2906

Message boards : Comments and discussion : 2906
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 14208 - Posted: 17 Dec 2015, 11:47:06 UTC

I'm not too sure that this one will actually render but since nothing else is in the queue we can at least give it a shot.
Some of those render settings are way crazy btw - that's just an insane amount of sampling going on for everything.

For similar future renders (here on BURP at least) you should consider setting AA to 1 and use a much higher resolution instead. Then we can split it into more parts and render it faster. Once you get the result you can then scale it back down and you get the same effect as AA.
ID: 14208 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
patr5589
Avatar

Send message
Joined: 3 Nov 15
Posts: 9
Credit: 104,348
RAC: 0
Message 14209 - Posted: 17 Dec 2015, 21:23:56 UTC - in response to Message 14208.  

I guessed 100 samples were insane, but I set it for that because I saw a heavy amount of grain on a test render with 50 samples.

I do apologies for any other settings I may have overdone, I'm pretty new to rendering things and it's not something I have much interest in, but it's necessary for my college education at the moment. I'm more of a 3D-Game kind of person than a visualizer/whatever.

I'm not sure if I can delete this submission myself, but if it's possible (or something else enters the queue, I guess) then you're pretty much free to do it since this project was scrapped. We got friday off so I don't need to present it in such a short notice anyway.
ID: 14209 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Professor Desty Nova
Avatar

Send message
Joined: 21 Mar 05
Posts: 97
Credit: 294,357
RAC: 3
Message 14210 - Posted: 17 Dec 2015, 22:11:36 UTC
Last modified: 17 Dec 2015, 22:26:12 UTC

This WU result is a bit strange, maybe a bug somewhere? BOINC says success, with a very small CPU time, and the "Stderr output" has a lot of errors...
I noticed that it says Python 2.66 on "Stderr output", but on my computer, the same WU says Python 2.76 on "Stderr output".

PS: It has crunched for 10 hours already, I hope it doesn't crash.
PPS: If I finish it, I hope there are other patient crunchers out there...


Professor Desty Nova
Researching Karma the Hard way
ID: 14210 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 14211 - Posted: 18 Dec 2015, 9:26:15 UTC - in response to Message 14210.  
Last modified: 18 Dec 2015, 9:40:54 UTC

Heh yeah that's certainly a bug. Expected run-time based on results returned is around 50+ hours or so on a very modern multi-core.

@patr5589:
It's no problem. It is just kind of a rule of thumb that single-frame renders on BURP go much faster if you use high-res to remove noise rather than high AA.
This session was split into the (current) maximum for HD which is 60 pieces. That means "only" 120 machines get to work on it in parallel. Double the resolution and we could have sent it to twice as many machines while at the same time getting the results back twice as fast because each peace is simpler with the half'ed AA. Double it again, and again, you get the point. There's usually around 400-600 machines available for low-mem jobs, so no reason to be shy about high-res. 8 AA is comparable to doubling the resolution 3 times.
For animations, on the other hand, it doesn't matter since there are usually enough frames to keep everyone busy.
ID: 14211 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
funkydude

Send message
Joined: 23 Dec 13
Posts: 275
Credit: 2,478,281
RAC: 0
Message 14213 - Posted: 18 Dec 2015, 13:15:02 UTC

If this is urgent you'd be better off cancelling and using the suggested method. The current method may take a month or so, if you're ok with that.
ID: 14213 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
patr5589
Avatar

Send message
Joined: 3 Nov 15
Posts: 9
Credit: 104,348
RAC: 0
Message 14215 - Posted: 18 Dec 2015, 18:59:49 UTC

@funkydude
As I said, it's not urgent anymore. If it's not a problem leaving it up, then it can take as long time as it wants.

@Janus
I'll try and optimize the render with the suggestions you've posted. If I'm successful, would it hurt any of the crunchers to cancel
this WU and upload the new one? Also, I might need help with this, know any text/video tutorials or something similar?
ID: 14215 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Professor Desty Nova
Avatar

Send message
Joined: 21 Mar 05
Posts: 97
Credit: 294,357
RAC: 3
Message 14216 - Posted: 18 Dec 2015, 19:20:04 UTC

After 28 hours it finished :-)
And since another person also finished it, it's completed and validated :-)


Professor Desty Nova
Researching Karma the Hard way
ID: 14216 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
patr5589
Avatar

Send message
Joined: 3 Nov 15
Posts: 9
Credit: 104,348
RAC: 0
Message 14217 - Posted: 18 Dec 2015, 20:40:51 UTC

Well, it looks like this blend file is broken anyway, so it's a waste of time to render it. Take a look at some of the completed bits, the floor is completely pitch black. I guess the texture didn't pack properly.
ID: 14217 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 14218 - Posted: 19 Dec 2015, 17:18:54 UTC - in response to Message 14215.  
Last modified: 19 Dec 2015, 17:28:02 UTC

I'll try and optimize the render with the suggestions you've posted. If I'm successful, would it hurt any of the crunchers to cancel this WU and upload the new one?

It is a bit too late to cancel this one but I'm sure everyone will be thrilled to see what it was supposed to look like - go ahead and upload a larger one with fewer samples and fixed floor when you feel like it.

[Edit:] Switched the session to "Best effort"-mode to use less rendering resources.
ID: 14218 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[AF>HFR]CgS

Send message
Joined: 6 Dec 15
Posts: 1
Credit: 349,890
RAC: 30
Message 14222 - Posted: 19 Dec 2015, 19:02:30 UTC - in response to Message 14218.  

I'll try and optimize the render with the suggestions you've posted. If I'm successful, would it hurt any of the crunchers to cancel this WU and upload the new one?

It is a bit too late to cancel this one but I'm sure everyone will be thrilled to see what it was supposed to look like - go ahead and upload a larger one with fewer samples and fixed floor when you feel like it.

[Edit:] Switched the session to "Best effort"-mode to use less rendering resources.


Computer 60396 seems doing only bad work, switching to "Best effort" will mark its (fast but bad) works as valid, and real (good) work as invalid !

example : work unit 2576986
task 8980997 is clearly invalid (look at preview) but confirmed, whereas task 8980761 is "invalid" while it is rather, in my opinion, the valid one !

(P.S. : sorry if I make mistakes, English is not my native language...)
ID: 14222 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 14224 - Posted: 19 Dec 2015, 20:42:16 UTC
Last modified: 19 Dec 2015, 21:15:55 UTC

Yup, going to look into that

[edit:] Will flip the valid ones for any that come in later. This will be done during next maintenance which unfortunately will not be until after New Year's Eve. Once it reaches 100% rendered I assume it will just stay around until fixed. Probably would have had been easier to just let it render normally with validation... oh well, hindsight 20/20.
This session is now officially completely broken - good thing that #2908 seems to be doing better.
ID: 14224 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
funkydude

Send message
Joined: 23 Dec 13
Posts: 275
Credit: 2,478,281
RAC: 0
Message 14225 - Posted: 20 Dec 2015, 14:06:44 UTC
Last modified: 20 Dec 2015, 14:07:12 UTC

I've had one of my tasks marked as invalid because of that machine it seems. I assume that will be fixed?
ID: 14225 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
patr5589
Avatar

Send message
Joined: 3 Nov 15
Posts: 9
Credit: 104,348
RAC: 0
Message 14229 - Posted: 27 Dec 2015, 2:20:26 UTC

Did this WU break?
ID: 14229 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 14233 - Posted: 27 Dec 2015, 16:32:02 UTC

Yup, fixing it after New Year's
ID: 14233 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4574
Credit: 2,100,463
RAC: 8
Message 14262 - Posted: 6 Jan 2016, 18:36:42 UTC - in response to Message 14233.  

Fixed and nuked. The remaining session was aborted.
ID: 14262 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
patr5589
Avatar

Send message
Joined: 3 Nov 15
Posts: 9
Credit: 104,348
RAC: 0
Message 14268 - Posted: 15 Jan 2016, 22:32:35 UTC - in response to Message 14262.  

Good thing we fixed the render settings for the other upload, then.
ID: 14268 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Comments and discussion : 2906