Viper animation

Viper animation

Description

Viper animation
Motion blur set to 1, AO 16, OSA 16.
A viper is tracked by a plane.

Message boards : Comments and discussion : 450

1 · 2 · 3 · 4 . . . 5 · Next
Author Message
loki
Send message
Joined: 7 Feb 07
Posts: 97
Credit: 158,266
RAC: 0
Message 5958 - Posted: 2 Jun 2007, 3:23:29 UTC

It doesn\'t seem to be having much problems running on my computer. So far, it looks like I had one error out at the beginning and it looks like out of memory, I don\'t really know...

http://burp.boinc.dk/result.php?resultid=2171205

There is some wu\'s in here that are taking about 18 hours to compute straight though, that may pose a problem for some computers.

Profile Jens
Send message
Joined: 1 May 07
Posts: 3
Credit: 1,762
RAC: 0
Message 5966 - Posted: 3 Jun 2007, 19:04:52 UTC

Yes, that session is VERY huge. My 2.4 Ghz Core 2 Duo worked about 5 hours on it and say, 45 min remain. I had an restart between, and so i would not get it in time... nearly 6 hours is very much. I\'m really interessed in the resuts. It must be something very cool... (hoping...)

mip2
Send message
Joined: 12 Jan 07
Posts: 26
Credit: 25,862
RAC: 0
Message 5967 - Posted: 3 Jun 2007, 19:29:57 UTC

take a look in your boinc folder and find /projects/burp.boinc.dk/450in.zip
expand, open in blender. ;-)

FreeLarry
Send message
Joined: 10 Oct 04
Posts: 42
Credit: 1,689,701
RAC: 0
Message 5968 - Posted: 3 Jun 2007, 19:44:07 UTC

Don\'t know about the times but have almost 14000 credit pending for this run ;)

a few erroring out but most going thru alright - one machine over 24 hrs on one unit currently.

Larry
____________

Profile Jens
Send message
Joined: 1 May 07
Posts: 3
Credit: 1,762
RAC: 0
Message 5969 - Posted: 3 Jun 2007, 20:22:03 UTC
Last modified: 3 Jun 2007, 20:36:20 UTC

I\'ve done, what FreeLarry said. It looks, like it is quite boring. OSA 16 and Motion Blur and 16 samples of AO are very computing-time-eating settings. Without these abnormal settings one frame can be renderd in ~15-60 min.
And a second thing is strange: I think an octree resulution of 64 is too smal fpr this scene... 256 should work faster.
EDIT: This Session IS a problem for some machines. If you look at the unfinished workunits, you can find many, which crunchers had errors or do not get the results in time. I think rendering with more parts per frame would do the job... no more crunching 5 hours with a very fast CPU on ONE wu. One hour is ok...

Profile Izarf
Send message
Joined: 29 Jul 05
Posts: 33
Credit: 13,637
RAC: 0
Message 5976 - Posted: 5 Jun 2007, 0:05:30 UTC

That is true, Jens. Pentium 4 3 GHz and ~13.5 hours on this one is way too much. Besides, I think the session should be aborted due to the waste of CPU time. Maby there would be some sort of filter that not would accept these kind of spoiling sessions in the future?
____________

loki
Send message
Joined: 7 Feb 07
Posts: 97
Credit: 158,266
RAC: 0
Message 5977 - Posted: 5 Jun 2007, 0:32:06 UTC - in response to Message 5976.

That is true, Jens. Pentium 4 3 GHz and ~13.5 hours on this one is way too much. Besides, I think the session should be aborted due to the waste of CPU time. Maby there would be some sort of filter that not would accept these kind of spoiling sessions in the future?


No, this is okay, thats why this one is currently in the \"background rendering\" state. This session is likely to take a long time to render, but thats what we\'re all here for, right? To put our computers idle time towards something meaningful. So what if this session is currently going to take the better part of a year to render, that just means we need to get more people working on it. Don\'t get what I said wrong, I\'d rather have long sessions because at least that way theres something to do constantly. Without this project I\'d probably have to go to a different project because there would be no work for this one.

loki
Send message
Joined: 7 Feb 07
Posts: 97
Credit: 158,266
RAC: 0
Message 5980 - Posted: 5 Jun 2007, 3:53:56 UTC

http://burp.boinc.dk/workunit.php?wuid=425328

23 hours at 2.4Ghz

Profile noderaser
Project donor
Avatar
Send message
Joined: 28 Mar 06
Posts: 506
Credit: 1,548,030
RAC: 45
Message 5981 - Posted: 5 Jun 2007, 5:43:00 UTC

I think the long estimated completion time is because it was assigned background status, when it was running with session 459 which went through a lot of failed WUs. As long as there aren\'t any other heavy work loads coming through the pipeline, I doubt this one will take as long as the estimates to complete. However, it looks like this one is generating failed work WUs here and there as well, so it probably won\'t be at optimum.
____________

Profile DangerNerd
Project donor
Send message
Joined: 31 Mar 06
Posts: 126
Credit: 263,029
RAC: 35
Message 5987 - Posted: 5 Jun 2007, 11:58:21 UTC

Hi everybody!

I don\'t know about those of you who are complaining. The alternative is no work at all, so why abort the session?

It is in low priority mode, so other sessions render first, so if you want something else to work on, why not submit a session of your own? :-)

Personally, I am overjoyed to see a test session actually push the boundaries of the project. That is what alpha testing is all about. The fact that we are in alpha test mode, and getting stable supply of WUs is a blessing, not a curse.

Of course if the session simply will not render on your machines, it may be time to switch projects until Janus builds in the ability to opt out of rendering a specific session. (Which I hope is soon.)

I hope that made sense.

DN.
____________
Our Advice is to support all useful BOINC projects. Smart people needed to give advice to those who seek answers: Give or Get Free Advice Here

Profile Zanthius
Project donor
Send message
Joined: 24 Mar 05
Posts: 94
Credit: 1,627,664
RAC: 0
Message 5989 - Posted: 5 Jun 2007, 14:20:57 UTC - in response to Message 5987.

Hi everybody!

I don\'t know about those of you who are complaining. The alternative is no work at all, so why abort the session?

It is in low priority mode, so other sessions render first, so if you want something else to work on, why not submit a session of your own? :-)

Personally, I am overjoyed to see a test session actually push the boundaries of the project. That is what alpha testing is all about. The fact that we are in alpha test mode, and getting stable supply of WUs is a blessing, not a curse.

Of course if the session simply will not render on your machines, it may be time to switch projects until Janus builds in the ability to opt out of rendering a specific session. (Which I hope is soon.)

I hope that made sense.

DN.


Agreed.

On a side note... one of these WU\'s is taking around x3 the normal...

http://burp.boinc.dk/workunit.php?wuid=426096

27 hours and only done 85%

Dont know if most machines will be able to do this within any time constraints BURP has.

____________

Profile DangerNerd
Project donor
Send message
Joined: 31 Mar 06
Posts: 126
Credit: 263,029
RAC: 35
Message 5990 - Posted: 5 Jun 2007, 16:33:57 UTC - in response to Message 5989.



Agreed.

On a side note... one of these WU\'s is taking around x3 the normal...

http://burp.boinc.dk/workunit.php?wuid=426096

27 hours and only done 85%

Dont know if most machines will be able to do this within any time constraints BURP has.


You aren\'t kidding... I have one at 32 hours and counting!

On the high side, that is 32 hours of work. If we dont\' get credit for these for some reason, I will not be happy, but other than that I am thrilled to be able to have all my machines crunching BURP exclusively for more than 3 days in a row. :-)


____________
Our Advice is to support all useful BOINC projects. Smart people needed to give advice to those who seek answers: Give or Get Free Advice Here

Profile noderaser
Project donor
Avatar
Send message
Joined: 28 Mar 06
Posts: 506
Credit: 1,548,030
RAC: 45
Message 5991 - Posted: 5 Jun 2007, 22:23:33 UTC

Are there any plans to have a standard workunit size, so work from slower computers is not lost due to the deadline enforcement?
____________

loki
Send message
Joined: 7 Feb 07
Posts: 97
Credit: 158,266
RAC: 0
Message 5992 - Posted: 5 Jun 2007, 23:24:38 UTC

If I recall correctly there was a post about the time limit on the forum somewhere and that it doesn\'t matter too much, it\'ll happily continue crunching away on it until it\'s finished. I\'m pretty sure it still counts it as valid unless it errors out or it\'s completed by other computers before yours is done, not too likely an occurrence when it\'s in the \'background rendering\' phase since it seems to only send requests for rendering one computer at a time.

Profile Zanthius
Project donor
Send message
Joined: 24 Mar 05
Posts: 94
Credit: 1,627,664
RAC: 0
Message 5993 - Posted: 6 Jun 2007, 1:17:33 UTC - in response to Message 5992.
Last modified: 6 Jun 2007, 1:23:17 UTC

If I recall correctly there was a post about the time limit on the forum somewhere and that it doesn\'t matter too much, it\'ll happily continue crunching away on it until it\'s finished. I\'m pretty sure it still counts it as valid unless it errors out or it\'s completed by other computers before yours is done, not too likely an occurrence when it\'s in the \'background rendering\' phase since it seems to only send requests for rendering one computer at a time.


I have seen one WU in this session that was completed by one or two computers, but all the others errored out, so the whole WU errored.

Might be an idea that if only one computer is able to render the part correctly, it should just keep trying until it get\'s the three, instead of just erroring out. (but that does say that there could be something wrong with the WU... I dont know)

Oh, and that other WU completed.... 121,138.17 seconds of work claimed credit: 506.42 -- mind you, that\'s on a quad core PC as well.
____________

Profile Janus
Volunteer moderator
Project administrator
Avatar
Send message
Joined: 16 Jun 04
Posts: 4467
Credit: 2,094,806
RAC: 0
Message 5997 - Posted: 6 Jun 2007, 6:21:36 UTC - in response to Message 5993.

121,138.17 seconds of work claimed credit: 506.42 -- mind you, that\'s on a quad core PC as well.

Wow! Now that\'s a lot. I\'m currently looking into the overhead/computation ratio in the case of splitting for 128 parts per frame, but to get 121k secs down to something useful would require an even bigger split.

csega
Send message
Joined: 8 Aug 06
Posts: 6
Credit: 141,511
RAC: 0
Message 6023 - Posted: 7 Jun 2007, 7:58:30 UTC - in response to Message 5997.

The solution for this problem is: checkpointing.

It works well by many projects with long-time WUs. If the developers can do this, they should do this (especially, if there will be even more long-time-wu sessions in the future). That\'s my opinion. What do you say?
____________

Profile DangerNerd
Project donor
Send message
Joined: 31 Mar 06
Posts: 126
Credit: 263,029
RAC: 35
Message 6025 - Posted: 7 Jun 2007, 14:24:57 UTC

Janus,

I just had multiple machines suddenly error on every result. Now they are at the 99 wu quota.

My guess is there was a batch of parts that are causing this, and once they pass, we will be computing again. If NO machines can finish them, however, this session is in trouble. :-(

Could you raise the threshhold? on this? Nobody is going to get too many of these at once due to their run time.

Thanks!
____________
Our Advice is to support all useful BOINC projects. Smart people needed to give advice to those who seek answers: Give or Get Free Advice Here

Profile Velociraptor
Send message
Joined: 17 Dec 06
Posts: 18
Credit: 8,400
RAC: 0
Message 6037 - Posted: 9 Jun 2007, 13:30:43 UTC

why is this Session still in sleep mode?

Profile DangerNerd
Project donor
Send message
Joined: 31 Mar 06
Posts: 126
Credit: 263,029
RAC: 35
Message 6038 - Posted: 9 Jun 2007, 14:51:07 UTC - in response to Message 6037.

why is this Session still in sleep mode?



Please read:

http://burp.boinc.dk/forum_thread.php?id=818

... and you will see that this session is likely to stay there. :-)

1 · 2 · 3 · 4 . . . 5 · Next
Post to thread

Message boards : Comments and discussion : 450