2016-04-02 1:31 GMT+02:00 Cedric BAIL <[email protected]>: > Hello, > > On Fri, Apr 1, 2016 at 3:19 PM, Andreas Volz <[email protected]> wrote: > > I compiled e20 on Raspberry Pi 2 with latest Raspbian and it works. I > > had to deactivate Composite Effects and switch to Software renderer > > because it says there's no OpenGL support. >
The latests rasbian images for rpi2 include the experimental VideoCore open source drivers :) https://www.raspberrypi.org/blog/another-new-raspbian-release/ I successfully built efl+gl on rpi2 using this driver and efl applications run quite fast. There is still (at least) one issue i faced: the driver seems not able to load big textures (something like 1000x1000px or more), and it slow down a lot when it fails. This for example make e20 unusable because it cannot load the e background texture. I did not investigate further, maybe it is just a stupid problem. Please let me know if you do some progress in this area, I really would like to see my emotion media center run smoothly on raspy :) > > Yes, RPi have still today a proprietary driver that is not following > standard integration with X requiring a custom backend to make it > work. I was hoping to see the open source driver sooner, but it is not > there yet. We do have in master a backend, eglfs, which will support > this kind of proprietary driver in the near future. Hopefully for > 1.18. It is currently done around hwcomposer, but should also be able > to support RPi and others. This backend will also allow the use of > Enlightenment Wayland only mode. > > The RPi even has something close to hardware layer. So if someone does > spend the time to optimize our backend for it, it should become quite > usable. I say quite, because there will still be some major slowness. > Even with wayland, there won't be support for Wayland EGL, so you will > fallback to Wayland SHM. Wayland SHM allocate at the moment the buffer > on the client side. This means that we can't give it a hardware memory > pointer that the hardware layer infrastructure could reuse. So we will > always need the CPU to do a useless memcpy from the client buffer to > the hardware layer buffer... Making it less useful, except maybe for > mouse cursor... So until there is an open source Open GL driver on the > RPi, it will lack behind on what we can do with it. > > > The reason because I did this is that like to run the Raspi with a > > 800x480 display and several fullscreen EFL applications which are > > stacked one over the other and the above one with alpha and shaped > > window. > > > > I use E20 only as layer manager and deactivated all other elements. > > Maybe E20 is oversized for that reason. > > > > I experienced now that the CPU load of the E process is really high > > even if nothing much happens. In idle the load is >10%, but if I render > > some edje content in my application the E process goes fast up >60-100% > > CPU. My application itself has very low CPU usage. > > Any change on the screen, should trigger a software rendering of just > the area that did change on screen. Note that shaped windows and alpha > are usually more complex to handle and require more ressource to be > drawn. With an anemic hardware like the RPi, it doesn't suprise me to > much that this is that slow/ressource intensive. > > > I tried to start my application with the default Gtk based window > > manager, but it's strange. As long as I activate alpha and shaped > > window support in my application the application is somehow broken and > > hangs in the beginning. I didn't yet trace it out where it hangs. > > > > Do you have any ideas or explanations? > > Not for the last problem. > -- > Cedric BAIL > > > ------------------------------------------------------------------------------ > Transform Data into Opportunity. > Accelerate data analysis in your applications with > Intel Data Analytics Acceleration Library. > Click to learn more. > http://pubads.g.doubleclick.net/gampad/clk?id=278785471&iu=/4140 > _______________________________________________ > enlightenment-devel mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/enlightenment-devel > ------------------------------------------------------------------------------ Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471&iu=/4140 _______________________________________________ enlightenment-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
