You're blocking ads, which pay for BlenderNation. Read about other ways to support us.

Free rendering on 250-CPU Grid

146

The Centre for Parallel Computing (CPC) at the University of Westminster, London are test-driving their 250-CPU Grid. It's built around Blender and they're inviting YOU to render your work there.

Zsolt Lichtenberger wrote:

We are the Centre for Parallel Computing (CPC) at the University of Westminster, London. As the name suggest, we are interested in parallel computing, particularly in computer grids. We would like to promote Blender on the Grid and we kindly ask for your help.

We created a simple rendering portal, which uses our 256 CPU AMD Opteron cluster. You can use it to render your Blender projects and we are offering this free of charge. Please give it a try!

We are looking for feedback because we wish to understand the needs of Blender users and to improve the portal.

Links

About Author

Bart Veldhuizen

I have a LONG history with Blender - I wrote some of the earliest Blender tutorials, worked for Not a Number and helped run the crowdfunding campaign that open sourced Blender (the first one on the internet!). I founded BlenderNation in 2006 and have been editing it every single day since then ;-) I also run the Blender Artists forum and I'm Head of Community at Sketchfab.

146 Comments

  1. @mav:
    i guess you don't want to see your files in public, because they're for commercial use. but if so, you hopefully have a budget to rent a proper renderfarm (or have your own).

    i think it's a great opportunity for hobby-artists to test a farm for free.

  2. mav, that's not true. Where did you get that idea? Rendering on this service does not release your work to the public domain. They stated very specifically on their homepage that they respect your copyright and do everything they can to keep their service secure.

    I tried it out, and it seems to work! After I submitted a job, it was running moments later and the render was complete very quickly. I'll definitely be using this service in the future.

  3. It would be good to see Yafaray supported. Although blender internal is not exactly fast it doesn't do GI and where render farms are an added benefit to ordinary desktop owners is that they can deliver amongst other things GI renders in a far shorter time frame and Yafaray, I am finding is a rather nice render engine. :-)

    Would it be possible to support yafaray?

  4. Yes, free render time is like "wow" and "WOW!!!" :)

    @Bart - how long do you think they will keep this free Blender rendering offer open? Some open projects would benefit from this resource. But even if it doesn't stay active for free use I think that as open source people we can, and probably will, choose to shovel the money out to a render farm when it's time to render our projects - that's what we do: find a way through every mountain. But I do hope it stays active. That would be sweet to render my stuff there. ;D

    Ciao!
    -b

  5. I did submit and render a job quite quick. Nice experience.

    However, now it always fails I think they got their wanted load and experience. I guess someone will fix it next week ;-)

    Carsten

  6. Dear All,

    First of all, thanks for the interest in the portal, we appreciate it!
    As you can see we had problems. Turns out it was just a silly little thing. It should be fixed now. I just tried and it works.
    However I'm afraid running jobs have to be resubmitted :(

    Check it out!

    @Yellow: If I get time allocated and you guys help me out with advice, sure. That's kind of why we started this.
    @BLENDER_BOY: I know I know... Can you give it a try now?
    @Orbit: I think you can render a single frame. But if you have a Core Duo or similar it will probably render faster on your desktop.
    @Lancer: I hope so.
    @Grzesiek: That is possible, yes.
    @jahmaica: Yes, Go give it a try!
    @banor: Obviously the dean has to see the benefit. That is where you guys come in: we would like to hear from you! We need feedback to improve it. Then we have to prove that the community actually needs this and it is worthwhile for the university to keep it running and for us to work on it.
    @Jj chilidz The user manual is under the help tab. I think we can prepare a technical paper as well. I'll ask.

    Zsolt

  7. @Zsolt:
    Thanks for your reply...I am still having the same error.
    I can connect to the site but shortly after I do it displays the following error message
    Look:
    http://i38.tinypic.com/15o6qmd.jpg

    I mean the website is reachable by my browser. There has to be something inside the portal itself that is causing the trouble.

  8. Well it worked great for me , very fast, but a few problems.

    One I dont know what version of blender its running, which isn't too much of a problem, but the by far the biggest problem , it renders an avi image sequence, and the avi movie wont play in quicktime. I'm on a Mac and can find no way of putting the sequence together, except manually, which sort of makes it useless, unless anyone can help me there.

  9. I have tried several times now.
    It seems to do its job just fine
    but than deletes all the files due to "exceeding user limit".
    Could you tell what these limits are exactly besides the 600 frames limit
    and the 100MB space limit
    (just for uploaded blends? or also for the rendered output?)
    also do you know if Rendering and compositing with FSA and the temp .exr files works?
    and which Blender Version is supported?

    But aside all those Questions
    Thank you very much for your kind offer.
    Maybe I get to render some of my stuff I stopped working on due to too large render times.

  10. Zsolt

    In case you get time to add Yafaray. :-)

    Is it Linux on your cluster?, if so what distro?, if precompiled binaries are not available then perhaps I could compile for you, so that it would just be the install across your grid with binaries and python scripts left for you to do?

  11. vlc may play it but I need a movie file rather than a sequence so I can use it in after effects or other compositing programes.

    I would need to put it all together by hand, which is very time consuming.

  12. @Jon: I think that with Perian, you should be able to read in .avi files as well. Haven't tried it, but I've had it on my system for years and I've always comfortably played and worked with both .avi and .mov files here.

  13. Thanks for the help bart, i've installed Perian and quicktime will play the online compressed avi online file you get with your render which is very poor quality, but will not play the image sequence.

    I dont what to sound like I'm complaining after all its free, I just frustrated, it took me 7 hours to render a sequence of a movie I'm making on my Mac and only 10 minutes in the render farm so I was really excited, but I've been searching all afternoon to be able to play the avi sequence and I cant find a way.

    If they're going to output a sequence wouldn't a jpg or png sequenece make more sense as most compositing programes can import these files and order them into a movie.

  14. I'd definately use this service. But I have a problem, I'm still a 3D Noob and don't know how to save all my external textures and materials in 1 single .blend file. I've tried it before and took my .blend file to my friend's computer and when I tried to render my scene almost everything came out black with no textures.

  15. Does the disk quota refer to the size of the output files, the .blend file, or both ? I have a 200mb file I'd like to animate, it's too big for my PC to handle.

  16. @all: our friend NextGens from CPC fixed the bug this morning. Say hello to him :)

    @BLENDER_BOY: I definitely want to sort this out for you. There is a redirection to https before the welcome page. Maybe a firewall issue or some security setting that kicks in. Have you tried going to https://rendering-portal.cpc.wmin.ac.uk:8080/gridsphere/gridsphere directly?
    @jon The avi is just a convenience, There is a zipped jpg sequence of course. Click on the green things in the output column on the manage rendering jobs tab! The manual is under the help tab. It's also accessible through a link on the welcome page without logging in.
    @marijn @Jonathan: Depends on you actually. The louder you are the easier it will be for us to convince those who make decision about our resources. So keep using it and let us hear your voice :)
    @jon, Bart: Yea, the video is dodgy. There was talk about setting up an rtsp server or at east tweaking the servlet to force the proper mime type. We'll discuss this issue on Monday.
    @testscreenings @Rhysy: blend only I think. What's your portal usernames? I'll increase your quota to 300MB for a while. The portal box has 160GB total at the moment but I think I can be bold enough to ask for a bigger disk tomorrow. Looks like Monday will be busy :)
    @yellow: Yes, The cluster runs SLES10 SP1. The portal machine is a separate box which I'd like to upgrade now that there is interest. I prefer Ubuntu, NextGens is a Debian guy, and CentOS is nice too... Yafray would have to be compiled for the cluster but that's no problem. I'll investigate it when I get the chance. Have to finish another project and get time allocated. That's how it works.
    @BigBoy: I guess this is a good place to ask. Unfortunately we at CPC know virtually nothing about using Blender.

    Also, please keep posting bugs and asking for enhancements! This is important for us to improve the portal!

  17. Wow, this is too good to be true. I'm going to be using this for my CgTalk entry. This is awesome. I'm off to see if I can get something running.

  18. Off the top of my head:

    * When a job is submitted, display its position in the que.
    * Add an option to receive an email when job is completed.

  19. Okay, I've submitted two jobs, (1) a 120 frame animation, (2) a 1600x1040 image.

    Both failed with this message:

    "Submitting workflow to DAGMan failed, error message of command "condor_submit_dag BlueBalls.dag" follows:
    ERROR: failed to transfer executable file /home/pgportal/pgportal/condor/bin/condor_dagman"

  20. @Alex: our Blender is version 2.45. Is that helpful?
    @Twinsnakes: I have added the feature requests. Submission fails at the moment. We are working on it.
    @testscreenings @Rhysy: Your disk quota should be 300MB now

  21. Many thanks Zsolt. I actually managed to get the file size down to 65 mb - I did try this before but it seems Windows wasn't reporting the file size correctly. Can't wait to try this out !

  22. Bender 2.45 is not the latest (2.48a is what everyone is using ATM)

    Also, fluid and cloth simulation animations can not be rendered because the baked movement data of the fluid/cloth is not stored in the .blend file itself, but it is stored separately in the /temp folder.
    This means you need to add the feature to bake the fluid/cloth before rendering, otherwise it will be motionless frames and nobody can use it for that purpose.

    Thanks for your support.

  23. Thanks for the reply!
    However the project i wanted to render with particles was made in version 2.48.

    maybe you guys could upgrade to 2.48 also?

    Anyway, still a great service!!

  24. Fair enough; I've just upgraded the version of blender we use to 2.48a.
    I hope I haven't created too much disruption on running jobs doing so :/

    Enjoy!

  25. Thanks for the answer Zsolt,
    I got an avi sequence, not a jpg seqeunce the first time I tried, my next render produced no output file and no error messages.

    hope that helps.

  26. I submitted OK, but I get the error message :
    zip warning: name not matched: frames

    zip error: Nothing to do! (try: zip -1 -r frames.zip . -i frames)
    When I hit visualise it looks like it's still doing something, though it says finished in the main window. Should I just try again ? I know the .blend is fine.

  27. @jon I don't really know Blender but there seems to be a setting that creates jpg. We got this avi sequence with a few public .blend files we tried, and jpg with others. Maybe somebody here can help you out.
    @endi: Which version of Blender are you using?
    @Rhysy: Please try again, We had to restart the portal and that messes up running jobs. Hopefully it will work. If not I'll look into it tomorrow.
    @Twinsnakes: Great! Cold you share a rough performance comparison?

  28. I get the following error in the logs:

    =>> PBS: job killed: walltime 302 exceeded limit 300

    Running version 2.48a... a 7M file... I am able to render at half the size on my laptop.

  29. Props to the portal progenitors, persevering, with perspicacity!

    Just tried it again, with a file that failed on the weekend. I realised it had a cloth modifier on it, so I exported the cloth mesh as a point cache (.pc2), removed the cloth modifier, re imported the point cache and re saved. It worked fine when I uploaded it again. So, that's one workaround if you want to render cloth. It's pretty damn fast! I hope this service stays around.

  30. ATM the management Tab only shows a help message for me. So I cant check my tasks.

    I did subscribe to the mailing list, I think we should not use this comments for error messages.

    Carsten

  31. Ermm, erm. I TRY to subscribe...

    - Results:
    Invalid confirmation string. Note that confirmation strings expire
    approximately 3 days after the initial subscription request. If your
    confirmation has expired, please try to re-submit your original request or
    message.

    Carsten

  32. A few frames rendered fine for me, everything looks as it should. Unfortunately I get the error message "122 frame(s) killed due to exceeding user limits" - not sure what those limits are !

  33. @Zsolt:
    First, thanks for offering this service and putting time and resources into it and Blender!

    Second: I was able to successfully submit and get results for two rendering jobs:
    however: my results come out as a bunch of .avi files ( I guess I will check my blender file and see if changing the default output to something like png or jpg yields different results).

    Third: If at all possible, support for fluids would be awesome! I'm guessing that there are other things as well that require baking that might not work either but I'd think others here would be able to point to those better than I. I've been trying to find a nice distributed renderer that would handle fluids for a while and haven't found anything simple enough ( I think I had it working once with Yadra but I think I just copied all the fluidsim data to the /tmp directory where each node was expecting the data to be but I don't remember for sure)

    thanks, again!!

  34. Dear All,

    We spent today on fixing some portal related issues on the cluster and eliminated a few bugs the portal code.
    As you know job submission is broken. The service is quite popular and the disk on the portal machine is full. We had a discussion and decided to allow you to use our cluster filesystem. This means 2 things: up to 11TB of storage and 1.2 Gb/s throughput. :)
    We had to take the portal down to migrate it to a cluster node. Since it was unusable anyway we decided to take it down without an announcement. This leaves a bad taste in my mouth but it will make migration much faster. Sorry for the inconvenience! We are working on it right now.

    For all those who I have yet to reply to: I'll do my best to reply as soon as possible but getting the portal up again has to take priority!

    Zsolt

  35. Well with all that processing power, isn't it a shame that blender is still single threaded when baking? My first thought was lets crank up the resolution insanely high and do some awesome water shots - baking is still single threaded and so it would still take a few days just to bake.

  36. I guess you could break down a large complex fluid simulation into a number of jobs each with only tens of frames. Then have the system's job management and LSF parcel the jobs out to individual cluster nodes (or node pools, depending on the cluster).

    I strikes me the Blender community could use a grid application similar to SETI's [email protected] BOINC application. I'd be happy to share my spare CPU cycles with fellow blender. (Yeah - right like we have spare CPU cycles.....)

  37. uploaded a fitness room full of mirrors let see how the britsh cope with that lol, I dont see any indication of how long it will take? I better check tomorrow.

  38. Hi Zsolt,
    I'm getting the same problem as "testscreenings", when my render finish i get this error: "X frame(s) killed due to exceeding user limits". What does it mean? I uploaded a 49Mb file, with 560 frames.
    Thanks for the help, and for this great project of you =)

  39. Icedeb, I think the frame limit is supposed to be 500. I've just realised I've been trying to render 501 frames, not 500, which is probably why my animation keeps failing. :)

  40. @Rhysy: but i tryed to render only 20 frames too, and got the same problem =\
    the idea of this project is great, but it will take some time to make it really good and have no bugs.
    **(sorry for the bad english ^^)

  41. Dear All,

    We are up and running again. But I'm sure you know that already. We didn't even have to announce anything, you rushed to the portal when it was restarted :)

    The portal was moved to a new machine last night and uses the cluster filesystem now. This improved job submission considerably. We've updated the welcome page with useful information.
    Workflow visualization is broken. Fixing it is a low priority at the moment.
    Much more important is the quota management. We are working on fixing this issue.

    @Icedeb: I have added useful information on the welcome page.
    @ysvry, @Icedeb: The walltime is 5 minutes. My guess is that a it takes too long to render, hits the wall and fails. Also, make sure it is possible to render the number of frames you want from your .blend file.
    @Rhysy: Your frame limit is also the default end frame you can see on the 'New rendering job' tab. This should be 600.

    Zsolt

  42. @ elwol , better make a multiplayer game that plays in blender and let the player share their free cpu cycles. something along the lines of the sims in space lol.

  43. Awesome, it worked ! I think I was probably hitting the 5 minute per render wall, it now works perfeclty. This probably would have taken 12 hours or more on my own PC, it was done in less than 2 thanks to this (I went away for a while so I couldn't time it exactly).
    Many thanks !

  44. I think this is great too can this be inbuild in blender so people allowing the feature will automatically form a render matrix? that would be uber cool.
    thx for the good work will this stay free or will it go comercial? and what will the prizes then be?

  45. another question could you make the wall time inversley related to the number of rames you render so say if you render 300 frames you can have a frame render for 10 min instead of 5 for the 600. Think it will result in the same processor usage per person, right?

  46. @Zsolt: thanks for the help. But how can i know if it will take more or less then 5 minutes? I tried to render a file with 1 frame and got the same problem...

  47. Hi Zsolt, i changed my file a little to make it faster to render. But now i get this error:
    cat: /tmp/tmp.rHvHas5956/*.stderr: No such file or directory
    zip warning: name not matched: frames
    zip error: Nothing to do! (try: zip -1 -r frames.zip . -i frames
    I don't know what this mean, Is it because of my file again? If it is, how can i fix it?

    Thx again =)

  48. Same here....

    ERR log of: SchPAL600 / Frames_451-600

    cat: /tmp/tmp.DYLUF24510/*.stderr: No such file or directory
    zip warning: name not matched: frames
    zip error: Nothing to do! (try: zip -1 -r frames.zip . -i frames)

    Carsten

  49. @Rhysy - Dec 1st, 2008 at 6:35 pm

    Maybe I'm mistaken but at least on my system it appears that Fluid Baking is multi-threaded.
    I'm testing using 2.48a 64bit on SuSE Enterprise Desktop Linux10 SP 2, Core 2 Quad Q9550 2.83.
    When I'm baking fluids, top shows that blender is taking upwards of 160% Cpu where no individual core seems to actually hit 100% but it seems to be spread over all 4 cores.

  50. Dear All,

    First, thanks for your patience! I'm struggling a bit to reply to all messages and to work on improvements at the same time.

    I'll start with a quick status update: last night the job scheduler stopped working due to a glitch. The load slowly wore off during the night as no new jobs could be processed and 9000 frames accumulated in the queue by 1pm today. We managed to solve the problem and frames are rendering now. Sorry for the inconvenience. I know the service is slow at the moment. We're working on it.

    @Icedeb @Carsten: We moved the /tmp (on the fly), that is why a few of you might see such errors. This should not happen any more after you resubmit the jobs.
    @ysvry: Yes, definitely a good idea, but all feature enhancements have to wait until we resolve the problems which can prevent you from using the service.

    This service is new and experimental, so, as the Welcome page says, please don't rely on its availability. We run into bugs and have to restart it from time to time. It will be much better once we've ironed out the wrinkles.
    Please also note that we are sharing the cluster, and it's under heavy load at the moment. You can check the load by clicking on 'Cluster Load Percentages' on the welcome page.

    So please bear with us!

    Zsolt

  51. @Giovanni: Could you please delete the workflow and create a new one? Chances are it got messed up while we were trying to fix the job queue. Thanks!

  52. Great service! Thanks!

    @Tomas: no, fluid baking is single threaded; I have a load-average of 1.9 on my Opteron dual-core, one processor at 100% the other around 60 (after a couple of seconds the other way round). That's simply one processor doing the physics, the other handling the output (disk, etc.).

    By the way, is there a way to do the physics baking from the command-line (i.e. without the GUI)? I somehow cannot find it, but there should be a possibility...

    Thanks!

  53. hey I get following error:

    Copying executable to resource failed!
    The failed command was: "globus-url-copy file:///home/pgportal/pgportal/portal_work/users/gizmo#/Firstfinal_files/Frames_1-92/submit.sh gsiftp://ngs.wmin.ac.uk/gpfs/rendering_portal/.grade-portal/Firstfinal_905cd8d0-5f8b-4cfa-9e85-dd2476004a93/Frames_1-92/submit.sh"
    - stdout begin -
    - stdout end -
    - stderr begin -
    error: [globus_gass_copy_get_url_mode]: globus_url_parse returned error code: -8 for url: file:///home/pgportal/pgportal/portal_work/users/gizmo#/Firstfinal_files/Frames_1-92/submit.sh
    - stderr end -

    I'm really new to blender and the render portal.. So i don't know really what I'm doing wrong.

    Thankfull for any tips...

  54. I know it experimental and dont mind hickups.
    aother thing not working here is the video button , it opens the movie player but then says end of file reached without showing a thing.
    could you inform me when the 5 min per frame wall is lifted?

  55. Dear All,

    Please tell us how much walltime you would like for each frame!

    If the rendering takes longer than the walltime (5 minutes at the moment) the frame is thrown out. Each frame is rendered on 2 dual core AMD Opteron 2218 CPUs (2.6GHz, 4 cores total) with 4GB memory.
    The walltime is essential for scheduling and to guarantee fair share as well as to protect against jobs caught in an infinite loop. We are working on our scheduling policy to fit all our users better (not just rendering users) and everybody has to specify a walltime for their jobs.

    @Giovanni: Hmm. Even if the cluster is under load and the frames are queuing the job status should be "running". This is what I see right now when I submit a rendering workflow, so I don't know. What's your portal username again?
    @Yeju: Your username confuses the URL parser of globus-url-copy (we use this to transfer files to the cluster). Create a new account and do not use whitespace or special characters (here "#" is the problem). Thanks! I'll implement a filter for this later.
    @TwinSnakes: I'm happy with the BlenderNation forum, it gets the job done and I can't manage two. Sorry.
    @ysvry: Please read the portal Welcome page, it's a known issue, video is disabled. We'll fix this one soon. The walltime can never be lifted but it could be increased. Please let me know how much you had in mind.

    Cheers,
    Zsolt

  56. Thx Zsolt!
    Now I can start the render job but somehow it only shows up 8 kb for the file size..
    Couldn't be right, could be?
    And it didn't come to an end..
    feel like one big noob.. hahhahah
    anyway. thanks for your service!

  57. Jennifer Blender on

    I have trouble downloading the finished render. If I click video I get a blank Quicktime player with nothing to play.
    If I click output, I get a zip file with multiple avi frames. What to do with that. I tried to google all this with no luck.
    Any help would be appreciated.

    Thanks for a great rendering service. I'm just not sure how to use the output.

  58. Jennifer Blender on

    Okay, I got Perian codec for Quicktime and it can now be downloaded, but it is just one frame.
    0K file. It does not seem to be rendering correctly.

    Sorry.

  59. Jennifer Blender on

    So I got to work on my XP machine. The file downloads to a 64kb file. Play it with VLC and it is a still shot. Not a 250 frame animation.

    Thanks

  60. i want as high a wall time as possible, what i want to do is try to put render settings as high as posssible and see what the effects are quality wise. Maybe its possible to send the screens how they look at the moment they hit the wall and not black screens. I can then evaluate the outcome and scale down some settings to avoid hitting the wall.
    Also is it not possible to have more clusterswork at the same frame to increase the speed? Im not so much intrested in big animations as in faster rendering with high settings. Mayb you can contact Ton to ask how you could spread frame rendering over more computers. I know the frames are rendered in seperate blocks so theoretically that should be possible.

  61. @ysvry yes, technically you could split your render into, say, 32 chunks, and have 8 different nodes render 4 of those chunks simultaneously. I think what you would find overall though is the cost of loading that image into memory off the HDDs 8 different times might, depending on what kind of an animation you're doing, not be so terrifically wonderful in the long run. Plus, different chunks of that frame would render at different speeds, etc.

    Rendering the entire project using just 1 CPU will give you the best PER CPU performance. You could split your project among all the CPUs Westminister has simultaneously, and get the fastest end-result render. However, your performance PER CPU would take a hit. Because other people are using this cluster, I'm pretty sure they would want to limit idle time on the CPUs as much as possible. Blah blah I don't know if this will even come out coherent.

  62. I'm getting a strange error : cat: /tmp/tmp.PlBQWS6147/*.stderr: No such file or directory
    313 of my 600 frames have rendered correctly but the rest haven't rendered at all. I'll try re-submitting.

    Personally I would use this service to render long animations I couldn't otherwise do. Given the high specs on the machines 5 minutes per frame seems pretty good to me as that's still much faster per frame than my home PC. Someone mentioned the idea of weighting the maximum time by the number of frames required, this seems like a good idea, especially for stills.

  63. No luck, I get the same error but this time hardly any frames have rendered. The only thing I changed was the number of x and yparts from 4x4 previously to 1 this time.

  64. Jennifer Blender on

    Yes, I set the start and end frame from 1 to 250.
    I render it on my machine and it renders fine.
    Just not rendering at all on the portal.

  65. Same error with my files...I think we just have to wait...looks like it isn't so easy to setup such a portal ;-)

    But keep on the good work...we all really appreciate this!!

  66. I give up.

    I reuploaded, re-submitted, re, re, re etc... ;-)

    cat: /tmp/tmp.WASgRt7956/*.stderr: No such file or directory

    I was really keen to try all, and feel how it is to have such service at my hands.

    However after the first tests all stopped to work, ok it is experimental, no commercial service, no problem for me!

    BUT: This "Forum" is not well suited for support, the email-link above would be much nicer however I got a mail back, telling me that this is only for Portal users (is me not using the portal?!) and I should use this comment "Forum"....

    Sorry guys, I really liked to help and give comments further than this rant.

    Carsten

  67. @Carsten: Sorry it didn't work for you. I think the error message you get is only possible if all your jobs got cancelled. This could have happened either because they hit your walltime of 5 minutes per frame or they needed more than 3.8GB of memory each. We call this heap of comments our "forum" simply because people started using it like that.

  68. Hi again =)
    I sent a very simple file this time, and it worked, no error messages this time. The only complex thing was some particles falling to the ground. But....... They didn't appear in the render =\ But i think the problem is because my blender version is 2.48, and yours is 2.45. I tried to run this file here, in my pc, using the 2.45 version, and it didn't work too.
    Maybe this is the reason for my other files also didn't render. ^^

  69. Hi,

    I submitted a job several times - all ended with the same result:
    Not a blender file, aborting ..
    Job exit code isn't 0!

    Yes, I deleted the jobs and submitted them again.

    Could the reason be, that I work on a 64bit machine?

    I have no further idea how I should get the thing running...

    Harald

  70. I think, there's something totally screwed up ;-)

    My jobs doen't render all frames...some are skipped...like 20/150, 24/150, 26/250, etc.

    If I look in the logs, there are more frames rendered as stored in the zip...

    When will it be running again? ;-)

    Cheers,
    Michael

  71. I have good news and results for Zsolt and the guys behind this grid :)
    I managed to render a 600 frame animation at 800x600 resolution correctly in less than 10 minutes!
    If I have rendered it with my computer, it would have taken about 12 hours of rendering time because each frame renders in about 1 minute and 12 seconds on my machine. On the grid it only took 10 minutes to render the entire animation!

    Awesome!
    Thank you guys :)
    keep it up please, the community is in dire need for such a service.
    (tell your dean about my story :D )

  72. I'm having problems uploading a 77 mb .blend file - is there a cap on loading or idle login time ? I've tried using IE and Google Chrome, both display a blank page after about 45 minutes. I've cleared my other jobs so it shouldn't be a quota issue.
    Thanks,
    Rhysy

  73. Another try - same result:
    Not a blender file, aborting ..
    Job exit code isn't 0!

    I created a new scene with a rotating cube, saved it with blender 2.46 and in another file with 2.48.
    Renders pretty on my machine.
    Submitted both files to the grid and ended in that error message.

    Frutstrating.

    Thanks,
    Harald

  74. I submitted a smaller file with no problems. It rendered half the frames fine in about 10 minutes (would take 5 hours on my PC !) and they look fine, but as in a previous attempt, half the frames haven't rendered, with no particular pattern to the sequence. No error messages at all this time.

  75. Dear All,

    It's been a while, I know. Sorry for not posting, I've been really busy. We had many meetings this week and one of them was about the Rendering portal. We made a user and a developer wish list as well as a crude list of bugs to see what we should work on. We have limited time of course so we can't deal with everything all at once. From our point of view the first thing that has to be fixed is the issue with the error logs. You do read those error logs do you not? Many of you run into the walltime (you have 5 minutes to render a frame at the moment and if it takes longer it gets killed) or exhaust memory (about 4GB per frame). In this case your frames are lost and unfortunately the error messages in the logs are... not very informative, sometimes they mention zip, pbs and whatnot. So we'll improve that and hopefully you'll be able to adjust your settings to avoid running out of time and memory and your frames will render. Then we'll create some sort of FAQ to help you better understand the error messages. Anything else comes after that.

    @Icedeb: Where did you see that our Blender is 2.45? I would have to fix that because it should be 2.48a. Says so on the welcome page.
    @Linoprit: NextGens has an idea why this is happening and if he's right we may be able to fix it. Are you compressing your .blend? If so try without compression.
    @BLENDER BOY: Thanks :) they read this thread
    @Rhysy: Yes, the portal will log you out after a while. And, I'm pretty sure your lost frames took more than 5 minutes to render or Blender consumed more that 4GB memory. Please submit it again and check out the log files in the zip.

    Zsolt

  76. Cheers Zsolt. I've tried a few times, but neither the 5 minutes walltime or 4GB limit should be an issue - I can render in 1-2 minutes on my dual-core PC with 1GB RAM (it's just a simple flyby animation, so there shouldn't be any major differences in rendertime between frames, certainly not memory usage).
    I've checked the log files, but they're.. umm... not all that helpful... here's one :

    *** glibc detected *** /opt/modules/Modules/3.2.5/bin/modulecmd: free(): invalid next size (fast): 0x08095ed0 ***
    ======= Backtrace: =========
    /lib/libc.so.6[0xf7d59cb1]
    /lib/libc.so.6(__libc_free+0x84)[0xf7d5b314]
    /usr/lib/libtcl8.4.so(TclpFree+0x1d)[0xf7e6fedd]
    /usr/lib/libtcl8.4.so(Tcl_Free+0x1d)[0xf7e770dd]
    /usr/lib/libtcl8.4.so(Tcl_DeleteHashEntry+0x7d)[0xf7ea6e6d]
    /usr/lib/libtcl8.4.so(Tcl_DeleteCommandFromToken+0x5f)[0xf7e7282f]
    /usr/lib/libtcl8.4.so(TclTeardownNamespace+0x163)[0xf7ebda03]
    /usr/lib/libtcl8.4.so[0xf7e70ade]
    /usr/lib/libtcl8.4.so(Tcl_EventuallyFree+0x8a)[0xf7ec9f8a]
    /usr/lib/libtcl8.4.so(Tcl_DeleteInterp+0x3a)[0xf7e73464]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x804d2ed]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x8051cdd]
    /usr/lib/libtcl8.4.so(TclInvokeStringCommand+0x69)[0xf7e70349]
    /usr/lib/libtcl8.4.so(TclEvalObjvInternal+0xfa)[0xf7e7149a]
    /usr/lib/libtcl8.4.so(Tcl_EvalEx+0x1ac)[0xf7e719ec]
    /usr/lib/libtcl8.4.so(Tcl_Eval+0x36)[0xf7e703e6]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x8052d15]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x8052b38]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x804d218]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x8051cdd]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x805dc2a]
    /lib/libc.so.6(__libc_start_main+0xdc)[0xf7d0b87c]
    /opt/modules/Modules/3.2.5/bin/modulecmd[0x8049a91]
    ======= Memory map: ========
    08048000-08063000 r-xp 00000000 08:02 81137 /opt/modules/Modules/3.2.5/bin/modulecmd
    08063000-08065000 rwxp 0001b000 08:02 81137 /opt/modules/Modules/3.2.5/bin/modulecmd
    08065000-081b5000 rwxp 08065000 00:00 0 [heap]
    f7b00000-f7b21000 rwxp f7b00000 00:00 0
    f7b21000-f7c00000 ---p f7b21000 00:00 0
    f7c98000-f7ca2000 r-xp 00000000 08:02 14577 /lib/libgcc_s.so.1
    f7ca2000-f7ca3000 rwxp 00009000 08:02 14577 /lib/libgcc_s.so.1
    f7cb4000-f7cf6000 rwxp f7cb4000 00:00 0
    f7cf6000-f7e11000 r-xp 00000000 08:02 13542 /lib/libc-2.4.so
    f7e11000-f7e13000 r-xp 0011a000 08:02 13542 /lib/libc-2.4.so
    f7e13000-f7e15000 rwxp 0011c000 08:02 13542 /lib/libc-2.4.so
    f7e15000-f7e18000 rwxp f7e15000 00:00 0
    f7e18000-f7e3b000 r-xp 00000000 08:02 13550 /lib/libm-2.4.so
    f7e3b000-f7e3d000 rwxp 00022000 08:02 13550 /lib/libm-2.4.so
    f7e3d000-f7e3f000 r-xp 00000000 08:02 13548 /lib/libdl-2.4.so
    f7e3f000-f7e41000 rwxp 00001000 08:02 13548 /lib/libdl-2.4.so
    f7e52000-f7ef5000 r-xp 00000000 08:02 14678 /usr/lib/libtcl8.4.so
    f7ef5000-f7efb000 rwxp 000a2000 08:02 14678 /usr/lib/libtcl8.4.so
    f7efb000-f7efc000 rwxp f7efb000 00:00 0
    f7efc000-f7f16000 r-xp 00000000 08:02 13535 /lib/ld-2.4.so
    f7f16000-f7f18000 rwxp 0001a000 08:02 13535 /lib/ld-2.4.so
    ff895000-ff8a0000 rw-p ff895000 00:00 0 [stack]
    ffffe000-fffff000 r-xp ffffe000 00:00 0

    Scary stuff !

  77. Dear all,

    I have just fixed two issues on the portal:
    - Compressed blender files should now be recognized properly
    - Rhysy's issue shouldn't occur anymore

    I suggest to anyone who was experiencing problems to give the portal an other try.

    Of course, please keep me posted shall any other issue arise.

  78. Thanks NextGens. I don't think the problem is quite fixed, this time I got 215/300 frames to render. There is at least an error file, it says :

    201:ngs.wmin.ac.uk
    1202:ngs.wmin.ac.uk
    1203:ngs.wmin.ac.uk
    1204:ngs.wmin.ac.uk
    1205:ngs.wmin.ac.uk
    1206:ngs.wmin.ac.uk
    1207:ngs.wmin.ac.uk
    1208:ngs.wmin.ac.uk
    1209:ngs.wmin.ac.uk
    1210:ngs.wmin.ac.uk
    1211:ngs.wmin.ac.uk
    1212:ngs.wmin.ac.uk
    1213:ngs.wmin.ac.uk
    1214:ngs.wmin.ac.uk
    1215:ngs.wmin.ac.uk
    1216:ngs.wmin.ac.uk
    1217:ngs.wmin.ac.uk
    1218:ngs.wmin.ac.uk
    1219:ngs.wmin.ac.uk
    1220:ngs.wmin.ac.uk
    1221:ngs.wmin.ac.uk
    1222:ngs.wmin.ac.uk
    1223:ngs.wmin.ac.uk
    1224:ngs.wmin.ac.uk
    1225:
    1226:ngs.wmin.ac.uk
    1227:ngs.wmin.ac.uk
    1228:ngs.wmin.ac.uk
    1229:ngs.wmin.ac.uk
    1230:ngs.wmin.ac.uk
    1231:ngs.wmin.ac.uk
    1232:ngs.wmin.ac.uk
    1233:ngs.wmin.ac.uk
    1234:ngs.wmin.ac.uk
    1235:ngs.wmin.ac.uk
    1236:ngs.wmin.ac.uk
    1237:ngs.wmin.ac.uk
    1238:ngs.wmin.ac.uk
    1239:ngs.wmin.ac.uk
    1240:ngs.wmin.ac.uk
    1241:ngs.wmin.ac.uk
    1242:ngs.wmin.ac.uk
    1243:ngs.wmin.ac.uk
    1244:ngs.wmin.ac.uk
    1245:ngs.wmin.ac.uk
    1246:ngs.wmin.ac.uk
    1247:ngs.wmin.ac.uk
    1248:ngs.wmin.ac.uk
    1249:ngs.wmin.ac.uk
    1250:ngs.wmin.ac.uk
    1251:ngs.wmin.ac.uk
    1252:ngs.wmin.ac.uk
    1253:ngs.wmin.ac.uk
    1254:ngs.wmin.ac.uk
    1255:ngs.wmin.ac.uk
    1256:ngs.wmin.ac.uk
    1257:ngs.wmin.ac.uk
    1258:ngs.wmin.ac.uk
    1259:ngs.wmin.ac.uk
    1260:ngs.wmin.ac.uk
    1261:ngs.wmin.ac.uk
    1262:ngs.wmin.ac.uk
    1263:ngs.wmin.ac.uk
    1264:ngs.wmin.ac.uk
    1265:ngs.wmin.ac.uk
    1266:ngs.wmin.ac.uk
    1267:ngs.wmin.ac.uk
    1268:ngs.wmin.ac.uk
    1269:ngs.wmin.ac.uk
    1270:ngs.wmin.ac.uk
    1271:ngs.wmin.ac.uk
    1272:ngs.wmin.ac.uk
    1273:ngs.wmin.ac.uk
    1274:ngs.wmin.ac.uk
    1275:ngs.wmin.ac.uk

  79. I've had a go with another file, same sort of problem I'm afraid (I'll try the original file again to check though). This time only 44 frames rendered out of 402. Again those frames seem to have rendered correctly. There is no error file in the output .zip, but there is the error in the log viewer :

    zip warning: name not matched: frames

    zip error: Nothing to do! (try: zip -1 -r frames.zip . -i frames)

  80. It looks like the problem is fixed - I tried again and this time it works perfectly. It could be this second file was hitting the walltime - I reduced the size of some of the texures, lowered the octree and increased the pixel cache for AAO. This seemed to make a nice speedup on my PC (from 6 min to 2). Anywho this portal has just increased rendering speed by a factor of 60, it would have taken 1000 minutes on my PC but actually took about 15 ! Awesome !

  81. Hi, Zsolt,
    sorry, but you said that the version was the 2.45, look:

    "Zsolt Edit Link Nov 30th, 2008 at 1:47 pm

    @Alex: our Blender is version 2.45. Is that helpful?
    @Twinsnakes: I have added the feature requests. Submission fails at the moment. We are working on it.
    @testscreenings @Rhysy: Your disk quota should be 300MB now"

    So, when the particles didn't appear in the final render, i though it was because of the diference of the versions... Sorry, but i had not seen that on the front page is written the version when i wrote my comment.
    So... Why the particles don't appear?

  82. Hello, can somebody help me?
    I've been tring to render something but it doesn't work. First of all any .blend file I upload, it always appear in the size column as an 8 kb file.
    Could it be a problem of my pc?

  83. Hello Zsolt,

    yes, my files have been compressed. Tried without copmressing and BONZAI - it worked.
    The testrender of an animated cube was rendered fine.

    Next I tried my testcube with a compressed blender file - worked too.

    Thank you very much for your help.
    Hope the portal stays open. Such a plattform was exacty what I missed.

    Harald

  84. Me again... :)
    I just tried rendering a short animation with particles (dupliverts actually), all set up in 2.48. It renders fine on my PC, but they don't appear when using the portal.

    Icedeb, I think the version was changed to 2.48a, so probably the particle problem isn't a version issue.

  85. 1. Fonts don't render as displayed, I am going to try convert the fonts to a curve..
    2. I am getting the following error...
    No Permission.
    qsub: cannot connect to server portal018 (errno=15007)
    3. I can't get Avi's to work... looks like maybe I need a special codec

  86. Dear All,

    Happy New Year! We're back in business.

    @Rhysy, OriginalBBB: Yes, we have problems with particles. We don't know why. Any ideas? Could it be Python? (we have 2.4)
    @rwgreen1173: 1: We probably don't have the font you are using. Sorry. Our font set is limited to free fonts. 2: This one has been solved. 3: Scroll up a little, there's been a few posts about getting the AVIs to work.

  87. @Zsolt:

    I think I know why you are having trouble with particles. When blender calculates particles (and the same for cloth and softbodies, if i'm not mistaken) it puts all the baked data into an external file. Since the users are only asked to upload the blend files and not their blendcache folder where the particle data is stored. But even if you ask the users to upload the particle data you have the task of distributing the right sim data files to the right frames.

  88. Hello. Thanks for this service! I'm trying to do particles too. It seems when I do hair or particles that cant be baked, it works, but when PS needs baking, it does not work. Is there a workaround for this, like a way to convert particles to objects with animation? (CTRL-SHIFT-A only makes parts real but looses all anim..)

    Cheers!
    Mkn.

  89. Hi Everybody,

    The project is not dead at all, it's just that most of the communication happens by email. I'll be more active in the forums from now on providing updates. I hope you'll be interested.

    We do have technical difficulties every now and then but we do our best to keep the service running. Let me share some statistics: you have submitted over 8800 Blender jobs generating more than 463000 frames totaling 180 GB in image files over the last 6 months. We've donated over 56000 CPU hours so far. The busiest month was July when you rendered 102000 frames.

    We've spent some time experimenting with the portal to see how we could get more processing power. Check back soon for announcements!

    Cheers,
    Zsolt

  90. Actually, here is a fresh announcement:

    Enhanced rendering services for our academic users - we need your help

    While keeping the operation of the current portal intact our team at the University of Westminster is working on a new rendering portal that will provide more advanced services for our academic users. This new version of the portal will allow access to large Grid infrastructures with much more processing power than what the current solution can provide. And it will still be free! In order to achieve our goal please help us to define the requirements of this portal. If you work at a university or research institute and are interested in using the new rendering portal
    for your teaching or research then please contact us on the e-mail address below: [email protected]

  91. Hello,

    Here's another announcement on behalf of the whole team at Westminster:

    Dear Rendering Portal Users,

    The portal developer team at the University of Westminster would like to
    thank all of you who tried and gave valuable feedback about our
    Rendering Portal Service during the past year. We have learned a lot
    from this feedback and have decided to make significant enhancements to
    the portal.

    Therefore, we regret to announce that the Rendering Portal service will
    be temporarily shut down for upgrades and maintenance. However, we will
    come back with an even better free service in the New Year.

    While no new jobs can be submitted right now you can still log in and download your files.

    Please do not hesitate to contact us if you have any comments,
    suggestions or questions. Our email address at
    [email protected] will still be active during the downtime.

    Hope to see you all on our enhanced service soon!

    The Rendering Portal Team

Leave A Reply

To add a profile picture to your message, register your email address with Gravatar.com. To protect your email address, create an account on BlenderNation and log in when posting a message.