The Cost Of ATI Kernel Mode-Setting On Fedora 12

[hobbes]

dodoent: glxgears is not a benchmark, it's not even anywhere close. It would be a much better idea to test with something more resembling real world use than three extremely simple non-lit, non-textured, rotating shapes. Phoronix's own benchmarks are probably the best thing for 3D testing on Linux, actually. Try those.

hobbes: on a quick search I can't find a report of that problem in Fedora or freedesktop.org Bugzilla, or Launchpad. You might want to file a report at fd.o or Launchpad, I guess. Have you verified that it works OK if you disable KMS (nomodeset)?

Yes, it does work with nomodeset. Sorry for the delay to answer.

This issue is still present today.

https://bugs.launchpad.net/ubuntu/+source/compiz/+bug/496653

[DoDoENT]

I don't get why fglrx devs are so overworked that they could not add support for new xservers in time, nv devs manage that too.

Maybe there are more of them and have bigger salaries than fglrx devs, so they are more motivated to add support for new xservers in time...

[Kano]

Kernel updates can be problematic, usually KMS is not that important, but there are really differences if a wlan driver works or not with a specific kernel version. For some specific hardware it could be needed even to downgrade to .28. The problem with fedora is, that the environment is relatively uptodate - and especially fglrx will not run. So you are forced to use oss drivers if you want em or not. I don't get why fglrx devs are so overworked that they could not add support for new xservers in time, nv devs manage that too.

[bridgman]

That sounds like a good approach; pushing development code out to everyone too quickly can be a really Bad Thing.

[AdamW]

bridgman: what usually happens with the kernel is the team tries to do roll-up releases every few weeks, so that we don't get a flood of kernel releases with small fixes, and each big release has a bit of time to get some vague testing done on it. The other components are a little less scheduled, but Dave will push an update when he thinks he's got everything that should go into it done right, I think.

[bridgman]

I guess the specific complaint here is "updating too much in the initial release and *then* updating too little"

The tricky part is that post-release changes may be a big help for one group of users but a big problem for others, and figuring out which potential improvements should be released as updates can require as much or more testing than the original release.

In that case I guess there's a good argument for making most of the potential updates available only for cherry-picking by the user rather than something more-or-less automatic.

[AdamW]

Indeed. This is because it's not as simple as slapping together a fix and stuffing it out, as that's what causes regressions.

It's funny, most people complain about Fedora updating too much, not too little.

[remm]

None of those packages are 'updates', yet. The developers are still working on them and haven't decided to submit them as official updates.

What actually happens is that, even in cases where current packages are demonstrably broken (as in, the user's installation no longer boots), propagating an update often takes days and days.

[AdamW]

There's a blog post by Dave Airlie about this stuff here:

http://airlied.livejournal.com/69074.html

it notes that some of the performance 'regression' is due to the new driver doing vsync, which we wouldn't want to disable (as you get tearing without it).

[legume]

These issues are known to the devs?

The XV high Cpu issue is known to the devs, so will hopefully get better with time.