How The ATI Catalyst Driver Has Matured Since The RV770 Launch

[phoronix]

Phoronix: How The ATI Catalyst Driver Has Matured Since The RV770 Launch

It has been two years since the ATI Radeon HD 4800 (RV770) series launched so we have gone back since that monumental hardware launch and have re-tested each Catalyst driver release since then to see how the performance has changed for the ATI Radeon HD 4850 graphics card. The Catalyst driver has certainly matured over the course of two years in speeding up the OpenGL performance with this hardware along with bringing new features to their proprietary driver, but it is not exactly smooth sailing.

http://www.phoronix.com/vr.php?view=15054

[NomadDemon]

hard regress.... dont like it. I changed 1 month ago from FGLRX to opensource drivers, 3d is still buggy, but 2d owns, plymouth works ok[with fglrx doesnt] and many good stuff for opensource drivers

i have 4850

[grigi]

Thanks Michael for this comparison. It is quite informative :-)

So there is a noticeable regression on 3D performance for 10.6 hmmm...
Is this because of the 2D engine change?

[rohcQaH]

Is this because of the 2D engine change?

possibly. There's only one way to find out: Michael, can you re-run the 10.6 tests with A2AA disabled? Or did you already wipe the test-rig?

[sabriah]

Very nice comprehensive test. Thanks a lot!

It was a nice find for ETQW.

[Fixxer_Linux]

I just can't wait that OSS stack can handle properly any kind of games, to get ride of fglrx.
What annoys me most in fglrx is video tearing.

[Exopaladin]

What annoys me most is the way that for the OpenGL beta 10.3/10.5/10.6 versions there's been horrible OpenGL bugs on the Evergreen cards that make certain games segfault on lauch or run slowly on 4 series cards. Even more annoying is that I'm pretty sure the bug was confirmed for 10.5 and announced fixed, but the fix didn't make it into 10.6.

[Fixxer_Linux]

What annoys me most is the way that for the OpenGL beta 10.3/10.5/10.6 versions there's been horrible OpenGL bugs on the Evergreen cards that make certain games segfault on lauch or run slowly on 4 series cards. Even more annoying is that I'm pretty sure the bug was confirmed for 10.5 and announced fixed, but the fix didn't make it into 10.6.

Sadly, when running linux, buying latest hardware can be deceiving, as manufacturer either don't give any driver and/or any testing device for the kernel devs to develop a driver.
More deceiving as we should attempt from AMD a premium handling of latest cards as they are particularly friendly with Linux / OSS world.
Nvidia is no better, as it took ages for their binary blob to drive the 8800 2 or 3 years ago, when those cards were ruling the 3D worlds.

Let's hope that with OSS stack and more unified code from fglrx side, this bad story won't happen again...

[Panix]

Tons of regressions illustrated via those tests. And you expect the Evergreen cards to be 'working' when the older cards aren't performing any better than older drivers (e.g. 10.5)? If you don't mind, poor 3D with FOSS drivers, then get ATI?

'Wish I didn't read that article..

I've been told by a few ATI *owners*, in particular HD 4xxx owners, not to bother and go with Nvidia. I heard good things with the 10.6 drivers but those aren't evident in those tests.

[Qaridarium]

Tons of regressions illustrated via those tests. And you expect the Evergreen cards to be 'working' when the older cards aren't performing any better than older drivers (e.g. 10.5)? If you don't mind, poor 3D with FOSS drivers, then get ATI?

'Wish I didn't read that article..

I've been told by a few ATI *owners*, in particular HD 4xxx owners, not to bother and go with Nvidia. I heard good things with the 10.6 drivers but those aren't evident in those tests.

don't waste your time on the past the future will be more successful.