Submitted by Adam Jackson has announced the availability of “the zeroth release candidate(s) of the next Xorg release(s).” The changelog is here. Before downloading, be warned, in the developers’ words: “The RC number is not accidental. This is unpolished and rough, and is only just at the point where we can usefully have large numbers of people testing it and fixing things.”
Xorg 6.9 and 7.0 Release Candidate Zero
About The Author
Adam Scheinberg
Technology Executive • Web Developer • Father • Foodie • Music Snob • OS enthusiast
Follow me on Mastodon @[email protected]
the opengl/xgl implementation is missing…hoped it would come with 7.0
It’s open source, patch it yourself
As the announcement says, there’s not a great deal of functional difference between 6.8.2 and 6.9… just improvements to what’s already there. And 7.0 is just a restructured version of 6.9.
All the cool stuff will hopefully start coming in after the 7.0 release, once they have the modular codebase to work in.
Xgl is nowhere near finished yet.
I hope that exa will land in 6.9/7.0, at the moment only the sis driver includes the new acceleration architecture.
any thing ground breaking new in this release?
any improvement on the performance front?
The modularity of 7.0 is ground breaking. You can finally develop parts of x separately from one another at their own pace. After that I gues the realy nice stuff will folow.
It seems the transition to XOrg has lessened my hate for X. Development speed has really picked up and it seems like things are looking up.
Why the heck did it take so long to get real transparency into X (retorical, I don’t really wanna know).
That sounds cool.
It’s good to see DRI development moving forward on ATI’s newer chipsets. I have a Radeon 9250(the last series supported by DRI) and have been happy with the DRI(especially since the ATI fglrx drivers don’t seem to work with PCI cards), not to mention it’s nice to have the drivers come preinstalled with the OS(with our any fuzy licensing legality), not having to install them afterwards.
I always wondered, is open-source DRI performance with a Radeon nice? I was thinking of trading a radeon 9600 for a nvidia board, but it would indeed be nice if I could instead use it with the new x.org. Or are the nvidia drivers better performers?
nVidia has always rocked the socks off of ATI’s half-assed linux driver set.
…and that includes an open-source DRI driver.
Open-source drivers for both ATI and NVidia are pretty slow. They’re good enough for light use, but if you’re a gamer you definitely want to get the proprietary drivers.
Because open-source drivers doesn’t contain cheats.
Nope, even without the cheats the open-source drivers are still really slow.
The DRI drivers for Radeons 9200 and under are quite good, and if you enable S3TC (which is unfortunately patent encumbered), the performance seems to be about on par with the Windows drivers. See here for S3TC: http://homepage.hispeed.ch/rscheidegger/dri_experimental/s3tc_index…
The current support for the Radeons higher than 9200 is experimental, I have no idea of the performance because I don’t own the requisite hardware to try them out. See here: http://r300.sourceforge.net/
AFAIK, there are no open source 3D accelerated drivers for NVidia cards.
DRI performance is quite nice. 2D and general functionality is better than in proprietary drivers. Of course people tell you that the 3D part is not that great, but it does work and gives you the ability to play most games – and it’s open-source. The support for 3D DRI in eg. 9600 (R300-series) is now in X.org CVS, which is a great accomplishment, even though it still needs a lot of work (join if you know stuff about video drivers and 3D!). see http://r300.sourceforge.net/ for developer information.
Anyway, NVIDIA doesn’t have any working open source driver that gives you 3D acceleration. Its proprietary drivers are still better than ATI’s proprietary drivers, but if you value open source and want to get rid of non-free software in your system, your best bet is ATI and the DRI development it has.
<quote>
The support for 3D DRI in eg. 9600 (R300-series) is now in X.org CVS
</quote>
Will it be released with X11R7 ?
Yes, the modularity of 7.0 is what’s most important. It is a better design and should make the development go faster with more innovative ideas being put in.
Either I’m missing something, or isn’t a release candidate defined as a version that could be a full release if there turns out to be nothing wrong with it?
If they claim it’s rough and not finished, shouldn’t it be a beta version, not a release candidate?
Maybe because it’s feature complete, or they’ve entered a feature freeze
Well, beta means feature complete and bug fixes only. What gives?
Actually no;
Normally when things are developed, there are three levels of development – things are prioritised.
Alpha is when the MOST important things are complete, Beta is when all the core AND major parts are complete, and the release candidate is when it is feature complete – new features *MAY* be added *IF* they meet release standard (which is rare).
Its rare to have only one release candidate – IIRC, Windows 2000 alone had 3 release candidates – it all depends on the scale and scope of the features been added.
any word if they ironed out some of the composite bugs?
composite is pretty slick but right now it is basically a toy because most people have stability problems with it, i know i do.
Composite is slow but it isn’t unstable. You probably have an nVidia graphics card – the drivers for -those- are unstable when doing composite work (especially with that EXPERIMENTAL RenderAccel option that every Gentoy ricer uses).
Composite may be stable but there’s no point to using it unless you also use Render acceleration. I’ve tried it. Software render is deadly slow when doing transparent windows. Stack two or three windows together and we’re talking an update every 3 seconds or worse (Athlon64 3400+, GF4MX, 1920×1200 32-bit).
The situation might be better with the new Render MMX/SSE work but I doubt it can compare to using video hardware which is designed for the job.
this rating system is really not working, i browse at -5 because people get low ratings for the most ridiculous reasons
if i see something as offensive i will dock them a point, but not just because i don’t like what they say, and that seems to be what people are doing
example.
RE: where is opengl extension??
By Anonymous (IP: 142.161.67.—) on 2005-08-03 19:14:26 CDT in reply to “where is opengl extension??”
It’s open source, patch it yourself
this was set at -2…. how is this offensive…. how is this a bad comment? At my default browser setting of -1 i would have not seen this comment and there are numerous others that i have seen like this… some very good and noteworthy comments at -3 or -4 because somebody didn’t like what they were hearing.
Yes, I agree. This rating system sucks.
I considered registering with OSnews but I will not untill they go back to the old non-rated system.
This rating system is realy equivalent to censorship.
I also agree. In fact I don’t like the new site at all. The old one was simple and perfect. why mess with it if it’s working? were there crowds cheering for the new one? I doubt so.
It is not equivalent to censorship since you can still see the comment by changing the settings. If you want the old site just browse at -5. Personally I like the change.
This rating system is censorship, that’s my opinion too.
You can mod good articles up, but not bad ones, that add nothing to the discussion, down.
You can only censor them, if they are insulting or sth. like that. But no correct way to say “this posting is not worth the 4 points it has”.
it wasnt that it was offensive, it that the reply was rather assholish
basically in that example, the poster is a prick and didnt add anything meaningful to the discusion, and frankly im glad those people get moderated down they are a waste of hard drive space to even have their comments saved.
and to the censorship troll, not it is not censorship, they are allowed to say whatever they want, however in response to that i am allowed to voice my opinion of his message in the form of moderation points, the power isnt in a single hand either.
it wasnt that it was offensive, it that the reply was rather assholish
Bzzzt, wrong answer! The post wasn’t assholish, it was the dry truth of Open Source. If it is not in the “official” release and the stuff can be patched in, you are Free to try and patch your own source tree to include the technology.
Not that technically adept? Too bad, then you will have to wait just like the majority of us.
Or do you think that it is assholish for some people to have more skills, more money, more real estate. Would you “mod down” a millionaire for being able to do a world trip on a whim, which would probably cost you your life savings?
This is Open Source people. You have the POSSIBILITY to be on the bleeding edge, not a DIVINE RIGHT to get bleeding edge features at the moment YOU want them delivered “take-away ready”.
To make a short summary: Get over yourself!
And you seriously need to chill out. Geez.
No. He’s right. Many OSNewsers and Slashdotters absolutely DEMAND things when, as if developers are their slaves. Those people are the ones who must chill out.
yes exactly, the dry truth, aka being a total prick
it was a waste of bandwidth to even say that in the first place. “get the source and patch it yourself” how insightful…
man you are a moron
i have no problem with a rating system but it should only be a positive rating system, then the moderators are free to mod down or eliminate those offensive comments
i don’t think it is good that weather or not somebody sees a comment (who uses default settings) be judged by people with a hot temper or people just out to cause trouble.
i appologize for starting this discussion…. i wasn’t intending on starting a flame….. how about we keep it on the subject from here forward…
My bad
We’re all geeks here. You’re a geek. I’m a geek. Geeks like performance.
How’s it perform? In all aspects, not just 3d.
All your speed are belong to 6.9/7.0?
I’m more concerned with reliability and features actually. If I wanted raw performance I’d probably use Windows
.
I’m gonna guess the RC performance isn’t that great; but it’ll probably improve before release and sub-releases will probably improve more. Overall I’d guess this release will be slower than 6.8.2, if anything. Simply because the largest change is a move to a modular design rather than monolithic design. However, the rate at which they develop new features is supposed to change. And things like XRender should get very interesting in the next year.
> Overall I’d guess this release will be slower than 6.8.2, if anything.
– MMX blending routines for the Render extension
(http://lists.freedesktop.org/archives/xorg/2005-April/007446.html)
– Added gradient, solid fill, and convolution filter operations to Render
(http://lists.freedesktop.org/archives/xorg/2005-May/008105.html)
And hopefully with more driver support in upcoming RCs
-New EXA acceleration architecture, with experimental support in sis(4) (more to come)
More Slow??
More quicker, you mean. The blending routines weren’t optimised, so with mmx they probably will be faster. They haven’t optimised for sse, because it would break in some plattforms.
The Exa, is claimed to more than twice fast than XAA, about the render issues, it probably could be more slow, but at least the operations are done imediatly…
By the way, about r300 project. I would like to give a big THANKS, for the project now it’s possible to play ET, and see kdeTV, without my screen going black. although they haven’t implemented the AA, and some fog extensions. In some games the r300 is almost on pair, with ATI, but in others the “cheats”(Specific Optimizations??) rules.
I’m execited wih this new version let’s wait and see.
It’s good to see via / unichrome coming along. I have been using Fedora Core’s stock drivers (i.e. vesa) to play DVDs for 1.5 years.
If I read that right, it said initial support for not running X as root. That’s pretty ground breaking for security. Also, I saw there was some work on MMX and RENDER extensions. That’d speed up some stuff, I’m sure. But as I’m not a programmer (I just know keywords), I don’t know how much of an improvement that is.
Since X is moving towards modules, how are we going to smoothly move from 6.9/6.8 to 7.0? Would just doing a simple yum/urpmi/apt-get/emerge xorg-X11 xorg-xserver handle this? Just curious.
Well I know I will be about to apt-get dist-upgrade to it. I don’t know about them yum users.
Might take a week for yum to do it’s thing. Same for the gentoo ricers, all that compiling…
Really though, if you use apt, this should work fine, as for the other package systems, I don’t know enough to make intellegent comments (as you can see from my trollin g above).
Is it possible to port X drivers for directFB ?
How difficult would this be ?
I didn’t see anyone talk about EXA. EXA is the new accelerating architecture. Is this going to solve the problems with composite?
EXA is a stop-gap acceleration architecture more than anything. The writing is on the wall: an X-specific acceleration architecture is no longer a good idea. EXA is mainly so Composite and whatnot can work on older hardware or hardware unsupported by the new X-on-GL infrastructure.
I’ve been using the latest X.org snapshots in my Gentoo linux system and Xorg is definately making progress as far as the composite extension goes. It’s much faster and almost bug free (combined with nvidia drivers). Much better than the “stable” release.
What would it take for NVIDIA support of EXA? Support in the drivers or X.org or both?
Actually Miguel is going to talk about Xgl tomorrow at OSCON – http://tirania.org/blog/archive/2005/Aug-02.html
Heya,
I don’t know what the license is, but people working ont he haiku-project are developping/have opensource 3d-drivers for nvidiacards:
http://haikunews.org/1074
http://web.inter.nl.net/users/be-hold/BeOS/NVdriver/index.html
I just found it interesting to report …
greetings,
Michel
Indeed, I was about to post about it, and then saw your comment.
The 3dnews page on that site has lots of interesting benchmarks too…
http://web.inter.nl.net/users/be-hold/BeOS/NVdriver/3dnews.html
As for the license, it’s X11/MIT. All of Haiku is under the X11/MIT license (non-copyleft like new-BSDL).
So if this guy continues to make such amazing progress, then these drivers could indeed be ported to DRI for X.org …
As matter fact that’s the easy part. The hard part is getting the hardware lingo nailed, which once it’s done will remain the same for any platform.
I wish this guy the best of luck, if he succeeds in getting some of the latter cards working, then we might finally have a good base to build off of in making future nVidia drivers. Since I would imagine it’s easier to keep up with them when you already have an opensource driver implementation for one of their previous cards. As it is now all we’ve got are drivers for TNT.
Honestly, I wish I knew what kind of tools would make hardware reverse-engineering easier. Since it would be great to set up a fund to get this ‘be-hold’ dude anything he needs to get this all figured out.
“I wish this guy the best of luck, if he succeeds in getting some of the latter cards working, then we might finally have a good base to build off of in making future nVidia drivers.”
I really would like to wish the same luck, BUT I have to notice that he generally wrote it glazing at UTAH project or something like that, where was showed how to get cards up to geforce 2 hardware opengl, he could also get it in cards up to geforce 4 because such are only ‘tuned’ geforce 2 chipsets. Next cards are _too_ different to gf4, and without specifics Rudolf will propably never get them working in hardware opengl mode
Sorry for my English
David
I wish that someone working at nVidia or ATi (that wouldn’t happen to include any OSNews readers, would it?) would take a look at this work. It’s really a shame that we can’t have leading-edge graphics support, given that we have highly capable people willing to write the drivers free of charge if they were given the appropriate specs.
Graphics architectures have progressed to the point where they are about as programmable as a general-purpose computer architecture. No one would buy a computer for which there are no publically accessible programming guides. It is no secret what instructions the P4 supports or what control registers are available on the XScale. If Intel would release a radical new CPU where there is no documentation on the programming model, only proprietary compiler support via ICC, then nobody would buy it.
Then why has the industry allowed these graphics companies to support undocumented features above and beyond VESA standards via proprietary drivers?
> Bzzzt, wrong answer! The post wasn’t assholish, it was
> the dry truth of Open Source.
Bzzzt, right answer! The post was assholish, because it served no other purpose than a putdown.
“Does it have feature A? Why doesn’t it have feature A? When will it have feature A?” All these are valid questions, worth discussing. Valid expectations and valid disappointments.
“Go code it yourself” is just 100% hostile and negative. It really serves no other purpose than to have the unfriendly person enjoy his 5 seconds of fascinating power of assholedom. There is a sense, a feeling of power and superiority in being unfriendly and dishing out putdowns.
Most of the time the person saying “go code yourself” doesn’t code anything himself. I don’t know where they come from, but often they are unable to have an intelligent discussion on “when exactly is feature A coming out.” Just frustrated and hostile.
And technically everyone of us knows that everyone of us can have his own fork of Xorg, so there is no real matter-of-fact reason to mention it.
I suggest that the moderators would encourage informed discussion on “when is it ready, what will it have, why, why not”, and silently remove all “go code yourself” messages, and everything that follows them.
Maybe we should actually read what the other one has written before shouting “assholish” and calling for censorship*… maybe I did get something wrong, but I think no one spoke of “code it yourself”, but of “patch it yourself” which is quite different. For example reiser4 has to be patched in manually if you don’t use the -mm kernels, however that’s a very minor work compared to writing it from scratch.
* ‘I suggest that the moderators would […] silently remove all “go code yourself” messages’ – yes, I think this can be called “calling for censorship”.
The good thing of now being modular is that driver including exa support can be released still later.
it’s not slow on my comp… in fact, as far as speed goes.. i can basically not tell it is on, but it is however unstable to the point that most of the time i keep it turned off…. (I hate to hard reboot).
and yes i do use an nvidia card with RenderAccel enabled lol
Depends about what problems you’re talking. EXA is there to improve speed, not remove rendering artifacts.