FC5, progression or regression?
berryja at gmail.com
Wed Mar 1 09:09:31 UTC 2006
On 2/28/06, David Timms <dtimms at bigpond.net.au> wrote:
> Dave Jones wrote:
> > On Mon, Feb 27, 2006 at 10:59:54PM -0500, Bill Nottingham wrote:
> > > Mike A. Harris (mharris at mharris.ca) said:
> > > > >a 3 GHz P4 processor for heaven's sake. It should run FC5 like the wind.
> > > >
> > > > That's why it is a test release, to find bugs. Don't use test releases
> > > > if you expect flawless OS operation and perfect performance.
> > >
> > > However, if it's *significantly* less performance, it could certainly
> > > signify something that's wrong.
> > Current kernels still have slab debugging on (Because we still have
> > bugs that this is tripping up that need whacking). The increased
> > overhead of this sucks up memory bandwidth, and I wouldn't be
> > surprised if shared-memory video chipsets feel some pain.
> With glxgears, I see the gears update about 15 times in 10 seconds, and
> the results are (AMD athlon 2600+, 512ram, nvidia fx5600)# glxgears
> 888 frames in 5.4 seconds = 165.811 FPS
> 798 frames in 5.0 seconds = 158.416 FPS
> 912 frames in 5.7 seconds = 161.284 FPS
> 798 frames in 5.0 seconds = 158.274 FPS
> 798 frames in 5.1 seconds = 155.004 FPS
> It's hard to tell if this is just a stobing effect ;-)
> I think it was at least an order of magnitude higher in FC4 - from
> memory approximately 5000 f/s (but perhaps only with the nvidia
> (non-open) driver).
Are you able to run glxgears with the nv (open-source) driver? It
does not work for me (x86_64 PCI-e 6600 GT):
X Error of failed request: GLXBadContext
Major opcode of failed request: 143 (GLX)
Minor opcode of failed request: 5 (X_GLXMakeCurrent)
Serial number of failed request: 22
Current serial number in output stream: 22
Same as Partha noted with an ATi card. Interesting that both of our
systems are 64-bit. I also cannot get my GLUT programs to work. They
give the same error, with the minor difference that the above 22s are
both 29 with the GLUT programs.
More information about the fedora-test-list