buddhi1980/mandelbulber2

Official repository for Mandelbulber v2

Martin

@acidhunter

everyone can get 300$ credit on google compute for free

with my documentation and dockerfiles its really easy to get up a cluster, nearly unlimited in size

Martin

@acidhunter

is it possibly to render in tiles?
i think this could give better performance

example:
a settings file will distributed to 4 worker, every worker renders 1 tile.
later the finished tiles get pushed to a server that merges the tiles to 1 picture

Krzysztof Marczak

@buddhi1980

Now it's divided to lines, which gives the same result as dividing by titles

Martin

@acidhunter

some days ago i tried to network render at 20k x 20k
i took 5 minutes to start the rendering

if every worker is standalone i think this could give a sppedup

Krzysztof Marczak

@buddhi1980

Ok, I agrre with this case, but it's not typical usage. If the image will be divided to tiles, then would be difficult to balance it

every tile will have different rendering time

Martin

@acidhunter

on small images the actual way of doing network rendering is nice
but if u scale up to 100 or more workers it would get rrealy slow

Krzysztof Marczak

@buddhi1980

And now every worker is independent. Can work even with very slow internet connection, because doens't need requests from server all the time. Every worker can decide what to render next.

Why you think that it would be slow?

Martin

@acidhunter

a 20k x 20k image needs some minutes to get the network clients start rendering

Krzysztof Marczak

@buddhi1980

The problem is lack of memory in this case

Martin

@acidhunter

yes that is the point

sry its difficult to explain, my english sucks a bit

beacause that i ask if it would be possible to render tiles

if a settings file get uploaded to the nodes, every node is only rendering a part and uploads the image afterwards to a ftp

so every node works alone

Krzysztof Marczak

@buddhi1980

What I can change is another allocation of memory for NetRender clients. Now it allocates memory for full image. I can change this to allocate only that parts which are actually rendered by given client.

It should solve this problem.

Martin

@acidhunter

i dont know if this works

just an idea

Krzysztof Marczak

@buddhi1980

I don't like idea with tiles, because it will be difficult to balance between nodes. Some of nodes can finish rendering very fast and some after very long time. Then it will be unefficient.

Martin

@acidhunter

i dont mean to implement this in netrender

Krzysztof Marczak

@buddhi1980

If you just ask to rendering parts of image, it's only partially implemented in the code, but not finished.

Martin

@acidhunter

another example,
uploading a settings file to 100 servers, every server renders a defined part and uploads it afterwards to a ftp or something else

so it would be possible to distribute it to a cloud provider to render images with unlimited size

after all nodes are finished the tiles can merged

Martin

@acidhunter

output could be something like this

http://mandelwerk.deviantart.com/journal/Worlds-biggest-mandelbulb-image-3-3-Gigapixels-214228096

Sebastian Jennen

@zebastian

@buddhi1980 please see latest commit. this may interfere with your code changes. watchout for merge conflicts! buddhi1980/mandelbulber2@2fb9571

Krzysztof Marczak

@buddhi1980

Thanks for your commit. It's good idea to test fractal formulas. Now I'm merging changes.

Krzysztof Marczak

@buddhi1980

I can take some time, because I have to solve problem with mandelbulber_cl_data.h inclusion

Krzysztof Marczak

@buddhi1980

I have merged all code. You can continue. But before please read comments added on GitHub in latest commit.

Martin

@acidhunter

hello there
is opencl included by default when building from git?

i can do some tests on archlinux and docker

Krzysztof Marczak

@buddhi1980

in mandelbulber2/qmake directory there is mandelbulber-opencl.pro. This will create makefile for opencl version. When you start Mandelbulber go to preferences and enable OpenCl on GPU tab.

Sebastian Jennen

@zebastian

i am having a problem with fractal.hpp, the enums are nested inside structs, like:

// basic combo
struct sFractalCombo
{
    enum combo
    {
        mode0,
        mode1,
        mode2,
        mode3,
        mode4,
        mode5,
        mode6,
        mode7,
    };
    combo modeA;
    //        combo modeB;
    //        combo modeC;
};

Krzysztof Marczak

@buddhi1980

doesn't work in pure C?

Sebastian Jennen

@zebastian

this does not work inside opencl:
/usr/share/mandelbulber2/opencl/../opencl/fractal_cl.h:278:2: error: type name does not allow storage class to be specified
typedef enum {

my first idea was to simply unloop this and put enums before the including struct, but we have got collisions with each other and the global namespace: acos

Krzysztof Marczak

@buddhi1980

It can be done if you convert them to correct names like: enumFractalComboCombo and items cound be: FractalComboComboMode0, FractalComboComboMode1...

another example: enumFractalMagTransformsMulti_orderOfTransf, and items like enumFractalMagTransformsMulti_orderOfTransfTypeT1, enumFractalMagTransformsMulti_orderOfTransfTypeT2 ...

they are long but will be no collissions

Sebastian Jennen

@zebastian

i see, thats problably the right way in the first place, and decouples the formula specific struct from the general enum types which are used cross formula

Krzysztof Marczak

@buddhi1980

I think we have no other choice

Sebastian Jennen

@zebastian

puh, got it

that was a lot of find and replace, if we want to change the names again, this should now be easier

Krzysztof Marczak

@buddhi1980

@zebastan, I have changed names of all enums to start with enum..., but now your script cannot rename items in emums body, so we have duplicate names. I have corrected all enums manually, but would be good to correct php script (if it is possible).

Where communities thrive

People

Repo info

Activity