by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
Martin
@acidhunter
But the moment that i got with that first 16k rendering from this cluster teached me that i have to love that bits
My first cluster rendering was done on a hackintosh,
That hackintosh was build by hacking uefi modules of my x79 mainboard, whole time on acid. After that we build that terabit cluster we decide to take even more acid (sry 4 that, u should know who i am)
Krzysztof Marczak
@buddhi1980
:)
Martin
@acidhunter
2 years later i upgraded from a lowend job to an advanced linux administrator
Without mabdelbulber, FOSS and my awsum friends that never been happend
Now we call it #BSDM
its liks bdsm but licensed with BSD
sry for the wall of text
Sebastian Jennen
@zebastian
latest commit fixes amd on ubuntu for me: buddhi1980/mandelbulber2@493e138
lets see if we can get this fixed with some spaghetti code...
Krzysztof Marczak
@buddhi1980
Does it work with spaces in directory name?
Now I'm updating cross compiler scrips (to ne folder locations and Qt 5.9). Tomorrow I can test it under Windows 10
Of course with nVidia. Then I will send exe file (or you cam use fresh cross-compiler scrips)
Sebastian Jennen
@zebastian
in an our or so appveyor will also have a fresh windows build, which i can test against amd.
Martin
@acidhunter
I got problems while building on ubuntu zesty
qt ui tools are unavaible 0o
I hope to provide cross compile pipelines at this weekend
Sebastian Jennen
@zebastian
just tested latest appveyor build: https://ci.appveyor.com/project/buddhi1980/mandelbulber2/build/job/kwdrvmsa11m9xhyu/artifacts
works with win10 / AMD card with and without spaces in path
tested with radean windows driver 17.1.1
Martin
@acidhunter
we should set travis to only build from master branch
image.png
Krzysztof Marczak
@buddhi1980
That's great it works on AMD properly. But now doens't work on nVidia with spaces. We are very close to the finall solution. I will correct the code to work again with nVidia with keeping code for AMD
Krzysztof Marczak
@buddhi1980
Code corrected. I have tested it with nVidia and now works properly. @zebastian , try again with AMD. I hope it's finall version.
Exe is in Dropbox/Sebastian-Krzysztof-Graeme-Robert/mandelbulber opencl for testing/
mclarekin
@mclarekin
Same include headers problem with win10 AMD
Krzysztof Marczak
@buddhi1980
Is this possible that @mclarekin and @zebastian use different drivers? Maybe older cards use another driver. @mclarekin do you use Windows built-in driver or use downloaded from AMD website?
Sebastian Jennen
@zebastian
@acidhunter these build bots on gitlab run like a breeze, pretty cool: https://git.kmpt.nz/mbulb/mandelbulber2/-/jobs/328
Sebastian Jennen
@zebastian
hmm, could not find a proper location on ebay to issue a report, so i just used the "report item" where you can not type any text on the topic.
Krzysztof Marczak
@buddhi1980
In next days I'm going to release 2.12 alpha7 - with DOF effect. Now I'm doing long run test to catch posisble problems (stability, memory leaks).
After that I will add OpenCL support for mesh export. This will be last feature in 2.12. Then we will start intense debugging and prearing for release.
Sebastian Jennen
@zebastian
IMHO a "render queue management software" could be the next big gain for Mandelbulber, see also this issue:
buddhi1980/mandelbulber2#372
With that it would be possible to manage (monitor, schedule, setup) a big render farm infrastructure for the software in cli mode.
I was thinking about contributing to: http://cgru.info/afanasy/afanasy
What are your thoughts on this, does someone of you have experience on this topic?
mclarekin
@mclarekin
1) sounds a goods idea
2) experience = zilch, but in theory it is simple :)
3) I find the ffmeg stuff at that site interesting too!
Sebastian Jennen
@zebastian
I looked around some more and the only open source solution which is stable, recent and modular seems to be afanasy.
Starmute
@Starmute
How can I stop openCL from crashing when refraction is used? Seems like it exceeds the memory limit - GTX 1070 card
Krzysztof Marczak
@buddhi1980
@Starmute do you have this problem under Windows? If yes, use this solution http://www.fractalforums.com/bug-reporting-b231/alpha-6-start-fails/msg103251/#msg103251
Starmute
@Starmute
thanks Buddhi - I'll try that once I've got a good Windows computer that isn't used too much
Martin
@acidhunter
brrr, it was a hard way
gitlab is migrated to the new server
image.png
Martin
@acidhunter
image.png
Sebastian Jennen
@zebastian
@/all right now i am working on a setup of cgru/afanasy. I also started on bringing the codebase of cgru "in shape" and then i want to integrate mandelbulber support into the program.
Hope this turns out to be a good match!
@acidhunter great news! can you put your changes back to upstream master?
Also we should reference these packages somewhere in the README / on mandelbulber.com?
Martin
@acidhunter
hi there
i had no time to finish the gitlab-ci stuff :-(
i can add id to master later this day
TasMania17
@TasMania17
How far away is 2.12 now? Looking forward to using more than one materials and textures and don't want to go back to 2.11.
Krzysztof Marczak
@buddhi1980
@TasMania17 Functionality of 2.12 with disabled OpenCL is not reduced in compare with 2.11. Materials and textures are not implemented only in OpenCL code.
TasMania17
@TasMania17
Sorry Krzysztof I should have tried it with OpenCL 'off'. It uses all cpu cores and is nearly as fast as OCL on. The message 'compiling OpenCL' takes a long time to complete and when rendering one core is used at any time.