OTOY Forums

fantome wrote: The Asrock X399 Taichi allow only 3 GPU because the case panel plug make contact with the last GPU card. you will need to do some tweaking to modify that. I have try to find a 90 degre adaptator on the net, but don' t find any.
if you are good with electronics it must be easy to create your own.

I have few of these bought for my DualXoen motherboards, Almost all MBs have that problem if you want to plug in PCI-E slot on bottom.

This is solution:
http://prntscr.com/pqvezy

If You could export scenes as ORBX and upload it we can try to test things out side by side on given standalone build. have few systems that I could check those files.

...uhm the .orbx file is hudge (about 500GB zipped).
How I could share it with you guys?

GIOLETS wrote:

If You could export scenes as ORBX and upload it we can try to test things out side by side on given standalone build. have few systems that I could check those files.

...uhm the .orbx file is hudge (about 500GB zipped).
How I could share it with you guys?

500GB or 500MB ?

Sorry... 500MB

GIOLETS wrote:Sorry... 500MB

you can upload that to www.wetransfer.com and it'll be there for 7 days (it's free up to 2GB).

Good idea!

Here the link:https://we.tl/t-pYYb3YP0k6

I have measured the time between I hit render icon and the moment when the image appear on Live Viewer.
For my machine (i7 4790K @ 4.0GHz) is 36 seconds in Cinema4d, 20 in Standalone (2018 version)

Planning my long overdue (10 years!) PC upgrade and was wondering if the latest AMD Ryzen CPUs would help, what with them running PCIE4 rather than PCIE3?

Would the extra PCI bandwidth help cram all that information into the graphics card faster?

Or is the long pause before seeing a render appear all down to CPU calculations?

I bet I'm not the only one who gets fed up of staring at 'scene evaluation' type messages. Never really knew which bit of the PC is slowing down this stage of the process - CPU, SDD, RAM, PCIE...

fantome wrote:@GIOLETS, you sucessfully setup NVlink with 2 * 2070 ? i thought they didn' t have Support for NVLink SLI,? how do you achieve that ?
you can run octane with 16G of ram with 2 * 2070 ?

EDIT :
https://www.nvidia.com/fr-fr/geforce/gr ... 070-super/
Gotcha, i didn' t know NV has release new 2070 with a new specs !

I'm pretty sure that NVLink on 2070/2070Super, 2080/2080Super, and 2080Ti are just getting SLI with that, but 50x faster. In other words, memory does not stack up until you get into Quadro RTX...

Also bro, I am completely mystified too why I can never find a solution (the couple times I've looked) to the motherf***ing 4th GPU problem with case headers and USB3 getting in the way. It sucks. When I bought my mobo, I sure as hell thought I'd be slapping four GPUs in this bastard...

FrankPooleFloating wrote: I'm pretty sure that NVLink on 2070/2070Super, 2080/2080Super, and 2080Ti are just getting SLI with that, but 50x faster. In other words, memory does not stack up until you get into Quadro RTX...

NVLINK is not available on 2070s, only on 2070super and up (2080, 2080super, 2080ti, TitanRTX and some older quadro, tesla cards).

The thing with NVLINK, You do not double VRAM, but before going out of core OctaneRender uses faster GPU memory (VRAM) of second GPU and that is way faster compared to going into system memory, thus when You NVLINK capable cards, You do not get so much of a penalty using OOC.

If Your scenes are small, You will not notice too much, but if You use 11GB 2080ti or 24GB RTX Titans, and store as much data over what GPU offers..then You can get back like 20% more speed, just because You do not need to go to RAM.

Hmmmm.. but if all GPUs are rendering Tom, then how often will that scenario be used? It would have rocked so much if RTX gpus could have also received the same treatment as Quadro (HBM2)...

Funny coincidence, I just ordered three EVGA 2070 Super Hybrids about an hour ago, to replace my EVGA 980Ti Hybrids... 2070 Super gets 220 OB, and 2080 Super gets 233. Yeah, I'm not sure that 6% faster is worth the extra dough for a 2080 Super. That would have been like another $600 more to get all 2080 Supers... But of course a day will come where I'll be trying to beat a deadline with all three 2070 Supers cooking, and me sweating bullets, cursing myself for not getting the 2080 Supers (or maybe even Ti)...