If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
ATI R500 Linux Performance -- With an X1300PRO, X1800XL, X1800XT, and X1950PRO in hand we compared the ATI 8.40 and 8.41 display drivers. The performance improvements for the Radeon X1000 series range from 50% to 90% faster.
Your testbase cards in the article are all PCIE cards..
Any chance you can lay your hands on AGP versions and an agp rig and kick some of these R5xx chips on AGP bus sets?
Currently my 9550 AGP (R3xx) kicks my X1650Pro AGP's (R5xx) ass sideways simply because fglrx_drm does not see the CARD as agp - it identifies the chipset as PCIE and never negotiates an AGP bus speed change. Doing anything that moves large amounts of data to the card (cough THINK about that statement) is sluggish in the extreme since the card is effectively in AGP mode0 (PCI standard speed)
What I'm reading here tho - I may just kick this driver in and give back a speed diff on the R3xx chips ... sounds like there might be a boost in there for me too....
Anyhow -- I"ll file version bump request @ gentoo and be waiting with baited breath for the release - bet yer ass I'm hauling this one down and setting up the ebuild for it in bugzilla.....
It actually says it won't work with 2.6.23 (general information), so you can't use it on Fedora 8 (specific information). Maybe I should re-use my signature with "people thinking distros are different kinds of linux"...
I didn't ask about brand new (not even released) 2.6.23. To me, and most others (iirc) with x86-64, you can't use anything after (including) 2.6.20. That's terribly weak.
I don't know what you're trying to say regarding distros, but surely distros do configure and even patch the kernels differently, so you can't compare a certain vanilla kernel version with for instance "Fedora 8". But even if you could, do you think it's reasonable to refer to Fedora releases instead of kernel versions? Am I supposed to know whatever kernel Fedora X uses? To me, that's completely irrelevant. I will never care even remotely whatever kernel version Distro X, version Y uses. I simply don't care, and I shouldn't have too, since I'm not using that Distro damn it!
Just wait and see. Michael said it'll be available in 1 week wtf.
Tell that to everyone else in this thread then? I'm not the only one commenting without actually having the driver. Or perhaps you just thought I was too negative, and thereby I needed to be shut up?
We all speak out of experience with the so far drivers. Many of us see a pattern of not caring about really fixing the sucky drivers, but just fixing a bug a month, and this time, making the driver get a better score on FPS benchmarks.
That's a terrible loss of knowledge to me then, omfg! I've gotta go back to school and learn these important things, before I give another statement on a forum on the internet. The whole fucking world could collapse!
Anyway. I've got some answers now.
* No RandR 1.2, because ATI has invented their own proprietary tool before RandR 1.2 was invented, and because of that (very interesting reason) they won't use RandR 1.2 even though it's superior.
* 2.6.22 _should_ work. Well, 2.6.20 _should_ have worked (if you ask me) a long time ago. Will they have fixed support for 3 major kernel versions in a row? OMFG. Unbelievable. And highly unlikely :/
Well well, we'll see... Maybe the "open source" driver ATI will release in a couple of decades, and which will have become well tested and patched a few decades after that, will be interesting.
But ATI surely doesn't understand open source development. Release a 50 mb .tar.bz2 and say "we're now open source" is lunacy. Release _early_ and _often_. This way, important and highly skilled people (I'm thinking of X.org developers) can pin-point flaws and comment how things should be interfaced, to work better with X.org.
Otherwise, it might be "open" but it'll take a _long_ time until it'll "work like a charm" with "just about any linux kernel, distro, X.org-version", which the currently support drivers more or less do.
But ATI probably just want being tagged as "open source friendly" instead of really being interested in community development.
Again, I'm speaking ahead of time, but it doesn't take many brain cells to figure out what's going on here. I'm not saying I'm guaranteeing the new ATI driver to suck, I'm just saying it most likely will.
AMD 8.41 Display Driver Preview -- Our first article on this new driver are simply the technical details such as noting this driver has been actually in development for over a year, the AIGLX support that is coming next month, and other relevant information.
ATI R300/400 Linux Performance -- We re-tested the Radeon X300 and X800XL with this new driver and it offers some HUGE performance improvements! The Mobility Radeon X300 was as much as 54 times faster compared to the previous ATI Linux driver. The improvements in these older ATI GPUs are simply astonishing.
Okay Michael I realize that PCIE is the hot new standard -- but there are a ***heck*** of a lot of AGP bus based cards out in the wild still -- can we get tests on AGP bus cards -- The cards in this article are again all PCIE based..... or IGP, which if sanity in chipset design exists are PCIE internally.
Michael: Will CCC integrate well with the new "Graphical configuration tool for X" coming out in 2.20? It'd be nice to have say an extra tab called "CCC" - sort of like how the windows forcewares work.
Also, I noticed all the benchmarks use x server 1.3 - is there x.org 7.3 /x server 1.4 support in the new driver? I can't find anything on the thread that confirms or denies this.