Author Topic: Frame buffer & Node rendering problem  (Read 619 times)

2018-03-14, 21:21:01

zules

  • Active Users
  • **
  • Posts: 78
    • View Profile
Hello in here !
Sorry to bother you again, but i got some problems on heavy scene when rendering with Corona...

First the main machine (from which I render) doesnt start rendering while my other WS do it (as a node).
As shown on the screenshot, the node do every passes.

So, I think it lead to the VFB pb, where strips appears. As the initial pass seems to bug on my main computer, VFB shows those black marks (sorry I needed to pain some black for discretion).

I get rid of this by restarting my main computer, but not very efficient on huge scene.

This WS is a bi xeon, 64 gb and the node the same in bi xeon and 128 gb. Corona 1.7 HF 2

Thank you !

2018-03-15, 10:31:26
Reply #1

Ryuu

  • Corona Team
  • Active Users
  • ****
  • Posts: 578
  • Michal
    • View Profile
Could you please try generating minidump from this situation? See the link in my signature for steps on how to do that.

2018-03-15, 18:34:18
Reply #2

zules

  • Active Users
  • **
  • Posts: 78
    • View Profile
Here they are !
I posted the minidump as Jules GAILLARD on your dropbox (300mb file size).

Thanks !
« Last Edit: 2018-03-15, 18:40:32 by zules »

2018-03-16, 15:39:02
Reply #3

Ryuu

  • Corona Team
  • Active Users
  • ****
  • Posts: 578
  • Michal
    • View Profile
That's weird, from the minidump I can see many Corona threads rendering.

Does this issue occur only on this single scene, or is it with any scene? Does everything work correctly when you render locally without DR? Is DR render still faster than just local render (maybe just the pass counter is wrong)?

If it is just this one scene, could you please share it with us for testing?

2018-03-16, 16:45:17
Reply #4

zules

  • Active Users
  • **
  • Posts: 78
    • View Profile
As I remind, just this project...

As everything is logging SO much, I cant really try to region render things... that's so annoying, I'm getting crazy as time past and deadline approach.

2 screenshots to illustrate, the DR one is the WS working as a node.

I joint you a Sini Forensic ss analysis to show the scene hierarchy.

The scene is 1Gb... and really private (competition)... sorry I cant.
« Last Edit: 2018-03-16, 16:51:57 by zules »

2018-03-16, 18:44:20
Reply #5

zules

  • Active Users
  • **
  • Posts: 78
    • View Profile
Ok I tried something..

I launched the scene on my WS02 (the usual node) and the render start prompt, 15 passes in 9min.
But the WS01 is style parsing the scene after 11 min...

Maybe do I need to format this WS01 ?

Specs :
WS01 : 2x 2690v2 64 gb
WS02 : 2x 2698v4 128 gb

2018-03-19, 10:22:39
Reply #6

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 9058
  • Marcin
    • View Profile
Hi, since you have different RAM amount on the master and the slave, maybe you are running out of memory on the master?
Could you tell us what is your RAM usage on the master and on the slave?

2018-03-26, 11:37:03
Reply #7

zules

  • Active Users
  • **
  • Posts: 78
    • View Profile
Yes there was a difference between the 2 computers.
I remember having swap on the master yes, but the node was ok...

2018-04-13, 17:41:36
Reply #8

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 9058
  • Marcin
    • View Profile