Weekly Edition
Return to the Kernel page
| |||||||||||||
Linux Graphics, a Tale of Three Drivers
James Bottomley has posted an essay on graphics drivers on the Linux Foundation site. "For Linux, the best way of demonstrating user satisfaction objectively is with the kerneloops project, which tracks reported problems with various kernels (an oops is something that's equivalent to a panic on Unix or blue screen on windows). For instance looking at the recently released 2.6.25 kernel one can see that both the binary Nvidia driver and binary ATI firegl driver account for positions in the top 15 oopses. If one follows the history, one finds that the binary drivers are always significant contributors to this list, whereas open source drivers appear and disappear (corresponding to people actually seeing the bugs and fixing them). This provides objective support for a significant kernel developer contention that it's harder to get fixes for binary drivers. The other bright spot is that the Intel graphics drivers rarely figure at all in the list also showing that if you want graphics to 'just work' then Intel is the one to choose."
(Log in to post comments)
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 15:32 UTC (Tue) by mbottrell (guest, #43008) [Link] I read the article and I suspect many would agree... that an open-source driver is indeed better. I recently moved *BACK* from running an Intel GMA950 (X3100) onboard graphics card to an nVidia based card (GeForce 8800 GT). As such I've moved from an open source driver back to the binary blog known as nVidia drivers. Why? Whilst I would prefer running a card that has opensource drivers, I found that many graphics intensive 3D apps (not games) crawled under the X3100, making them either unable to run or so slow they were unusable. I'm personally haven't run an ATI card, so I cannot state how well their Radeon-HD cards compare, particularly with open source drivers, yet the state of their open source drivers/performance. If Intel want to become dominant on the Linux desktop in graphics drivers I would recommend: 1. They start producing graphics chipsets that can handle something more than low-end 3D. 2. Dedicated add-on cards instead of onboard graphics cards would also be advisable, or at least the option. Onboard graphics cards for laptops seems reasonable, but just doesn't cut it for desktops. They are woefully under powered. Whilst I'm stuck on a binary-only driver and it isn't ideal, I'm at least seeing the 3D performance some of these desktop apps require. I'll stay on the nVidia train for the time being until Intel can produce a card that matches the performance of even the 7000 series cards. In a perfect world nVidia would open-source up their graphics drivers, but I'm not holding my breath.
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 17:13 UTC (Tue) by tjc (subscriber, #137) [Link] Whilst I would prefer running a card that has opensource drivers, I found that many graphics intensive 3D apps (not games) crawled under the X3100, making them either unable to run or so slow they were unusable.Which apps are you speaking of?
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 17:47 UTC (Tue) by sfeam (subscriber, #2841) [Link]
> Which apps are you speaking of?
I don't speak for the original poster, but I can attest to similar
problems arising from the molecular visualization programs used routinely
in my lab. Most are built on OpenGL. Performance using Intel graphics
chips is adequate but not great. Aside from uninspiring performance per
se, it is also annoying that even the latest intel drivers still cause the
machines to lock up hard occasionally. So, reluctantly, I'm back to
fitting out new machines with nVidia cards even if they were originally
configured with Intel graphics.
If you want example apps to download and try, I suggest
PyMol http://pymol.sourceforge.net/
ccp4mg http://www.ysbl.york.ac.uk/~ccp4mg/
Coot http://www.ysbl.york.ac.uk/~emsley/coot/
VMD http://www.ks.uiuc.edu/Research/vmd/
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 19:34 UTC (Tue) by k8to (subscriber, #15413) [Link] I had the same experience running the *ahem* productivity application known as World of Warcraft under wine. The performance was not impressive, although I didn't care that much, and then the system would hard lock for around 30 seconds and reset. I blame thermal issues, but it was using the heatsink etc as supplied from manufacturer, and applying much more prodigious heatsinks did not resolve it.
Linux Graphics, a Tale of Three Drivers Posted Jun 25, 2008 8:53 UTC (Wed) by Wummel (subscriber, #7591) [Link] IIRC Google Earth does not run with the free nvidia driver. You need the binary driver with 3D acceleration to run the application. But I guess for most uses Google Maps is good enough.
Linux Graphics, a Tale of Three Drivers Posted Jun 25, 2008 7:46 UTC (Wed) by tajyrink (subscriber, #2750) [Link] I very much support / only accept using the open source drivers, but I do acknowledge the lack of performance in Intel graphics. I think the article was a bit (not too much, but somewhat) biased. Some points I think are worth mentioning: - Intel drivers are good, especially the new 2.3 series, but not perfect - they still require manually forcing overlay support (instead of textured video) and greedy fallback EXA functionality to perform at useful speed in all situations I encounter. Implemented like that in Ubuntu. - Intel is, while fixing drivers, fixing X and architecture at the same time. I believe that by the end of this year full EXA acceleration can be enabled without problems. - Intel's 3D performance is only at most 50% from what the hw would allow, because of the delays in implementing TTM/GEM stuff in kernel and batchbuffer functionality (without regressions) in the DDX driver. It will get better from the current situation. But, all in all, Intel is the best we have in terms of open source drivers, and they're coming better all the time. ATI's open drivers are quite good, but before AMD opened specifications there were bad problems with specific chipsets, together with some lack of functionality (like sw rendering if anti-aliased lines were enabled in any OpenGL program), which hindered the quality. The next round of driver releases (DDX, Mesa) will have a lot of those fixed, and AMD's R300-R500 (9500 - X1950) might become the preferred choice for many since they outperform Intel solutions so greatly. The ATI's open drivers do not reach the full potential of the cards yet either, though. The X2000 - X4000 series 3D support is probably not ready this year yet, or maybe towards the end of this year compiz might be running... dunno. ATI's binary drivers have sucked, but have finally been improving to a state they actually might eg. not crash on a 64-bit computer or when using both (!) VGA and DVI outputs like in my case, but I wouldn't install those anyway. The open drivers are better in every other aspect except for the raw 3D speed. Lastly, the reality is that for many casual users NVIDIA's binary drivers seem to deliver the best experience. 3D performance is reaching the full potential of the hardware, and they are relatively stable at least for some. The same casual users blame the non-functionality like lack of Randr1.2 support (ie. ordinary tools for changing resolution do not work) on the "Linux", not the drivers. And likewise if the system crashes, which it does on many configurations. And likewise if the NVIDIA's drivers do not support latest X, it's X's fault. But generally for those that the drivers work, they're getting speedy 3D. So, it's easy to understand the cheer for NVIDIA from time to time. But Intel is starting to kick serious ass in any non-heavy 3D use, ie. normal computer use, while ATI has reached binary driver parity with NVIDIA (from what I've heard) and they're starting to have some good open source drivers finally. When the X2000 - X4000 get 3D support, they can be easily recommended over NVIDIA as they have better support for eg. standard X features like Randr1.2, and they work out-of-the-box on any distribution without the need to install drivers. But that'll be next year.
Linux Graphics, a Tale of Three Drivers Posted Dec 5, 2008 11:46 UTC (Fri) by josvazg (guest, #55445) [Link]
Hope you are right and I am just impatient, but the current state of things regarding intel graphics drivers in Linux is quite SAD, as one can discover in this launchpad bug:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-vi... And you can even see in this video: http://www.youtube.com/watch?v=7cWW-12gxdY
Just as a summary:
Linux Graphics, a Tale of Three Drivers Posted Dec 5, 2008 12:19 UTC (Fri) by njd27 (subscriber, #5770) [Link]
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 16:00 UTC (Tue) by yanfali (subscriber, #2949) [Link] I recently purchased an AMD/ATI graphics card (HD3650). It's been about 5 years based on bad experiences with ATI driver support. I did this based on the quality of the work going on in the radeon and radeonhd trees over at xorg. I only need 2D to work right now, but so far I'm really impressed. The card pretty much worked out of the box. Now that AMD has released docs for the R500 series, and the R600 following closely behind, I hope to stop buying Nvidia cards in the very near future; open the specs and I'll reconsider that position. I also recommend Intel based graphics at work, and for people who don't want 3D, based on the quality of the Intel xorg driver. In a business setting, the existence of a high quality stable 2D open source driver is important. Nvidia is a hardware company that does software development on the side. After a few years Nvidia has no commercial incentive, other than goodwill, to maintain drivers for obsolete platforms. Xorg does.
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 16:39 UTC (Tue) by NAR (subscriber, #1313) [Link] The free drivers (when I last checked) either haven't supported my card (HD2600, at least one year old) or haven't supported xv, so I can't watch movies, TV, etc. fglrx is pretty buggy too, I'm afraid. So I find myself using Windows XP more and more...
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 17:41 UTC (Tue) by drag (subscriber, #31333) [Link] That's to bad. I enjoy the stability of my Intel systems to much, I guess. I don't remember the last time X has locked up.. Used to be (say a year ago) reliably unreliable no matter what Linux system I used. The only repeatable issue I run into nowadays is when resuming from suspend-to-ram the keyboard input is ignored occasionally. So I have to use a usb keyboard and reboot the system. Right now if you want better-then-Intel, but retain open source drivers, is to get a R520-generation ATI video card. The 3D support isn't quite complete yet, but it's usable for some games and for compiz stuff. AMD's support is helping things move along much quicker now... we shouldn't see the huge time lag in open source ATI support with the newer stuff that we experienced with the r200-r500 generations support. The r500 3D support is bleeding edge right now. I think that Fedora 9 may support it somewhat, but I don't know all the details. I don't know how stable it is. In terms of real cards you'd probably want to look at ATI 1650 and 1650pro.
Linux Graphics, a Tale of Three Drivers Posted Jul 3, 2008 19:11 UTC (Thu) by anton (guest, #25547) [Link] Right now if you want better-then-Intel, but retain open source drivers, is to get a R520-generation ATI video card.Why do you suggest that? Probably any ATI video card will give better performance than Intel's integrated graphics, and the support for up to R4xx is much more mature than that for R500 and up (not that I know much about the latter). The best 3D performance as reported on free3d.org is from a Radeon X850XT (R480); actually I don't see any R500 or R600 results there, which indicates that the R500 driver is not yet even mature enough to run glxgears. So my suggestion for people who want 3D performance with free drivers now is an ATI card with a chip up to the R480. In the last years I have happily used a Radeon 9600 (RV350), a 9250 (RV280), and an X850XT (R480), always with free drivers.
HD2600 = R600 Posted Jun 24, 2008 18:01 UTC (Tue) by tialaramex (subscriber, #21167) [Link] That's an R600, I'm a bit dubious about the "at least one year old" claim for a chipset that ATI launched only in May 2007. Did they really put the mid-to-low end HD2600 variant out less than a month after the flagship product ? Until sometime earlier this month these were the most advanced chips from ATI, so you were on the bleeding edge as far as driver support goes. I guess I'd say that, for now at least, you don't want to buy any graphics chipset that only just came out. Actually that goes for Windows users too, none of these drivers are bug free in the first few weeks of a new chipset, why pay extra to go through the pain of being an early adopter? It makes sense for games developers, and teenagers looking for something to brag about, but not for anyone else. ATI has only released a fraction of the documentation for R600 chips. To make Xv work on your hardware someone probably needs to program the shaders to implement color conversion and scaling in hardware, which needs more documentation than is public. There may be people with that information, but if so they're busy working not talking. The new chips either don't have old-school separate hardware video assist, or its only available on the low-end variants, which no-one is likely to write a separate driver for. So a shader solution is the only way out. In Fedora 9 I gained shader-based video acceleration for an R400-based laptop and a machine with nVidia onboard graphics, so it's happening.
HD2600 = R600 Posted Jun 24, 2008 19:40 UTC (Tue) by NAR (subscriber, #1313) [Link] It felt like a year, but maybe I actually bought it last autumn, not last summer. Actually I bought it for gaming purposes too. Anyway, it's a little frustrating - I've bought ATI instead of Nvidia, because they've opened their specifications, but I end up using Windows at home (at least for a while). It's just sad that a couple of years ago my desktop could hibernate fine (with an Nvidia video card and maybe a 2.4.x kernel?), nowadays hibernate doesn't work at all on my desktop. It just doesn't look like that the "year of the linux desktop" is coming anytime soon :-(
HD2600 = R600 Posted Jun 24, 2008 19:48 UTC (Tue) by rfunk (subscriber, #4054) [Link] There's a definite tradeoff here. Buying an nVidia and using their binary drivers would (probably) make things work well now, while an ATI with free drivers might have problems now. But with the ATI card the driver is likely to only get better with time, while with the nVidia card the driver is likely to get harder to make work. So it should be interesting to see how things are with the same card running next year's software. Or even this fall's software.
HD2600 = R600 Posted Jun 26, 2008 12:38 UTC (Thu) by NAR (subscriber, #1313) [Link] By the way, is there relatively pain-free method to test the latest and greatest drivers (and X)? Doing a "dist-upgrade" can lead to nasty non-graphics related surprises, a live CD with some applications (e.g. googleearth) would be more useful for this purpose.
HD2600 = R600 Posted Jun 26, 2008 12:48 UTC (Thu) by rfunk (subscriber, #4054) [Link] My colleagues and I use the most recent Ubuntu Live CD for that purpose. But those only come out every six months. If I want to get more fine-grained, there are places I can get .deb packages of the most recent driver versions for certain cards; installing those while retaining backups up previous known-working versions works for me.
questionable data Posted Jun 24, 2008 16:52 UTC (Tue) by joey (subscriber, #328) [Link] This article uses kerneloops.org as its sole data source, and tries to use that to compare the nvidia and ati kernel modules with the intel X driver. Doesn't that seem a bit of an apples-and-oranges stretch? I'd be highly suprised if kernel modules, binary or not, didn't produce more oopses than user mode code (even register-twiddling X code). It also seems less likely that any oopeses produced by user mode code would be as easy to track back to that code, compared with oopses produced by kernel modules, where the backtrace should make clear that the module was involved. I also doubt the effecacy of using oopses as a measure of how well graphics drivers work. My experience is that when graphics drivers fail on my laptops, they tend to fail on resume from suspend, or crah the X server -- but not produce a kernel oops. These failures still seem just as bad as a kernel crash to most users. I think a better indication of the problems of binary kernel modules is just that in the top 15 oopses on kerneloops.org, there are always a high proportion of binary kernel modules (and generally the same ones).
questionable data Posted Jun 24, 2008 17:30 UTC (Tue) by tialaramex (subscriber, #21167) [Link] All three (five, six?) 3D drivers** have a kernel module, so this is very much apples vs other apples. In theory (but sadly not always in practice) these drivers do all the dangerous low-level stuff like setting up DMA transfers and reacting to interrupts, and they should be bullet proof so that even a fairly serious bug in the OpenGL driver can't crash the machine. So bugs in Intel's code are just as likely in principle to cause an Oops as are bugs in nVidia's or ATI/AMD's code. ** nVidia's proprietary driver, ATI's FireGL driver, the two or three Free Software ATI drivers (2D stuff may not need the kernel driver), the Nouveau driver, and the Intel driver, plus possibly more. You're right that most failures appear as hangs to the user, either full system hangs or at least graphical hangs (and most users don't want to SSH into a machine and try to fix it). Some of those hangs cause an Oops, a WARN_ON or similar which may or may not be logged. Hangs are also not necessarily a reliable indication of driver error, some of this commodity 3D hardware is rather unreliable anyway, and it's not unheard of for there to be undocumented hangs where the driver did everything right and the chip froze up solid anyway due to a timing problem.
questionable data Posted Jun 24, 2008 17:51 UTC (Tue) by joey (subscriber, #328) [Link] Ok. I don't really know anything about the non-free graphics drivers, having never used any of them. For intel, are you referring to the intel_agp module? I've always considered this a fairly small and simple module and none of the intel video bugs I've experienced (and been happy to let Keith fix when he was nearby) have involved it. So I discounted/forgot about that module, apologies.
questionable data Posted Jun 24, 2008 18:45 UTC (Tue) by drag (subscriber, #31333) [Link] For current open source drivers that impliment both 2 and 3d drivers you need: * X.org DDX driver for XAA or EXA support for 2D acceleration * Linux Kernel DRM driver for allowing userspace drivers to interact with hardware. * DRI 3D driver. DRI is basically done by taking the Mesa OpenGL stack and trying to run as much as possible in the video card hardware. Then in addition to that you can have a Linux kernel VGA driver for virtual console access and/or Linux kernel framebuffer driver for allowing access directly to the memory on a video card.. used for virtual console on non-x86 systems as well as other things. Needless to say, having 4 or 5 different drivers from different projects trying to coordinate with each other to control one peice of hardware is pretty bad design. Of course all of this is slowly starting to evolve into different things. For example we are starting to see DRI2 drivers for Intel. A Gallium3D API is being developed for the ability to use a (relatively) simple device-specific driver to accelerate all sorts of application-level APIs. The DRM in-kernel driver is starting to take over mode setting and other responsibilities. The idea being that X.org will end up being a non-privilaged application and will use a generic API, just like any other graphical application, rather then have it's own set of drivers for twiddling with the hardware.
questionable data Posted Jun 25, 2008 7:13 UTC (Wed) by daniels (subscriber, #16193) [Link] The AGP module, oddly, only provides AGP services. The DRI wiki has <a href="http://dri.freedesktop.org/wiki/DRM">more information</a> on what DRM is, which you may find enlightening for future discussions about the kernel components of DRI drivers.
questionable data Posted Jun 24, 2008 19:00 UTC (Tue) by AJWM (guest, #15888) [Link] > You're right that most failures appear as hangs to the user, either full system hangs or at > least graphical hangs (and most users don't want to SSH into a machine and try to fix it). Is there any way to fix a hung X session other than restarting X? If you're a desktop user that kills all your work anyway, so you might as well just reboot. (And if you're not, why is X running? ;-)
questionable data Posted Jun 24, 2008 20:42 UTC (Tue) by tialaramex (subscriber, #21167) [Link] A full screen video game can hang without necessarily taking X with it. This is sadly not common with 3D games, it seems that usually the game is tied up waiting for DRI, and DRI is still waiting when X tries to clean up after the game exits, so you're still hung. But it has worked sometimes for me, so when I'm stupid enough to leave anything important running while trying to play a 3D game, I still try it just in case. I really don't get the impression that the vendors (3D chipset vendors I mean) care as much about this as they ought to. It seems that Microsoft has similar concerns. You can plug a cheap mouse, cheap scanner, cheap SATA drive and cheap sound card into your PC and, despite all the bargain basement gear in there, the expensive 3D card will be the most unreliable component by a mile. I have my doubts as to how much Free Software can achieve without these vendors losing a few hundred million dollars from a class action suit or a forced recall.
questionable data Posted Jun 24, 2008 17:56 UTC (Tue) by drag (subscriber, #31333) [Link] > This article uses kerneloops.org as its sole data source, and tries to use that to compare the nvidia and ati kernel modules with the intel X driver. I think it's fair. As a end user all I care about is my ability to use 2D and 3D features of my video card. Nvidia and ATI proprietary drivers, as well as the Intel drivers, all provide this to a different extent. You see the same problems with Windows. Microsoft had to recently disclose email communications under a court order because of the misleading term 'Vista Capable' for Intel GMA950 video cards. One of the emails had bug report statistics caused by driver errors grouped by vendor. Guess what vendor was at the top of the list? (hint: aidivN) You can find the emails here: http://blogs.zdnet.com/hardware/?p=1391 (there is a link to the pdf from that article) The information I am talking about is on page 47. Top 5: 1. Nvidia Corporation 28% 2. Microsoft Corporation 17% 3. Unknown 17% 4. ATI Technologies 9% 5. Intel 9%
questionable data Posted Jun 24, 2008 22:08 UTC (Tue) by dberkholz (subscriber, #23346) [Link] Is that scaled by total number of users on each card?
questionable data Posted Jun 25, 2008 0:04 UTC (Wed) by njs (guest, #40338) [Link] My guess is that it's not -- but that actually only makes things look worse for Nvidia/ATI, since Intel has larger market share. More problematic is that these numbers appear to be for all hardware, not just video drivers... but again, Intel has a more diverse range of hardware than Nvidia or ATI, so if we wanted to figure out the numbers for video drivers only, then we probably have to subtract a lot from Intel and not as much from Nvidia/ATI, and the gap would get larger, not smaller.
questionable data Posted Jun 25, 2008 18:52 UTC (Wed) by Los__D (guest, #15263) [Link] This is not quite fair. Maybe the Intel user count is higher, but I'm fairly sure that the nVidia GPUs get a lot more workout. Noone buys a computer with an Intel card to run 3D games. (Yet, but it seems like Intel is starting to move)
questionable data Posted Jun 26, 2008 2:56 UTC (Thu) by drag (subscriber, #31333) [Link] Intel dominates the video chipset market. Your looking at 2 Intel video card users for one nvidia card, easy. And you have Intel nic cards, intel wifi, intel audio, etc etc. etc. Which without a doubt are far far more popular then any Nvidia setup. Then you have ATI which outsell Nvidia also. At least last time I checked it was more popular with OEMs then Nvidia. Probably what is happenning is that Nvidia just had a very shitty driver release for a while there. These stats were pulled early on in Vista's existence and it's quite possible that bugs were fixed.
questionable data Posted Jun 29, 2008 10:29 UTC (Sun) by pointwood (guest, #2814) [Link] Yes, it's pretty well known that Nvidias Vista drivers wasn't quite ready when Vista was released. === My current laptop is a Thinkpad Z61t, specifically bought that model instead of the Z61m because it have Intel graphics and I've not had any problems in that regard. Kudos to Intel! I'll be building a new desktop PC soon (probably in august) and it will be with an ATI card (likely one of the new Radeon HD 4870), simply because I want to support ATI's good efforts with creating open source drivers.
Linux Graphics, a Tale of Three Drivers Posted Jun 24, 2008 22:02 UTC (Tue) by lovelace (guest, #278) [Link] This article appears to have a fairly serious error in it. It claims the nvidia driver on linux doesn't have all the functionality that the windows driver does. But, as far as I know, the nvidia driver shares a common codebase between their windows and linux drivers and has all the same functionality between them. So, if that is the case and the paper is wrong on that particular point, what else is it wrong on? Or, have the nvidia linux and windows driver codebases diverged and the linux driver does actually lag the windows driver now?
Linux Graphics, a Tale of Three Drivers Posted Jun 26, 2008 8:24 UTC (Thu) by nim-nim (subscriber, #34454) [Link] Some functionality depends on interfacing correctly with the rest of the system (for example xorg) and nvidia never did this correctly. It does not matter if the core driver code is capable of all kinds of fancy stuff if it does not expose the interfaces standard apps used to control this fancy stuff under Linux expect.
Linux Graphics, a Tale of Three Drivers Posted Jun 26, 2008 18:14 UTC (Thu) by Shewmaker (subscriber, #1126) [Link] In my experience, the nvidia drivers have excellent performance and many advanced features (i.e. framelocked stereo graphics) that are not available with other hardware/drivers. Nvidia cards are the only real option I know of for high end powerwalls. I've set up a cluster of workstations with nvidia quadro fx 4500s driving an array of 24 Christie Digital projectors at 1280x1024 at 100 Hz and the result was fantastic. I will say that I did initially have driver stability problems, because almost no one drives LCD projectors at that rate, but nvidia's technical support was responsive and fixed this corner case as soon they were able to replicate the problem.I would also prefer for nvidia to open source their driver, but I don't think the article gave their driver its due when it comes to features. Also, people should note that nvidia does provide source for certain things like their configuration utility, which indicates they aren't opposed to open source on general principles. Hopefully this article, people's responses to it, and the nouveau project will cause them to re-evaluate their belief that they gain more by keeping the driver development closed.
Linux Graphics, a Tale of Three Drivers Posted Jun 25, 2008 8:50 UTC (Wed) by daniels (subscriber, #16193) [Link] It's pretty unfortunate that the essay is so ridiculously sycophantic towards Intel and so dismissive of AMD/ATI's work in this area (their spec release predates Intel's), but that seems to be pretty much par for the course these days.
Linux Graphics, a Tale of Three Drivers Posted Jun 29, 2008 22:13 UTC (Sun) by johnh500 (subscriber, #49452) [Link] Intel does such an effective job of marketing that few people can get beyond the hype to see the real situation. The real situation is that the performance of their integrated graphics is many orders of magnitude worse than either ATI or NVIDIA's discrete graphics cards. You can't even run about 30% of the 3D games on Intel graphics, and don't even dream about pushing the envelope with really useful stuff such as powerwalls.
Linux Graphics, a Tale of Three Drivers Posted Jun 26, 2008 21:05 UTC (Thu) by caitlinbestler (guest, #32532) [Link] If you want a graphics adapter that "just works" as in "just barely provides a minimal display" then Intel chips are definitely what you should select.
|
|||||||||||||