Wow, the current radeon works exceptionally well!

[Porter]

Does the versions of Mesa and Radeon in Ubuntu X-Swat fix the performance regression that I had with the X200?

Because for some reason 3D is messed up these days with the radeon-rewrite.

What was the performance regression, and in what app? Maybe I can check it for you.

[Melcar]

What's the current state of rovclock? Last time I used it, it did not support anything newer than r3xx.

[Ant P.]

Btw a tool to check video clock would be awesome - to see if the dynamic clock actually works :-)

`rovclock -i`?

[Extreme Coder]

Does the versions of Mesa and Radeon in Ubuntu X-Swat fix the performance regression that I had with the X200?

Because for some reason 3D is messed up these days with the radeon-rewrite.

[bugmenot]

yeah, radeon works awesome!

but i would like to see scaling options like in catalyt on windows (on linux catalyst does not work at all, so i could not test), so that i can use the full screen of my tft tv, connected over hdmi. the picture on the screen is too big, so i cannot see the whole picture.

Thanks!

[juhuuu]

Hi
I do agree that the radeon driver provide good performance.
However I only get 1800fps on glxgears - IBM T60 1,8ghz x1400 mobility.
I am have to use kernel 2.6.29 because witj 2.6.28 and 2.6.30 I get x-freezes all the time, I really really hope this will be fixed in the future because I dont wont to be stuck with jaunty forever :-)


What I still find very to be annoying is that I get strange, non consistent pixel corruption after resuming from standby.
By "not consistent" I mean once I resume I have corruption in some areas or whole windows, resizing windows sometimes helps, but when I go back to standby and resume again everything is gone and all is fine....very very strange.
I am using x-squat drivers currently, when I use tormods newer git drivers this corruption issue is a lot more severe and working with the notebook is impossible.

And powersaving could be a little better ;-) .. but I know u guys are working on that. Btw a tool to check video clock would be awesome - to see if the dynamic clock actually works :-)

cya

[legume]

Turning off bicubic filtering did change things a bit! Definitely sharper. I did some experimenting, and it seems like it actually looks best with bicubic on and a mild unsharp mask filter on. Better than bicubic off (with or without a filter), actually. Obviously this depends heavily on the source material.

I guess that combination would be nice - with unsharp affecting perception and bicubic doing a nice smooth scale.

http://en.wikipedia.org/wiki/Acutance

[Porter]

Turning off bicubic filtering did change things a bit! Definitely sharper. I did some experimenting, and it seems like it actually looks best with bicubic on and a mild unsharp mask filter on. Better than bicubic off (with or without a filter), actually. Obviously this depends heavily on the source material.

[legume]

HD-quality video works great, but I found lower-res (SD) video to be a bit soft when scaled to full-screen, so I added a 3x3 unsharp filter to the MPlayer options within SMPlayer, which helped immensely. Refer to the docs on that one if you want to try it, or post in this thread and I can help.

On r500 there is an xv attribute for bicubic filtering, which defaults to 2 (auto) - which means size > 2x gets bicubic filtering rather than bilinear.

You can use xvattr to force it off if bicubic looks too soft for you.

Edit - beaten to it again :-)

[agd5f]

HD-quality video works great, but I found lower-res (SD) video to be a bit soft when scaled to full-screen, so I added a 3x3 unsharp filter to the MPlayer options within SMPlayer, which helped immensely. Refer to the docs on that one if you want to try it, or post in this thread and I can help.

The default filtering for r3xx-r5xx is bicubic. Some people find this too soft. You can disable it (to use linear filtering) by changing the Xv attribute XV_BICUBIC to 0, e.g.,
xvattr -a XV_BICUBIC -v 0