GIMP AppImage (continuous integration)

Your understanding of the apt output is correct.

Hi!
First of all, thank you for this incredible work.
I tried to run all versions of gimp-cce and in none of them I can open the side panels (tools, layers, etc). When I click on single-window mode, the app crashes. Everything works fine with the standard gimp appimage. I’m using Ubuntu 16.04.
Is there something I am missing?

Hi! I’ll be providing new packages in a short while, including a CCE one… I propose that we first see if the new package has the same problem, and then eventually investigate.

hi @Carmelo_DrRaw, it’s been found that an “03” flag is not set in the RT5 Ubuntu download, and this seems to lead to a 5x loss in speed. I just thought I’d mention this in case it also needs setting in the AppImages you are kindly working on! (Recall I thought GIMP sluggish…) There’s more detail here -

cc @heckflosse
Best wishes

PhotoFlow is compiled with

-O3 -ffast-math -ftree-vectorize

For GIMP, I am using the default compiler flags that are set by the ./configure script… I should check with the GIMP devs if -O3 is safe in the GIMP case.

Thanks for the suggestion!

Remember I know precious little about all this software, but I couldn’t help notice you have -ffast-math and the post I referred to earlier says (in a different context?) “Using -ffast-math is dangerous” …

It should be, I have been using it for years.

1 Like

hi @Carmelo_DrRaw, I was just wondering whether to make a start with the current AppImage, or wait for one with the O3 speed-up. Do you think you’ll be building one soon please?

I’m compiling a new AppImage with up-to-date sources and -O3 flag for GIMP right now, will be online in a few hours…

Latest one I downloaded is slightly slow. Is that normal ? I have a very fast computer and regular Gimp 2.8 installed through the Debian system runs just fine. :frowning:

Maybe once @Carmelo_DrRaw posts the new appimage itll be faster

A new AppImage was just posted, you should try that one. Also which version of Debian are you using?

@paperdigits I’m using the alpha of debian 9 right now, so basically Debian testing.
Do you have link for the latest appimage ?

I’m thinking of making an appimage of the newest rawtherapee in debian for people with other distros

It is on the Community Built Software page!

1 Like

thank you!

@paperdigits has been faster than me… indeed, I have uploaded a standard gimp appimage this morning. For the CCE version it will still take some days…

@Carmelo_DrRaw that’s fine I’ll use the regular. Can’t wait until 2.10 is out :smiley: You have no idea how encouraging it is to see all the things that are happening in this community. I’ve been waiting and hoping for so long and finally I get to see the day when FOSS overtakes those stuck up commercial alternatives. These are good times to live in :slight_smile:

1 Like

hi Stefan, @Carmelo_DrRaw , all, how are you finding the new GIMP AppImage? Sorry to say I’m finding it rather slow. I repeated the Levels test I did previously (20Mpxl image, 32bit floating point, move the middle slider in Levels with preview unticked) and it’s taking 16s (i7 6700K cpu). That’s with the whole image displayed. The more you zoom in, the quicker it responds. Ubuntu 16.10 System Monitor is showing only one cpu core being used. My system is snappy with RT.

@RawConvert thanks a lot for the packaging of the appimage first of all :smiley:
I’ve just run it and it is faster than the old version that was up. it is however still roughly half the speed of gimp 2.8.18 which comes with Debian. I understand an appimage will load slower …could it also be the reason for it running slower overall? perhaps it’s optimization it needs, perhaps it’s just about there being more calculations done period. Now I did notice the slowing down both in 8bit mode and 32bit integer, pretty much the same. so it’s either more things being done in this version or the fact that it’s an appimage.


@RawConvert the mouse also seems to lag a little in it. weird

@Carmelo_DrRaw packages the appimage, not @RawConvert :wink: