[Assist] Having Problems with [AMD/ATI] RV730 XT [Radeon HD 4670] GPU

Kuba Kukielka kzer-za at cryptolab.net
Thu Nov 20 20:52:13 GMT 2014

Hi there,

Thank you for your reply! Here's some anwsers to your questions.

> When trying to make it work, try with KDE first, I had some issues with
> gnome3, it only did work correctly with 2 screen that had the same
> resolutions setup.

I have tried the most popular GUIs (GNOME, KDE, LXDE and XFCE) and they
work fine but had the monitor problems that were mentioned before (using
KDE at the moment though I have used GNOME previously though my display
manager is stil GDM because I have had the best experience with it so
far (But I might try KDM because of the customisation)).

> I guess you're using gnome3 with the fallback (you have an ATI GPU...).

I have used it at the time but now I use KDE.

> I'm happy to see that at least something does work, you're using
> xf86-video-ati, right? (I only gave a very quick look at the log).

Yes, I use xf86-video-ati.

> How are your monitors connected?

I have one GPU that are connected to two monitors, one VGA (1360x768)
and the other DVI (1920x1080).
> That's xrandr -q's output?

No, I just ran xrandr by itself but xrandr -q displays the same thing.

> Here it tried to probe the monitor trough i2c and failed to find the
> EDID trough that.

I am not sure if the kernel (libre package) has the radeon i2c enabled
but I am trying older versions of linux-libre (checking from disros that
have the monitors working, using uname -r). I am in the process of
intalling one.

> You can also get it trough sysfs, example mine is at:
> /sys/class/drm/card0-LVDS-1/edid
> So:
> 1) do we really need the firmware to get an i2c bus out of that card?

Not sure, maybe it does contain important information but I don't know
how to check.

> 2) can the VBIOS somehow get you the EDID, for instance by connecting
>    both screen at boot...

I do have the screens connected at all times and I don't think it is the
VBIOS' fault because both monitors work at boot (though mirrored but I
can't fix that) and because other systems worked with my resolution. But
I might be wrong because I don't know how to check.

I will also reply when I have the kernel compiled.


Kuba Kukielka <kzer-za at cryptolab.net>

More information about the Assist mailing list