Seth Nickell posted some screenshots and videos showing the experimental Luminocity window manager & Cairo which enable a XOrg-based desktop to get accelerated GL eye candy graphics. update: more here.
XOrg Catches Up with Longhorn/Mac OS X’s 3D Desktop Acceleration
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
“2) Why linux users always say that eye candy effects are useless and for “noob” when they don’t have it, and why they are working for an X server that allow those effects.”
It’s because there is not a single group of people who all think the same called ‘Linux users’.
This is a little off topic but…
If you look at ten random computers running Linux, they will be running different distros, with different desktops, and perhaps even on different architectures. The diversity of choice is astonishing.
Some pure console users even refuse to use bash as a console because it’s too ‘bloated’, preferring sh or csh.
Others are using LookingGlass with a full 3d accelerated desktop.
And yet, I see the same comments assuming that there is some consensus and groupthink going on. I find the opposite to be true, and am puzzled as to why people assume all Linux users think the same way.
@finalzone:
Thanks for trying to help. I did restart X properly and the Composite extension is running (fglrx reports it has deactivated DRI due to the enabled composite and xdpyinfo lists it too) but Damages and XFixes are not. I even put the following in the config file:
Section “Extensions”
Option “Composite” “Enable”
Option “XFIXES” “Enable”
Option “DAMAGE” “Enable”
EndSection
and in the logfile I get:
…
(**) Extension “DAMAGE” is enabled
…
(II) Initializing built-in extension DAMAGE
…
yet xdpyinfo doesn’t list Damage and xcompmgr and luminocity complain that they need the Damage extension. I’m really puzzled about this.
I find it really curious that nobody has asked it before:
When can we see this awesome technology in an ordinary distro, like Fedora or Ubuntu?
Is it in the timeframe of gnome 2.12 or is it more realistic that we have it in 2.14?
Thnx!
I wonder if itll work on a my p2 450/voodoo 3 agp/384mb ram machine.. would b ecool if it did.
“Considering that Xorg with Composite/Xcompmgr is buggy as hell (and I can’t imagine how buggy is the new GL server + Cairo), yes, by the time Longhorn will be out maybe this stuff will be usable.”
Yes, Longhorn has no bugs and its development is not late…
Longhorn is late, yes. But it will be delivered sooner than usable, integrated Linux hardware accelerated compositing, even if is ships at 2007. Do you believe the mess of several uncordinated projects (Cairo, Xmopmgr, Xorg etc) will be integrated fully any time soon? And what with the drivers not all supporting proper OpenGL accelaration et al.
Also, while Longhorn will have bugs, and many, none or very very very few of its bugs will have to do with compositing. For users, compositing will be a non-issue, it will just work. Minor glitches could exist, but nothing major regarding compositing, hardware accelaration etc. It’s just a part of an API that MS can’t afford to get wrong, and severe testing will ensure it is not wrong.
Deep down inside, Linux is ugly like some fat geek’s buttox. Grandma still isn’t supposed to install her leenucks and get on AOL.
All of that said, by all means have some fun and share it with folks! It’s useful, it can be stable, and it’s just plain awesome. Why complain?
Nicely put.
Linux is ugly because geeks neither know, nor care about beauty (design, fashion, etc) except in the form of abstract “beautiful” maths or code and such.
You only have to look at how geeks and programmers are dressed to know that this people cannot possibly design anything pretty.
It would be nice if the Linux community had designers and illustrators that could help (it helps Mac and Windows -many designers use windows). Sadly, it doesn’t with the exception of a handful.
(Yep, some geeks use Gimp et al and consider themselves “designers”, some even land designing jobs. But of the *real* designers -i.e not people making some lame websites and pizza delivery pamplets, people that design book covers, cd covers, magazines, etc- I doubt even one used Linux as a work platform.
Some people likes FEATURES, not eyecandy
Which just goes to show that you are a working drone with no aesthetics whatsoever.
Is it in the timeframe of gnome 2.12 or is it more realistic that we have it in 2.14?
I think this is the stuff we’ll see in Gnome 3.0 and KDE 4.0, not before.
Cairo, xcompmgr and Xorg not integrated? What the hell are you talking about? They already are, for heaven’s sake!
xcompmgr is not meant to be a finished product, it’s a testbed. The whole idea of the Composite-extension is to let the window manager take care of it, like Luminocity does. If that’s not integration, I don’t understand what it is you want integrated.
The problem is, this is very new technology in the X11 world. It’s much easier to do on Windows and OS X since they are free of the client-server model, but we have to take different approaches. And once the Composite extension in X.org matures, there are no obstacles. This is technology that works here, now, today. It will certainly be stable long before 2007. A lot of people use it with great stability right now.
The non-accelerated drivers are not an issue, there have long been software-rendering support for OpenGL on Linux. It’s much slower, so people without hardware support would probably avoid it, but the vast majority of Linux users using NVIDIA, ATI and Intel cards will be able to run this very smoothly.
Once the Luminocity stuff gets implemented in Metactiy, compositing will be a non-issue for Linux users as well. And I can’t wait. I just tried a Mac OS X desktop the other day, it’s REALLY useful! Especially the Exposé window management. That thing rocks! A very good example of how this stuff can increase productivity and provide a much more effective GUI environment.
– Simon
is there accelerated support for the older 3dfx cards like the voodoo3 ? even beos has 3d acceleration for them.. so im assuming yes.. think this would work ok on one? if not ill have to wait till i get all the part for my gaming tower together.. i currently have an nvidia 6800 256 sitting here crying
These video show eyecandy… maybe some think it’s useless but try to understand what is possible… the swinging windows are neat but it’s the live pagers that show whats possible why wouln’t you want to monitor what goes on live on your virtual desktops without the performance hit?… and I think eyecandy can be effective when used correctly… like toolbars that are transparent when not in use eg. office on OSX. This is technology that will mature and be a part of everyday computing.
“Yep, some geeks use Gimp et al and consider themselves “designers”, some even land designing jobs. But of the *real* designers -i.e not people making some lame websites and pizza delivery pamplets, people that design book covers, cd covers, magazines, etc- I doubt even one used Linux as a work platform. ”
talk about insulting. how come both gnome AND kde have more attractive iconsets, hell, are just plain more attractive then windows xp? how come windows xp, made by a company that should have the money to hire someone BETTER then those washed up artists (like jimmac), looks like a teletubby exploded on the desktop?
Longhorn is late, yes. But it will be delivered sooner than usable, integrated Linux hardware accelerated compositing, even if is ships at 2007. Do you believe the mess of several uncordinated projects (Cairo, Xmopmgr, Xorg etc) will be integrated fully any time soon? And what with the drivers not all supporting proper OpenGL accelaration et al.
well, i do believe they “will be integrated soon” because they already are. xorg != xfree86. the drivers not supporting ogl will need a better software emulation layer then exists now, but that will most definately be done by 2k7, and the vast majority of users that would want this sort of thing anyways will own an nvidia or ati card. (which both have ogl support)
Also, while Longhorn will have bugs, and many, none or very very very few of its bugs will have to do with compositing. For users, compositing will be a non-issue, it will just work. Minor glitches could exist, but nothing major regarding compositing, hardware accelaration etc. It’s just a part of an API that MS can’t afford to get wrong, and severe testing will ensure it is not wrong.
dude, “severe testing” has yet to make any product in the history of software development bug free. i have yet to see the game that launches and has no “it doesnt work with my vid card” complaints. expect avalon to be usable by 2k7-2k8. i wouldnt be suprised if this is great by the end of the year on most major vid cards, and it goes typical linux incremental improvements from there.
some people want FEATURES, not eyecandy
christ, you really have gotta be pretty dumb. linux is the king of features, it has more features then any other os ive used, its the nonexistant or god awful interfaces that make it so much less attractive to most people.
forget that attractive interfaces are more efficient then utilitarian ones (google for don normans emotional design), forget that stuff like expose wouldnt be possible before on linux, and isnt possible on windows, forget the fact that a vecoctorized desktop means resolution independance (when you bring the res up, stuff looks better, not smaller).
people here saying that this is pointless are either extremely dumb, or just dont know what they are talking about.
I followed the directions posted in the blog and got it up and running (woohoo!). Aparently the API is undergoing some changes(who woulda thunk)-so when I built it I kept getting the same error messages mentioned by ‘foo’, ie. missing @XTHREADS_CFLAGS@….
So not knowing anything about the code I just sat down I tried to do some good ole pattern recognition- I noticed some files referenced :
-DXTHREADS
the @ symbol in a makefile, AFAIK, just symbolizes some library which was not resolved in the autogen.sh/configure process. So I simply edited the respective makefiles and swapped @XTHREADS_CFLAGS@ for -DXTHREADS and everything compiled fine with little annoying errors about some important change (!). Anyway it compiled and runs here…actually rather neat…unfortunately it doesn’t run quite as fast and smooth on my machine as seen in those demos….(2.4 GHz Celeron/GF5700)…but not much slower, much less smooth…only a whee bit….
Also just to note I have been running xcompmgr fulltime for about 2 months with very few problems..older gtk apps (xmms, mplayer) are the only real sources of problems nowadays….It’s still not quite there-(surprise!)-but everything works-it’s kind of funny how anything on your screen which is being dynamically updated (ie, the system monitor panel applet, or a terminal which is being updated) shows through any opengl apps running in full screen mode. Made the mistake one time of lauching enemy-territory ina terminal window with xcompmgr running-kind of wierd seeing a black rectangular box showing through while running e-t…hehe…
The last couple of years I always bought ATI cards, but their drivers are really crappy, so yesterday I bought myself a brand new Nvidia 6800 GT card and it works great with composite!
So Linux users, buy yourself a Nvidia ,ATI is really no good for linuxusers.
karl I’m getting an error trying to build xserver and it’s complaining about @XTHREADS_LIBS@ what’d you end up putting there?
Seeing the massive overhaul of desktop concepts that is planned for Gnome 3.0, I’ll guess this will be in Gnome long before 3.0 is even started.
My guess is 2.12 for cairo and som of the xcompmgr effects. Luminocity will probably leak some features into the 2.14 release. 2.16 will probably be 3.0 but that would mean that they will go long over the 6month release schedule.
Ok, sorry, all of you saying “don’t buy ATI” are weak.
All ATI cards up to the Radeon 9200 are supported by the DRI. There are a couple (important) aspects to those cards for which ATI didn’t provide docs, such as HyperZ and fragment programs. HyperZ has been reverse engineered and implemented in the DRI drivers. Fragment programs have been 90% reverse engineered, and are one developer’s pet project. He will almost certainly have them finished soon.
r300+ chipsets are being very quickly reverse engineered. So, I recommend this: Either buy cards <= Radeon 9200, and write ATI & NV and tell them that you would have bought a $200 card, except they didn’t provide free, complete docs for them, or buy a card > Radeon 9200 and help with the reverse engineering. (Or, if you own an NV card get off your butt and start reverse engineering.)
A modern graphics card is basically a second, specialized CPU… and what sort of fool would buy a CPU from AMD or Intel without proper documentation? No one, other than your Packard-Bell using grandma. Get a clue, and don’t buy undocumented hardware unless you intend to reverse engineer it… no matter what lame “not our IP” excuses the companies might make, and no matter how fast the undocumented card is!!!
The DRI desperately needs a new vram memory manager, and that is being worked on. After than, the drivers are going to rock. (They already do.) Don’t bitch, help.
p.s. Even Windows users should be pushing for Free Software drivers from the card manufacturers… I don’t know why you trust benchmarks without them.
I only saw a problem as regards:
@XTHREADS_CFLAGS@
not:
@XTHREADS_LIBS@
I won’t even pretend to understand how the software works. All I did was do some basic pattern recognition and guessed-and guessed right!-which is more a matter of luck than knowledge.
I cannot tell you:
a) the difference between @XTHREADS_CFLAGS@ and -DXTHREADS
b) the difference between @XTHREADS_LIBS@ and @XTHREADS_CFLAGS@
my working assumption is that when you see a gcc error message, or a Makefile, referencing somthething surrounded by @ symbols, that this means that the autogen.sh/configure process failed to resolve whatever is found between the @ symbols.
I was building with jhbuild and got the @XTHREADS_CFLAGS@ error as reported by gcc. jhbuild offered me a choice of options as to what to do next. I chose ‘4’ which dropped me back to a shell in the working directory where the compile failed. I then looked around in the configure file to see if I could see what was going on-somewhere I saw a reference to -DXTHREAD and thought, hmmm, that naming convention isn’t purely coincidental is it? so I opened up the Makefile in a text editor(nano) and searched for. Sure enough I found 2 references to @XTHREADS_LIBS@. I then replaced each occurence with -DXTHREAD and then exited (exit)the shell landing back in the jhbuild menu and chose 1 to restart the compilation. Some of the packages consist of mulitple subdirectories each with their own Makefile- xserver for example- in which case I edited each of the Makefiles making the substitution described above. Luckily I only had to edit and change about 6 or 7 Makefiles total…
I imagine you could just switch into your JHBUILD_ROOT_PATH, which by me was /root/cvs/gnome2 and simply do something like the following(note I haven’t tried this-so don’t shoot me if it doesn’t work)
find . -name ‘Makefile’ -exec sed -i ‘s/@XTHREADS_CFLAGS@/-DXTHREADS/g’ {} ;
find . -name ‘Makefile’ -exec sed -i ‘s/@XTHREADS_LIBS@/-DXTHREADS/g’ {} ;
notes:
1) not sure about my sed syntax here….don’t know if @ or _ is a special charachter….
2) I know this works for @XTHREADS_CFLAGS@ -but I am not sure about @XTHREADS_LIBS@-try it-if it doesn’t compile I was wrong
oh and a little foot note about my last post…
it seems as if the site strips away ‘back slashes’ ie. the opposite of / (forward slashes)
Insert one of these ‘back slashes’ prior to the ; at the end of the two command lines I posted above-otherwise BASH won’t like you
If you don’t want to try the hack I listed above wait a couple of days and maybe this stuff will be resolved in cvs….
oh and if any of the readers here, perhaps one of the devs
, actually know what causes this error, or what the difference is between @XTHREADS_LIBS@/@XTHREADS_CFLAGS@/-DXTHREADS is, please, by all means, let us know 
r300+ chipsets are being very quickly reverse engineered
Ummm. The 9700 (first R300 chipset) was released in late 2002. What exactly is your definition of “quickly”? I bought a Ti4200 in late 2002 and have had good accelerated drivers since I bought it. Yes, these are closed source drivers. If you must have open source drivers, don’t buy either ATI or nvidia. Neither company actually supports open source drivers anymore. ATI used to, but that is in the past. I think Intel actually has the best record right now with open source video drivers (not sure where Matrox stands nowadays).
Cairo is not going into Gnome. Cairo is being integrated into gtk+, infact is it already in Gtk+ head. Cairo will use glitz for opengl hw accel. on most computers.
So, if I don’t really care about jittery windows and other eye-candy, and I just want windows that move reasonably smoothly and don’t leave trails all over the place on my i810, is this going to help?
Ummm. The 9700 (first R300 chipset) was released in late 2002. What exactly is your definition of “quickly”?
My definition of “quickly” is “moving fast.” What other relevant definition could there be?
The r300 reverse engineering project started in the Fall, roughly. They’ve been making steady, incredible progress since then.
They still have a lot of hard work to do, though, and for some people this has been their introduction to reverse engineering. (The r400s are very similar to the r300s from a driver standpoint, so both are effectively being reverse engineered in parallel.)
And, yes, Intel does seem to be the best right now about providing documentation to Free graphics driver developers. However, many people don’t have the option of installing agp or pci Intel graphics cards (maybe i915 ones are out that I missed) so I omitted them… but they’re great for laptops. (Intel provided fragment program documentation that others have withheld.)
“how come windows xp, made by a company that should have the money to hire someone BETTER then those washed up artists (like jimmac), looks like a teletubby exploded on the desktop?”
Yep, that’s what it always comes down to. Big ass company like MS and the only reason they’re so prevalent is they got to the scene first. Massive Windows division and they’re still “borrowing” from Apple.
actually, apple got to the scene first. ibm/microsoft managed to gain ground after a series of supendously expensive machines comming out of apple (lisa, first mac). what many people dont realise is that microsoft has never competed (until now with linux) on the strength of their product, they competed by undercutting the competition, while offering solutions that were “good enough”. the truley ironic thing is that is the way that linux is spreading now. microsoft is being forced to justify the cost of their product for the first time ever, which is why we have seen so much improvement in the last few years.
my point with that statement is that the origional poster seems to be talking about 1990 linux. things have changed alot, and i think you would be hard pressed to find someone who doesnt think that most of the art in linux is inferior to that of windows. we have some very talented artists now. look at the oo.o iconset jimmac did, and compare it to words. compare the crystal iconset with the windows one. compare plastik to luna (personally, i have never been a fan of industrial, but hopefully gnome will be getting something a little nicer as the default look sometime in the near future).
“and i think you would be hard pressed to find someone who doesnt think that most of the art in linux is inferior ”
should be
“and i think you would be hard pressed to find someone who thinks that most of the art in linux is inferior “
just run xcompmgr without any option.
Trolltech has just hired well-known KDE hacker Zack Rusin to work on X11, Mesa and fbdev.
Yes, they’ve made progress lately, but by the time they have a good driver for R3XX/R4XX, R5XX will be out with a completely different architecture. Reverse engineered drivers are always going to have this lag problem. The simple truth is that most users don’t care if their drivers are open source and the graphics companies aren’t loosing sleep over this issue.
ATI’s closed source drivers SHOULD be a lot better than they are. This would be excusable if they provided specs, but they don’t anymore. At least nvidia puts significant effort into their Linux drivers. Supporting ATI because somebody was able to figure out the R3XX architecture in spite of ATI basically rewards ATI for its non-commitment to Linux.
As long as I can turn it all off, go for it I say.
You wrote:
Deep down inside, Linux is ugly like some fat geek’s buttox. Grandma still isn’t supposed to install her leenucks and get on AOL.
All of that said, by all means have some fun and share it with folks! It’s useful, it can be stable, and it’s just plain awesome. Why complain?
Nicely put.
Linux is ugly because geeks neither know, nor care about beauty (design, fashion, etc) except in the form of abstract “beautiful” maths or code and such.
You only have to look at how geeks and programmers are dressed to know that this people cannot possibly design anything pretty.
It would be nice if the Linux community had designers and illustrators that could help (it helps Mac and Windows -many designers use windows). Sadly, it doesn’t with the exception of a handful.
(Yep, some geeks use Gimp et al and consider themselves “designers”, some even land designing jobs. But of the *real* designers -i.e not people making some lame websites and pizza delivery pamplets, people that design book covers, cd covers, magazines, etc- I doubt even one used Linux as a work platform.
Clearly you never worked at NeXT nor Apple. Quite a large number of engineers are stylish in wardrobe and other areas of life outside of writing eloquent code. And that includes mechanical engineers, astrophysicists, electrical engineers, linguists, so on and so forth who all share high aesthetics for style on both sides of the Interface Paradigm.
Perhaps that is why we are all considered elitist snobs who worked at NeXT.
Being that I’m writing this on a PC running Debian Sid/KDE and upstairs is a laptop with OS X I can say that both OS’s have room to improve and continue to progress.
When KDE releases a stringent UI Guideline document I look forward to seeing the improvements. With this it will push OS X to continue to improve as well.
Competition is great.
Just upgraded to an nvidia 5700 from ATI 9200 I know it is not much but Now everything is flying with the xcomposite changes compare to when I used ATI.
KDE support for xcomposite surpasses gnome at the moment, I have both right now (KDE 3.4 and gnome 2.8.2 on Gentoo).
ATI has the benefit on running on lower watts, but the lack of good drivers in linux is not helping them.
I got a NVidia quadro FX 4000 at work which does around 12000 fps (glxgears) on SuSe 9.2… The first thing I do is to get the KDE 3.4 and composite thingie working on it. I think beside nice looks composite adds to efficiency. You just need to get used to it and find your ways with it.
It is still little buggy but with the speed Xorg is moving I think Microsoft is going to have a very tough competitive in a year or 2.
It was very nice, but unstable on my setup. I’ll try it again when I’ve got a new version of X and the nvidia driver installed. Otherwise, though, it looks very good.
saw it now in kde 3.4 on slackware 10.1, xorg 6.8.2, p4-2.4ghz and nividia6600gt-agp
it’s nice but not very useful.
I created a new user for kde, so I can stick with xfce/rox on my main account.
I’m really happy with xfce/rox, it’s damn fast 🙂