GPU type and scene loading times

Newtek Lightwave 3D (exporter developed by holocube, Integrated Plugin developed by juanjgon)

Moderator: juanjgon

Post Reply
User avatar
BorisGoreta
Licensed Customer
Posts: 1413
Joined: Fri Dec 07, 2012 6:45 pm
Contact:

How come that GPU type affects scene loading and preparing times in this example:
Capture.PNG
I am not sure what software is that but renderer is Octane.

https://youtu.be/VMu7KUcaxgQ?t=566
User avatar
juanjgon
Octane Plugin Developer
Posts: 8867
Joined: Tue Jan 19, 2010 12:01 pm
Location: Spain

I have no details, but probably some of the preprocessing happens on the GPU side, like for example the textures preprocessing, the mesh subdivision or the vertex displacement.

Thanks,
-Juanjo
User avatar
Lewis
Licensed Customer
Posts: 1101
Joined: Tue Feb 05, 2013 6:30 pm
Location: Croatia
Contact:

Andrey usual uses Cinema4D.


But what struck me in that video is that when he used Out Of CORE in Octane he said: "...if I would start render from frame 0 it would start right away, BUT i was interested to start render from frame 1500....).
Why is that chokinng Octane + out of core+octane so much is mystery to me.

P.S. He says he switched ot Redshift due BAD Out Of Core in Octane.
--
Lewis
http://www.ram-studio.hr
Skype - lewis3d
ICQ - 7128177

WS AMD TRPro 3955WX, 256GB RAM, Win10, 2 * RTX 4090, 1 * RTX 3090
RS1 i7 9800X, 64GB RAM, Win10, 3 * RTX 3090
RS2 i7 6850K, 64GB RAM, Win10, 2 * RTX 4090
Iceman9
Licensed Customer
Posts: 217
Joined: Sun Jun 09, 2013 1:22 am

Lewis wrote:Andrey usual uses Cinema4D.


But what struck me in that video is that when he used Out Of CORE in Octane he said: "...if I would start render from frame 0 it would start right away, BUT i was interested to start render from frame 1500....).
Why is that chokinng Octane + out of core+octane so much is mystery to me.

P.S. He says he switched ot Redshift due BAD Out Of Core in Octane.

Two different things getting confused as one thing:
-Out of core rendering
-Simulation generation

His screen might have said, "Scene Preparation" but what was happening in the scenario above is the sim was having to run up to frame 1500. If he'd cached the sim he'd had experienced different results. That's my take anyway.

Andrey is bright and creative but he makes mistakes. For instance around the 4-5 minute mark he shows he doesn't understand Nvidia's relationship with Samsung and AIB partners.
User avatar
james_conkle
Licensed Customer
Posts: 162
Joined: Sun Oct 27, 2019 10:28 pm
Location: Brooklyn, NY
Contact:

In past videos he said if you want a linear workflow you need to check “neutral response” in the camera imager without even using a LUT so that checkbox did nothing. And he never mentioned converting textures to linear space. So not totally buying it.
Last edited by james_conkle on Sun Dec 27, 2020 2:08 am, edited 1 time in total.
Octane C4D Plugin 2021.1 / C4D R25.015
UNIT 0: 2080ti / 6700K / 64GB / 500w
UNIT 1: 2x 3090 / 3970x / 128GB ECC / 1600w
UNIT 2: 5500M / 9980HK / 32GB / 95w
RNDR Artist Calculator
User avatar
Lewis
Licensed Customer
Posts: 1101
Joined: Tue Feb 05, 2013 6:30 pm
Location: Croatia
Contact:

Iceman9 wrote: Two different things getting confused as one thing:
-Out of core rendering
-Simulation generation

His screen might have said, "Scene Preparation" but what was happening in the scenario above is the sim was having to run up to frame 1500. If he'd cached the sim he'd had experienced different results. That's my take anyway.
But eeve if is just simulation time still why would that do any such noticeable difference in START of rendering?

If is VRAM usage then Titan RTX should be roughly same speed as 3090 'coz they both have 24GB and Titan RTX was 23 minutes vs 4 minutes on 3090.
That is very big difference. I could understand that 2*2080Ti with NVlink would go slower (i still don't see Octane properly usign NVlink, it shows VRAM combined but somehow still uses OOC in many cases so i think there is still some error in Octane that prevents it to properly use Nvlink).

But on the other hand as you say maybe he mixed up something in his test.
--
Lewis
http://www.ram-studio.hr
Skype - lewis3d
ICQ - 7128177

WS AMD TRPro 3955WX, 256GB RAM, Win10, 2 * RTX 4090, 1 * RTX 3090
RS1 i7 9800X, 64GB RAM, Win10, 3 * RTX 3090
RS2 i7 6850K, 64GB RAM, Win10, 2 * RTX 4090
pixym
Licensed Customer
Posts: 597
Joined: Thu Jan 21, 2010 4:27 pm
Location: French West Indies

How can such a scene take 56mn to prepare ???
Work Station : MB ASUS X299-Pro/SE - Intel i9 7980XE (2,6ghz 18 cores / 36 threads) - Ram 64GB - RTX4090 + RTX3090 - Win10 64
NET RENDER : MB ASUS P9X79 - Intel i7 - Ram 16GB - Two RTX 3080 TI - Win 10 64
User avatar
glimpse
Licensed Customer
Posts: 3740
Joined: Wed Jan 26, 2011 2:17 pm
Contact:

have both Titan RTX and 3090 - haven't notice any difference in loading.

my guess is that something else have to do with this, like Iceman9 pointed.

however I've seen other artist pointing similar experiences:
BEN.JPG
Source: https://www.facebook.com/ben.mauro.3/po ... 9411097318

& some discussion that followed the post:
discussion.JPG
again, if user felt the difference - that's cool, I'm not saying he's lying or so. I'm just really curious where that comes from and would love to be able to leverage that.
Post Reply

Return to “Lightwave 3D”