A question about "Software rendering for gnome-shell"
Adam Williamson
awilliam at redhat.com
Tue Feb 21 23:40:29 UTC 2012
On Tue, 2012-02-21 at 16:35 -0700, Michal Jaegermann wrote:
> On Tue, Feb 21, 2012 at 05:12:13PM -0500, Adam Jackson wrote:
> > On Tue, 2012-02-21 at 13:10 -0700, Michal Jaegermann wrote:
> >
> > > Running 'LIBGL_ALWAYS_SOFTWARE=1 gnome-shell --replace', as proposed by
> > > drago01, indeed starts an intended mode gnome session, both before and
> > > after installing these scratch packages, but results are spectacularly
> > > useless. 'gnome-shells' eats whatever CPU it can get so one can look at
> > > results but no hope of doing anything useful.
> >
> > CPU usage is going to be higher. That's what "software rendering"
> > means. I've not found it onerous on an early c2d, but again, I really
> > need to know what kind of CPU is being complained about.
>
> This particular one is a 64-bit (albeit quite old) processor:
>
> vendor_id : AuthenticAMD
> cpu family : 15
> model : 5
> model name : AMD Opteron(tm) Processor 142
> stepping : 1
> cpu MHz : 1600.062
> cache size : 1024 KB
>
> on a board with 2GB of a physical memory. I know some machines around,
> and doing useful job, where this is a quite powerhouse in a comparison.
> When forced with LIBGL_ALWAYS_SOFTWARE=1 gnome shell was taking all the
> time between 94% and 96% of CPU and a response latency for a keystroke
> or a mouse movement was in a order of few seconds. My guess is that
> more of CPU would be grabbed if only it would be available. At the
> first moment I thought that the whole thing just locked up and only
> after some delay I realized that I was mistaken.
>
> I have no idea how much this CPU usage would have to be reduced before
> this setup would pass a "laugh test" but probably 10% for gnome-shell
> would be way too much.
Can you test with a release (not debug) kernel?
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
More information about the test
mailing list