Author Topic: Computing secondary GI taking ages  (Read 19511 times)

2016-05-24, 19:33:59

serch

  • Active Users
  • **
  • Posts: 39
    • View Profile
Hi there,

I am having issues with the secondary GI. Sometimes it takes very long to process it, I've notice that if a close MAX and re-open the file it will temporarily fix the problem.
It gets a little annoying when I'm trying to test materials or lighting.

Using Corona 1.4
All the settings on default

Thanks in advance.

2016-05-25, 14:04:18
Reply #1

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12759
  • Marcin
    • View Profile
Maybe you are running our of RAM?
Are you getting any warnings/errors?
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2016-05-25, 14:49:42
Reply #2

PROH

  • Active Users
  • **
  • Posts: 1219
    • View Profile
In case you're using Max2016 then make sure you have SP3 installed. There was a severe memory leak before that.

2016-05-25, 17:56:22
Reply #3

serch

  • Active Users
  • **
  • Posts: 39
    • View Profile
No warning or errors,
It could be the RAM, I'm actually running 2016

I notice if a close some of the other programs it will go fast and smooth, so I will try with the SP and a RAM upgrade.
Thanks for your replies

2016-05-26, 02:06:16
Reply #4

serch

  • Active Users
  • **
  • Posts: 39
    • View Profile
I notice CPU and RAM usage it's not at least at 80% and 16 mins for GI calculation for a 800x400 test image it's way to much :(

Any thoughts?, when I was using version 1.3 I didn't have this issue.

Thanks in advance. 

2016-05-26, 12:07:55
Reply #5

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
I have experienced similar on rare occasions, although speed picked up again 20 seconds after letting it run slow. Never had again since 1.4.
Shot in the dark... But try to type these three commands into the bottom right max script command window.
Code: [Select]
gc() Garbage Collection.
freescenebitmaps() Cleans up in texture memory.
clearUndoBuffer() Removes your Undo's
Logic being, that in heavy scenes all Undo steps are saved and can add up to gigabytes with heavy changes respecrivly. We may crash past our max's ram limit and it started paging out parts of max to compesate, which is a classic ruiner of performance.
If this is the case, then undo delete won't immidietly pickup the speed again, but maybe 10 secs later.
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-05-26, 14:36:34
Reply #6

vkiuru

  • Active Users
  • **
  • Posts: 320
    • View Profile
I never had this until 1.4, so far I've found two solutions. The first one being merging the problematic scene to a blank Max file, the other one being freeing up tons of space on the HDD I have my virtual memory/swap files on. I don't entirely get why freeing up virtual memory helps when the scene is using about 3 gigs out of the available 32 gigs of RAM.

Oh and sometimes running Corona Converter and running the Start Conversion, clicking on disabling legacy mode and Coronabmp -> Bitmap helps, even though there is nothing to convert :P I'd like to send a scene if this is something else than a memory problem but right now I'm on a tight schedule. Would still be interested in more expert opinions since this can be seriously time consuming. Can it really be the virtual mem?

2016-05-26, 14:49:17
Reply #7

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
Can it really be the virtual mem?
Just a shot in the dark, since you mentioned 80% ram usage. Paging starts at ~90%
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-05-26, 15:00:40
Reply #8

lixai

  • Active Users
  • **
  • Posts: 11
    • View Profile
try to disable displacement in the common tab in render settings. maybe something you imported has displacement activated

2016-05-26, 17:10:59
Reply #9

vkiuru

  • Active Users
  • **
  • Posts: 320
    • View Profile
Can it really be the virtual mem?
Just a shot in the dark, since you mentioned 80% ram usage. Paging starts at ~90%

To be clear it wasn't me, but the OP who mentioned the ram usage % :)

2016-05-26, 18:39:19
Reply #10

serch

  • Active Users
  • **
  • Posts: 39
    • View Profile
I never had this until 1.4, so far I've found two solutions. The first one being merging the problematic scene to a blank Max file, the other one being freeing up tons of space on the HDD I have my virtual memory/swap files on. I don't entirely get why freeing up virtual memory helps when the scene is using about 3 gigs out of the available 32 gigs of RAM.

Oh and sometimes running Corona Converter and running the Start Conversion, clicking on disabling legacy mode and Coronabmp -> Bitmap helps, even though there is nothing to convert :P I'd like to send a scene if this is something else than a memory problem but right now I'm on a tight schedule. Would still be interested in more expert opinions since this can be seriously time consuming. Can it really be the virtual mem?

I did tried merging the scene to a blank max, it was fine for a few test renders and then the performance drop again. I will try with the corona converter fix and let you know my findings.

I have experienced similar on rare occasions, although speed picked up again 20 seconds after letting it run slow. Never had again since 1.4.
Shot in the dark... But try to type these three commands into the bottom right max script command window.
Code: [Select]
gc() Garbage Collection.
freescenebitmaps() Cleans up in texture memory.
clearUndoBuffer() Removes your Undo's
Logic being, that in heavy scenes all Undo steps are saved and can add up to gigabytes with heavy changes respecrivly. We may crash past our max's ram limit and it started paging out parts of max to compesate, which is a classic ruiner of performance.
If this is the case, then undo delete won't immidietly pickup the speed again, but maybe 10 secs later.

I will try this sounds interesting.

try to disable displacement in the common tab in render settings. maybe something you imported has displacement activated

I did, not a change on GI cal.

Thanks for the replies and solutions, will give and update as soon as I try everything.

2016-05-27, 11:00:08
Reply #11

vkiuru

  • Active Users
  • **
  • Posts: 320
    • View Profile
Just a thought, try rendering the scene with an override neutral Corona Material and see if it changes things. If it does, start excluding materials from the override.

I got my scene to work smoothly again doing this, going through the exclude-list keeping priority on the most complicated materials or the ones that had otherwise questionable setups. It turned out to be my wall material which had some gimmicky bump map combining round corners, composite normal maps and standard bump maps. Once I simplified the bump channel the scene has had no hiccups since.

2016-05-27, 20:12:51
Reply #12

serch

  • Active Users
  • **
  • Posts: 39
    • View Profile
Just a thought, try rendering the scene with an override neutral Corona Material and see if it changes things. If it does, start excluding materials from the override.

I got my scene to work smoothly again doing this, going through the exclude-list keeping priority on the most complicated materials or the ones that had otherwise questionable setups. It turned out to be my wall material which had some gimmicky bump map combining round corners, composite normal maps and standard bump maps. Once I simplified the bump channel the scene has had no hiccups since.

After several minutes of trial and error I found the issue, a converted proxy, once that I deleted that proxy the performance increased.

Thanks you for the suggestion

2016-05-28, 14:21:55
Reply #13

vkiuru

  • Active Users
  • **
  • Posts: 320
    • View Profile


After several minutes of trial and error I found the issue, a converted proxy, once that I deleted that proxy the performance increased.

Thanks you for the suggestion

Nice to hear you got it solved :-)

2016-05-28, 21:23:52
Reply #14

fellazb

  • Active Users
  • **
  • Posts: 281
    • View Profile
Do you have any idea why this proxy was "corrupted"? Have you tried to reproxy it again and was it ok then? Curious to know why it caused this problem.