Can only see half a stretched screen (as if it's double height) on 337.12

This is a hard one to describe. Resolution is 1080p (and the TV confirms this when X starts) but what I see is the top half of a screen (~540 lines) stretched over 1080 lines. Width is correct. Make sense? I can draw a diagram or take a photo if required.

To complicate things, all I have to do to fix this is turn the TV off and then on again. As soon as I do that, it’s fine. But if it’s all turned off for a while (eg overnight) and then turned back on, it looks dodgy again.

I don’t know enough about the underlying handshake that goes on with EDID but everything I’ve read (and I’ve spent a month looking into this now) seems to point that way. I’ve no idea how to remedy this though. I’ve tried and reset a couple of things but it hasn’t worked. As far as I know, this is not underscan. It’s far too big and precise (it’s exactly half the screen).

This is on a GT 630 with a Samsung 3DTV (in 2D, “just fit” mode). There is an AV Receiver in the middle (Onkyo) but the issue persists with a direct connection too. I have also tried different HDMI cables and ports.

I actually had something very similar with Ubuntu a couple of year ago with an Ion and a LG LCD TV but this is completely new hardware on a completely fresh install.

There’s nothing in the nvidia-bug-report but I’ve attached it anyway. The Xorg.0.log does contain EDID information if that’s at all useful. Both logs were taken while the system was in a “broken” state.

X: 1.15.0/7.7
Nvidia: 337.12 (334.21 showed this too)

This is how it looks when I turn the system on:

External Media

And this is how it’s supposed to look, after I’ve turned the TV off then on again:

External Media
Xorg.0.log (233 KB)
nvidia-bug-report.log.gz (89.2 KB)

Plz share photo for this issue

I’ve added before and after shots. Before means when I first turn the system on, after means after I’ve turned the Samsung TV off then on again.

I can confirm exactly this issue. Any driver later than 304.X leads to the stretched image on my Samsung UE40F6100 3D TV. Turning it off and on again solves the problem, but of course, this is no solution because everytime XBMC changes the refresh rate (24fps,25fps,30fps), the image is stretched again.

xorg.conf is the default one from the driver’s installation routine.

filed bug 1512542 to track this issue.

Any news here? By the way: How can I access the bug tracker? There’s no possibility to register on https://nvdeveloper.nvidia.com ?

For some reason I didn’t get a notification about the updates here but how do I track that? My forum login doesn’t work on the nvdeveloper subdomain.

How do I track the issue?

Does this help:

nvidia-settings --assign CurrentMetaMode='DFP-0: 1920x1080'

OliWarner, the bug tracker is not public. I just looked it up and our QA people are trying to find hardware that can reproduce the problem.

can you put how you got it to install properly for me i’ll i’ve gotten was bits and pieces no complete info so far trying to install a version slighter high than yours 337.25 thanks in advanced

or atleast a direct link to something that says how on 14.04 64bit

nvm discovered sudo add-apt-repository ppa:xorg-edgers/ppa
sudo add-apt-repository ppa:ubuntu-x-swat/x-updates

sudo add-apt-repository ppa:xorg-edgers/ppa
sudo add-apt-repository ppa:ubuntu-x-swat/x-updates

adds newest ones into driver manager or additional drivers app

Sorry, no. Does nothing.

Thanks for checking. This might be model-specific to Samsung. I’ve only just noticed nvbenutzer’s reply above. My Samsung is the UE46F6100 (just a bigger version of the UE40F6100).

If they need EDID information (or a tester), I’d like to think I can help.

Hi, I also can confirm the same issue
Samsung UE46F6100
GeForce GT630 (GK208)
driver nvidia-331.38

I also tried the GT640 and the GS8400 in between. Same issues. I can’t believe that there is no possibility to find out the differences between driver versions 304 and 331.

(I also updated my TV’s firmware to version 1031 yesterday… no solution.)

Edit: One more comment - latest 340 also does not help.

Again me. Please, could anyone have a look at the corresponding bug tracker entry? I wonder if there’s anything happening here, or if I really have to buy that ATI crap for beeing able to use hardware acceleration with my native screen resolution. :(

But if it’s all turned off for a while (eg overnight) and then turned back on, it looks dodgy again.

OliWarner, I think in this case system can suspend, screen-lock, TV goes in power saving mode etc. In which case did you observe this issue? We don’t have exact TV but we try to reproduce issue with EDID but still no luck to reproduce.

Also what desktop env you guys using KDE, GNOME, UNITY etc …?

To be clear, it doesn’t have to power cycle. As others have said since, a simple resolution or rate change can break it.

I’m usually not using a DM at all (I’m just loading X for XBMC) but I have tested this as broken under Unity and OpenBox. I don’t think it’s relevant to the problem here.

I think given from the replies, this has to be a quirk against this particular range of screens.

DM: LXDE or X only (xmbc-standalone)
Happening: As soon as the resolution or refreshrate is changed, also on first boot of course

Is there no way of finding out differences between the current drivers and the 304 branch which did/does work?