177.70 (BETA) for Linux discussion & problems thread

[philipl]

Originally Posted by Kirurgs

Feedback about 177.70.

Here (see signature) results are as follows:
* hibernate broken completely (black screen after resume), however worked fine with 177.13

I had this problem as well (XPS m1330, 8400GS) when I turned MSI on. If I leave MSI off, I can resume from suspend or hibernate.

[artem]

Originally Posted by dinosaur

The only problem I encountered so far with 177.70 (as well as with all previous 177.* beta drivers) is related with CoolBits: as soon as I try to touch to the GPU or memory clock settings (even by *underclocking* them), the driver crashes the computer (blank screen, keyboard unresponsive, sshd unresponsive, no disk activity) as soon as I press the "Apply" button.

A fix is on its way.

[mathew.winters]

heres some problems with this driver and the last release version..

http://www.nvnews.net/vbulletin/showthread.php?t=115209
http://www.nvnews.net/vbulletin/showthread.php?t=116946

nvidia have not appeared to acknowledge this problem in those threads, maybe they might notice it here!

[Jon Nettleton]

I am not sure what it does but I definitely like the results. I was seeing improved performance with the newest drivers and settings, but things like cairo-dock still rendered very slowly when my cpu scaled down. I was randomly playing around when I noticed that IPP could be changed from 1-4. If 2 was better than 1, I figured 3 should be great. Well I have to say it is!!! It has fixed all the speed problems I was seeing with Compiz and my 9600gt card. I then tried 4 and found it dog slow again. hmmm.. Oh well back to 3. Hope it helps someone else out, or the devs to figure something out. Let me know if I can provide any information.

I take that back. Seems like 3 acts very much like 1. Cairo-dock very fast and smooth, moving windows smooth, resizing windows very slow and jerky. I still think I like 1 and 3 better than 2 but it isn't perfect.

[AaronP]

I'm not entirely sure why it was designed this way and why the X driver doesn't reject it, but 3 isn't a valid value for InitialPixmapPlacement.

[Jon Nettleton]

I see that now. I wonder what it is doing them. It certainly works better than any other value. If only we could look at the source. ;-)

[Astar]

I have all kind of regression slowness with firefox on 177.70, comparable to 177.68 if I use it under xfce, but under kde it works well. For example any "quick-reply" element (on this forum, or livejournal's) becomes unusable, cause x-server start hanging with 100% cpu load.

177.68 worked well under both environment.

[madmage]

Originally Posted by Astar

I have all kind of regression slowness with firefox on 177.70, comparable to 177.68 if I use it under xfce, but under kde it works well. For example any "quick-reply" element (on this forum, or livejournal's) becomes unusable, cause x-server start hanging with 100% cpu load.

177.68 worked well under both environment.

same here (177.70 on a KDE4, nvidia quadro nvm135m), sometimes, when switching quickly among desktops, or when the wallpaper changes (I have a slideshow), X hangs with 100% cpu load

[how]

xvideo flickering
resume after hibernate broken

[taupter]

There's a trick for those who have slowness in plasma when using composite: try adding

export QT_NO_GLIB=1

to a place that will run before KDE4's window manager. It will disable Qt's Glib event stack support (if it was enabled at compilation time), and things may speed up a lot. It seem there's an issue with Qt with Glib support enabled and KDE4's window manager when compositing is enabled. KDE devs are trying to fix it (maybe it's already fixed in SVN).
AFAICT recompiling Qt4 with Glib support disabled causes no problem to KDE at all, but I can't say it for sure. Please test and report problems.

[eliwap]

taupter -- The line seems to have had a significant and positive impact on my 6600 card.

I put the line at the bottom of /etc/X11/xinit/xinitrc. So the bottom of the file looks like...

export QT_NO_GLIB=1
nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1

However...

?????? WHERE ARE THE LEGACY DRIVERS ??????

[spectred]

Originally Posted by taupter

...KDE devs are trying to fix it (maybe it's already fixed in SVN)...

afaik it is fixed in svn (i read something about this on planetkde) but i'm not absolutely sure.