards) I also want to slap two 1080Tis (waiting for 3xxx series for upgrade) I’m using now. Unfortunately second x16 slot on Designare will be by default assigned to that Quad NVMe card which is in the box (will use 8 NVMe drives, up from 6 on existing Zenith Extreme X399 system). My question is:

How using Gen 3.0 VGA in Gen 4.0 x8 slot may affect the performance. I don’t talk about gaming here. Interested only in rendering. Using CUDA/IRAY ecosystem (DAZ/Blender/MD/bits of Substance and Resolve).

I know that for gaming there is no issue, even ignoring the fact there is no support for SLI these days, all of my … few… games (like ME Trilogy) are old and never break 60 FPS so couldn’t care less about gaming.

Problem is that rendering is much different thing. When you load GPU, memory and frame buffer to full with tasks and textures suddenly that x8 electrically starts to have some impact. Read multiple articles and most of them direct me at the 80%-85% of nominal performance.

Sadly can’t test it in any way. Don’t have time or hardware to do that. All input much appreciated.

This may help. Last time I checked bandwidth approximately doubled each generation so your 8x 4.0 should support 16x 3.0 speeds

1 Like

While 8 lanes of 4.0 would be as fast as 16 of 3.0 both devices have to agree to talk at 4.0 to get that benefit so unless you are using Gen 4.0 GPUs your only going to get 3.0 x8 speeds stil


The issue is that even 2080Tis barely saturate 8x 3.0 so I doubt the bandwidth issue will cause any major problems beyond 1-2%.

1 Like

Correct Not saying its going to be an issue, just making sure they know it will be operating at 3.0 not 4.0 so the double bandwidth isnt a thing right now with his current 1080ti set up


Yes, I’m aware of this GN article. However I clearly stressed that I’m not interested in gaming.

When you dealing with constant (10, 30, 50h) 90-100% load on VRAM and 100% on GPU and frame buffer with few GB to 30GB+ of RAM assigned to the scene render, game performance means nothing - and I clearly stated that for gaming I know there is no problem. No game in existence can utilize that much resources on one 1080Ti let alone two.

I’m also aware that Gen 3.0 card will not magically start working in 4.0 standard.

Thanks for that [email protected] link tho. Interesting, haven’t done [email protected] in years, but if he can model proteins via USB riser link I shouldn’t have problems with x8 electrical stability - even considering that older versions of DAZ clearly were hampered by IRAY performance on cards running in x8 configuration.


As you can tell the load type doesnt really affect the results of that it still doesnt matter

Now that’s what I’m talking about. Thanks man for this. :+1: Although Octane for licensing reasons is out of my reach, but it clarifies the issue at hand.

I’m aware of Puget Systems, but not from USA or that part of the World in general.

They do the best reports on stuff like this to be honest