Linux GPU Performance 373
CrzyP writes "AnandTech.com has benchmarked the most popular graphics cards from ATI and NVidia on the Linux OS (SuSE 9.1). It is interesting to see that they have also written a custom benchmarking tool which can also be downloaded from the article. Take a look at Kristopher Kubicki's "Linux 3D AGP GPU Roundup" to see how each of the mid to high end cards performed on the Penguin flavored system."
Kristopher Kubicki (Score:5, Funny)
It's funny. Laugh.
Re:Kristopher Kubicki (Score:3, Interesting)
They seem to know something [kde.org] about improving performance. Try a recent KDE release, you'll be surprised.
Heading back to the topic, I've been very impressed with the Nvidia drivers when using SuSE 9.0. Fast, completely stable, and dead easy to install - and that's with the standard Nvidia installer. With 9.1, it's supposed to be even easier...
Re:Kristopher Kubicki (Score:3, Interesting)
KDE's performance is fine. What they need is better usability, better defaults, and most importantly, a better looking QT. Even when you theme QT and GTK with Bluecurve to make them look as identical as possible, open konsole then open gnome-terminal and place them side by side. konsole, like all QT apps, has needless bordering everywhere, whereas gnome-terminal, like all GTK apps, is so much cleane
Re:Kristopher Kubicki (Score:2)
My oppinion is I like QT, I like the way apps that use QT look compared to GTK, which is why I use KDE over Gnome.
Re:Kristopher Kubicki (Score:3, Informative)
>them side by side.konsole, like all QT apps, has >needless bordering everywhere,
funny, i see many more borders is gnome-terminal ?!?!
>This is why many notably amazing Linux apps
>(GIMP, gaim, evolution, hell even firefox)
>reject QT in favor of GTK.
gimp is the origin of gtk, i would be surprised to see gimp using qt.
gaim, evolution and firefox (and nautilus and abiword and
Re:Kristopher Kubicki (Score:3, Informative)
Really? Every time I have compared KDE and Gnome on a lowend computer (300-500MHz 128-256MB RAM) the result have been the same. Gnome was way faster than KDE.
What they need is better usability,
Maybe KDE is not perfect, but I don't know any GUI which is better than KDE.
better defaults,
I'm sure most people (including me) will agree with that. But that is all they could agree about. Because everybody want different settings, and no default will satisfy everybody. Better just
Except (Score:2, Interesting)
What good is good GPU performance if you have to run an old kernel to run the GPU at all?
I don't know (Score:2)
Re:Except (Score:4, Informative)
Penguin-flavored? (Score:4, Funny)
What does a penguin taste like, anyhow?
Re:Penguin-flavored? (Score:5, Funny)
Re:Penguin-flavored? (Score:2, Funny)
Re:Penguin-flavored? (Score:5, Funny)
Fat chicken.
KFG
Re:Penguin-flavored? (Score:2)
Re:Penguin-flavored? (Score:2)
Maybe Lola Granola would know...?
"Darn it, it doesn't soar!" -- Lola Granola
ATI vs nVidia (Score:5, Insightful)
Re:ATI vs nVidia (Score:5, Insightful)
Re:ATI vs nVidia (Score:3, Insightful)
During publication of this review, we received some information from ATI about some upcoming Linux announcements which they are working on. We will keep you informed of the details as we hear them.
Is it true? Is ATI finally getting it's act together with linux? Can linux finally become a game platform with everyone able to play?
Looking at past experiences, don't get your hopes up methinks.
Re:ATI vs nVidia (Score:2, Informative)
http://anandtech.com/video/showdoc.aspx?i=2044
It's a driver thing (Score:2)
Re:ATI vs nVidia (Score:5, Informative)
Re:ATI vs nVidia (Score:3, Informative)
Re:ATI vs nVidia (Score:2, Informative)
There wasn't Quake3 test in this review, but ut2004 was said to run faster. As for Q3, I've tested it myself: runs 5-6% faster.
Re:ATI vs nVidia (Score:2, Interesting)
Maybe they're built with super-mega-optimizing flags on linux that they don't use on Windows. Maybe it's the sound or input routines. Both were developed in linux, AFAIK.
My point is, it's true enough to say that Q3 performs better in linux, but that doesnt necessarily mean the OpenGL drivers for linux are better than Windows. I'd say that, at best, they'd be pretty much the same.
Re:ATI vs nVidia (Score:3, Informative)
Re:ATI vs nVidia (Score:3, Interesting)
Re:ATI vs nVidia (Score:3, Insightful)
"The only reason I'm not getting decent drivers from them is because their aren't enough of us gaming on Linux, not because we didn't purchase their expensive video cards."
Agreed. I have two (primary use) PC's at home. Number one is an XP2400+1Gb+GF4-4200/128 and it runs Windows. Number two is a Dual PIII/933+512Mb+GF4MX400/64 and it runs Linux.
The first one is obviously the better system for gaming, and in fact that is what I primarily use it for. It is good enough to run Far Cry with most settings o
...vs. same cards with Windows? (Score:5, Insightful)
Re:...vs. same cards with Windows? (Score:2)
I'd like to see that comparision as well, but with Linux using open source drivers on documented hardware. (I know that would disqualify a lot of nVidia and ATI chips from the test.)
Re:...vs. same cards with Windows? (Score:5, Informative)
My linux GPUbreakdown from independant tests: (Score:5, Funny)
NVidia: Sort-of
ATI: Kind-of
Are you saying (Score:3, Funny)
Better drivers and licensing please (Score:5, Insightful)
Comment removed (Score:5, Insightful)
Re:Better drivers and licensing please (Score:3, Insightful)
Chip Spec's would be better (Score:3, Insightful)
nvidia has always fscked around trying to get mangled source into XFree
ATI where unable to understand what this linux thing was but people who ati trusted wrote drivers...
solution - provide the chip spec's
frankly with CG and other GPU compilers the hardware should be patented if not then what harm is there ?
giving out spec with COPYRIGHT stamped all over it only supports you
for example those XGI people could capture the market by opens sourceing the drivers (they failed under SIS bec
Re:Better drivers and licensing please (Score:3, Informative)
There is a market. ATI has sold X cards. These cards, hardware wise, can be used identically on Linux or Windows, its only software thats needed to adapt to the system. If 90% of those cards went to Windows users and 10% went to Linux users, here is why the Linux driver is more expensive per unit sold.
Let's assume driver development costs a fixed number per OS, Y. Let's assume that its equal for Linux and Windows. Thus, every driver ATI develops costs them Y out of their profits.
The
Re:Better drivers and licensing please (Score:3, Informative)
I'm not sure what distro you're using, but with suse 9.1, no such contortions are needed. I simply checked the box in yast that says "install nvidia drivers" and a message popped up saying "nvidia drivers will take effect next time X is restarted".
The artic
Re:Better drivers and licensing please (Score:3, Informative)
Re:Better drivers and licensing please (Score:2, Informative)
If you don't have the kernel source that was used to build the kernel you're running, you'll need to either obtain it, or in many cases, obtain the latest kernel source and build a kernel to match it.
Wasting the space to have the kernel sources around is pretty sucky just to upgrade a video driver. I'm thinking small, set-top or embedded gaming machines, like something jammed into an arcade cabinet.
But, what the hell, I al
Linux Gaming, In Summary (Score:5, Insightful)
"High performance gaming on Linux certainly isn't for everyone. We spent weeks preparing for this analysis and we still ran into problems that we could not correct. So many times, we came to a solution for a problem only to find our Linux distribution had some files in a slightly different place or our file dependency tree was completely broken. These are the things that scare away people from Linux."
That is the 100% gospel truth. I couldn't have said it better myself. How then will the Linux community and game publishers overcome this (IMHO) enormous obstacle?
Comment removed (Score:5, Insightful)
Re:Linux Gaming, In Summary (Score:3, Interesting)
Which is why I believe the "desktop Linux" is doomed to fail, ultimately.
For servers, you want all of the customizability you can get. You want users who can tweak the most minute of details, because the system's performance and security depends on it.
On the other hand, that doesn't bode well for desktop applications, which rely on certain system assumptions in order to work properly. Microsoft's
Re:Linux Gaming, In Summary (Score:2)
Which is why I believe the "desktop Linux" is doomed to fail, ultimately.
But that's exactly why we have custom distro's to handle these things. Mandrake/Suse/Gentoo/etc. simply write wrappers (RPMs/ebuilds/etc.) around apps that handle the problem of tweaking each app to fit their particular distro.
Admittedly, they don't always do a fantastic job of this, particularly when close-source drivers are
Re:Linux Gaming, In Summary (Score:4, Insightful)
"Oh, but you maintain it for them!"
Sure I do. But I'd be doing the same damn thing if they ran Windows. Except then I'd be scanning for viruses, removing spyware, etc. There's really no difference.
As for consistency, that's a mute point. There already is a huge number of statically compiled apps which run in any distro. And apps that aren't are provided by your distro's package manager. Why is consistency even an issue at this point? Because you want a ubiquitous distro? Sure, that'd be nice but the world doesn't work that way atm.
If you're talking about ease of use and installation, I'll have sympathy for you. But once installed, the Quake3 tastes just as sweet. All it needs is a little popularity and the real world performance materializes nicely. Stuff like that is proof that Linux IS ready for the desktop and that people are just unwilling to change due to some legacy nonported application or just their unwillingness to learn something new.
To such people I say "what do you really need windows for?" In most cases the answer is nothing except proprietary games. In which case I convince them that games are not as important as running a proper and moral OS (in the sense of free and open as well as unpirated), or I at least encourage them to dual boot. (Even I do that.)
Besides, Linux gets more games all the time. There are only TWO games left that necessitates my windows install. When I stop playing them or they are ported, I will never maintain a windows install again.
You say Linux needs consistency to be a successful desktop OS and I say it needs time. I'm switching people. Friends of mine are switching people. Convincing people to drop the MS monopoly like the bad habit it is is a painfully one-person-at-a-time process.
But each person is worth it.
Re:Linux Gaming, In Summary (Score:2, Interesting)
Re:Linux Gaming, In Summary (Score:2, Interesting)
I'm running an nVidia card on Fedora 2 (x86-64) and the installation went flawlessly. After a few trivial X11 config changes to let the X server know about the new video card, DVI and accelerated OpenGL worked like a charm.
Odd (Score:2)
Re:Odd (Score:2)
Read the article for details on the problems they ran into; most of the problems were with ATi's drivers and specific bugs/glitches in games.
Re:Odd (Score:2)
As a Linux user, I have to tweak configuration files. Yeah, that's fine for me because I'm a geek. Try to get your mother to do that.
Why should you have to put your videoram in your X config anyhow? It can figure it out from the card!
Re:Odd (Score:3, Insightful)
Re:Odd (Score:2)
Re:Odd (Score:4, Informative)
NVidia isn't a problem under Linux because they actually put out decent drivers for it. But ATI support is horrid, because the company barely puts any effort into Linux drivers.
Re:Odd (Score:2)
And how'd you figure out how to do all of this? I would have *never* figured this out. I can't imagine a whole lot of other people would be able to figure this out, either.
Re:Odd (Score:3, Informative)
installing games (Score:2)
Re:Linux Gaming, In Summary (Score:3, Interesting)
Maybe if a site analogous to the Linux Counter were established, to count the number of Linux gamers, recording hardware, games played under Linux, games we wish we were playing under Linux, etc...
The root of the problem is a lack (though obviously it's not as bad as it was) of communication. So maybe if we could tell more companies that Linux gamers are numerous enough to be worth pleasing, we could get some res
Re:Linux Gaming, In Summary (Score:2)
Re:Linux Gaming, In Summary (Score:2)
"Linux" is a shitty platform for binary compatibility though, and all indications suggest that this will continue to be the case for the near future at least. Don't expect non-free games and other software binaries to work with "Lin
Re:Linux Gaming, In Summary (Score:2)
Unfortunately, the decision to buy new hardware constantly goes hand in hand with the decision to play some new game - and if it's a gaming machine you want, then Linux isn't the operating system that you need.
IOW, new hardware (especially graphics cards) is purchased in order to run software that can not adequately run on yesterday's gear. Even though the graphics card manufacturers do release Linux drivers, it's
We need... (Score:2, Insightful)
Until EVERYONE adheres to some sort of guidelines (HA! Yeah, RIGHT!), people are going to be dealing with oddball dependancies, kernel/driver issues, and filesystem layout annoyances.
Documentation all around needs to improve too, for both the Linux distros and the game makers.
Re:Linux Gaming, In Summary (Score:2)
They won't.
Just throw your collective hands in the air and admit that everyone should just use Microsoft products to play games.
Or Nintendo.....
Or Playstation.....
Re:Linux Gaming, In Summary (Score:4, Informative)
It's all Mesa's fault! (Score:4, Interesting)
Remember ALSA? It sticks its gun to GPL right down to the driver, and Creative actually donate SBLive driver for it, when the company was already crushing everyone else (Aureal included) sound card market! This should be how Mesa license the code, not the lame, bogus, xfree licence.
Re:It's all Mesa's fault! (Score:3, Insightful)
Slower? Please explain to me how you've come to the conclusion that NVidia's closed source drivers are slower than the open source ones, especially given that the open source ones don't support 3D acceleration at all?
"Penguin flavored" (Score:2, Funny)
No Americas Army? (Score:3, Informative)
On an nVidia MX 400 card, AA is playable and actually pretty fun online, but shadows are mostly chunks of squares on the ground. Otherwise, rpg's and smoke grenades look fantastic. I wonder why they didn't do comparisons of at least AA? I would think that's one of the first games people download for Linux especially because it's free.
Oh yeah, I had some original difficulty installing the nvidia drivers on a knoppix hd install with the 2.6 kernel, but I finally got it running well and documented the installation here: http://www.knoppix.net/forum/viewtopic.php?t=10314 &highlight= [knoppix.net]
What about DRI? (Score:5, Informative)
Eric Anholt's benchmarks of DRI on FreeBSD are here. [freebsd.org]
Roland Scheidegger's comparison of the three drivers available for the Radeon 9000 (DRI, FGLRX, XIG) is here. [hispeed.ch]
It's a bit surprising that the Radeon 8500 series is completly absent from this comparison. The 8500 and FireGL 8800 are still remarkable video cards.
Re:What about DRI? (Score:2)
Proprietary driver hell (Score:5, Interesting)
For a few months I ran Nvidia's proprietary driver but found that their support was poor. Countless people would report the same problem and Nvidia would basically just shrug and not even reply to the postings on their website. Stuff like "not our problem". They were very slow to support 2.6.
And as a gentoo user, I hated the binary installation program.
I finally dumped their stuff and went to the OSS driver. It is much slower, even when just opening new browser windows or xterms. But not having to mess with nvidia installer hell each time I gen a new kernel (which is pretty rare, actually) makes it worth it.
This was a great article, however, because it shows just how much chance and luck there is in getting these drivers to work. Buying the latest and greatest MB and CPU for use with Linux is still a huge unknown for the novice and experienced Linux user alike. And then there is the very real fear of whether it will work after you upgrade your kernel, etc.
Sad to see that Nvidia is the most Linux friendly vendor??
Re:Proprietary driver hell (Score:2)
Re:Proprietary driver hell (Score:4, Insightful)
Oh, little things, like the reassurance of being able to continue to use the hardware I've paid for even if nVidia don't feel like continuing to develop the drivers if-and-when the kernel API changes - like with the recent 4k stacks issue. That, and Free drivers are more stable that proprietary drivers in my experience, and when they aren't, you can look at the code to try to figure out why, rather than crossing your fingers and waiting for a driver update that may never come.
--
Re:Proprietary driver hell (Score:4, Insightful)
Oh, little things, like the reassurance of being able to continue to use the hardware I've paid for even if nVidia don't feel like continuing to develop the drivers if-and-when the kernel API changes
Yeah. You mean like my Zoran TV tuner card that hasnt worked since the 2.2 series, despite their being public specs and sources for it? Drivers are only maintained as long as the developer is around. And unless you have the skills to write your own drivers (and most of us, including large numbers of application developers, *don't*), the having specs/source or not is irrelevant.
...like with the recent 4k stacks issue...
You mean the "issue" that nvidia had *working drivers* for within weeks after it was *even an option* in the kernel? You mean the "issue" that "open" drivers like *ahem* ATI have and NVidia does not?
That, and Free drivers are more stable that proprietary drivers in my experience, and when they aren't, you can look at the code to try to figure out why
Again. you are in the vast minority in being able to do this. So don't bash NVidia for catering to the rest of us.
Open your wallet (Score:5, Insightful)
You mean like my Zoran TV tuner card that hasnt worked since the 2.2 series, despite their being public specs and sources for it? Drivers are only maintained as long as the developer is around. And unless you have the skills to write your own drivers (and most of us, including large numbers of application developers, *don't*), the having specs/source or not is irrelevant.
Why not offer to pay an open source developer to update it ? At least you have that option, independent of the manufacturer's support for doing so.
You going to have the same problem with Linux kernel version 4.0, when Nvidia don't provide a driver for their XYZ card in 2010. Problem is, at that time, you won't have access to the specs, so you won't even have the option of paying an open source developer to update the driver for you.
Standardization (Score:5, Interesting)
After both ATI and nVidia clobber each other with better framerates and better overall performance, I think that a new competitive advantage will develop... perhaps this may be better graphics quality or easier installs.
The real question is (Score:5, Interesting)
nvidia (Score:4, Informative)
You have to realise what you are asking here. The binary drivers that you are mentioning (which, by the way, never lock you to a certain kernel) are using the code made by ATI/NVIDIA to take advantage of their hardware's features. PixelShading, 3d processing. Each have their own way for their hardware to preform 3d functioning. This is not something they are going to disclose (they are a buisness too, remember?) and thus the open source drivers for video cards are always going to be horrid compared to the avalaible binaries.
Seriously though. Get a card that works great on linux despite the binary packaging. And I still don't see how it locks you into one kernel? Could you explain further in depth what you mean for me? Because upgrading nvidia (which must be done each time you update/replace your kernel) is as simple as shutting down the X server, rerunning the nvidia binary, and then rebooting the X server.
Re:nvidia (Score:5, Informative)
Besides, you didn't answer his question - he said "What's a good card with solid open source drivers?" You said "Nvidia has open source drivers but they suck, you shouldn't care about the binary only drivers."
I'd still advocate a Nvidia or ATI card. ATI makes regular code drops to the DRI and Mesa projects, and the open source drivers are of reasonable quality, and the nv drivers are high profile, with lots of work going into them. These cards are the most likely to see solid render acceleration in the future as XAA is replaced with a new acceleration architecture, so even with the Open Source drivers you'll see best performance with stuff like Composite (the basis of much of the X11 6.8.0 eye candy) with these cards.
Of the two, ATI and Nvidia, the open source drivers seem to be of roughly the same quality in my experience, but the Nvidia binary driver is far superiour to the ATI binary driver. ATI has got more bang for your buck, the GATOS project is working to support a lot of ATI's extra features, and ATI seems minimally more involved in the community with an eye to becoming moreso.
I think that pushes things solidly in ATI's favor if you're absolutely commited to the open source driver. If you're willing to use the binary driver, things become more even - it's ATI's price versus Nvidia's better support for the card under Linux/BSD
Re:nvidia (Score:3, Informative)
Correct, as it should apply to any kernel specific module. This isn't something to whine about, as each nvidia vinary wants to use your latest kernel headers or configurations. That shouldn't be a big deal, especially when the binary process takes about ten seconds.
>>And what about 4k stacks?
This was already fixed, oh a few months ago. About
Re:The real question is (Score:2)
If you don't care about 3D support then, it doesn't matter much which card you get. Basically any card from a major vendor will do just fine playing movies, displaying your desktop apps, etc. I recommend just buying something nice and cheap.
Re:The real question is (Score:2)
Re:The real question is (Score:3, Informative)
Re:The real question is (Score:3, Informative)
I am not sure how fast the ViaCLE266 [sourceforge.net] is, but it does not matter since it is a chi
ATI multi-monitor support a shocker (Score:2, Informative)
However, having read the article, ATI claim to have some Linux announcements in the pipeline - with any luck, maybe these drivers will allow me to use both my monitors with some decent 3d acceleration.
Anyway, to anyone thinking of get
Re:ATI multi-monitor support a shocker (Score:3, Informative)
That's still poor; I'm using the Free ATI drivers from the 2.4 kernel and XFree86 4.3.0, and I get about 800fps with glxgears (lousy benchmark, BTW). On my two year old P4 2.4 and Radeon 7500.
--
Why Linux sucks (Score:5, Insightful)
Any questions?
This binary driver thing has got to go. As Linux gains desktop market share, pressure will increase to open up the hardware interface to the driver. It's not like hooking OpenGL to the card involves any technology that isn't well known in the industry.
No... (Score:3, Insightful)
Little quirks (Score:3, Informative)
The ATI drivers are bad, but not *that* bad (Score:4, Informative)
While there *is* an "auto-installing" driver-package from ATI you'd better avoid that (unless they fixed a good number of bugs). Just running the package resulted in an error for me, googling around i found some hints and managed to install them: run the package in extract-mode, make manually, ignore error, make install accompanied by some messing with
there is also (for Suse 9.1) an rpm-package [suse.com]. Following the README in that path closely will get the video driver installed. Like nvidia ATIs driver combo too consists of a kernel driver and a n X-driver, and as usual the kernel-driver is a little fiddly to install. There is *no* (longer?) need to compile a custom kernel, you need to install the kernel source though (and really, read the README!).
Be careful though when configuring the XF86config. fglrxconfig is *not* a good idea since it asks you about mouse settings, monitor modes and whatnot, things that are running perfectly well and shouldn't be touched anyway. NVIDIA does a much better job just telling you the few lines you have to change in the config, fglrxconfig produces an XF86config-4 that is mostly useless and contains heaps of garbage.
To make the kernelmodule load automatically add two lines to the "modprobe.conf.local" (i think the first is unnecessary):
install fglrx
In XF86config load "glx" and "dri" in the Modules sections and put
Driver "fglrx"
Option "UseInternalAGPGART" "no"
in the device section. If you've got access problems put:
Section "DRI"
Group "video"
Mode 0666
EndSection
After restarting the X-server (twice to be sure, and check if the kernel module loaded) "fglrxinfo" should tell you something about ATI (and not Mesa), if that works do a "sync" for good measure and try tuxracer.
In my experience the nvidia-drivers are definitely easier to install, but it's really not impossible to get the ATI-stuff running.
Re:Indeed (Score:5, Informative)
I don't know about the ATI drivers, but this isn't true for the NVIDIA drivers. You can download an installer from NVIDIA that will create a kernel module for you and places it with the other modules. No need to recompile the kernel at all. Just load the module (if the installer doesn't do this for you) and restart your X server.
mod the parent up (Score:2)
There is -no- need to recompile the kernel unless they are referring to the old 2.6 kernel bug with the old nvidia drivers(the stacks bug). But that was a GLITCH that only lasted a week or two, and something quickly fixed (not to mention only really experienced by Fecora Core 2 use
Re:Indeed (Score:4, Informative)
All in all, it does work really well...until you upgrade or replace your kernel, and then X of course won't work. Many times you can run the installer again, though. Simple enough!
Re:Indeed (Score:3, Informative)
What's the beef with rebooting? (Score:4, Insightful)
The same with my slackware machine. About 25-30 seconds for a reboot. None of that bothers me because I -know- 30 seconds on my home machine doesn't mean a damn thing. I'll enjoy that time to rub my eyes, refresh myself with maybe spending that 30 seconds taking all the dishes out of the room back up to the kitchen.
Don't treat 30 seconds as a long and unbearable time unless you want to start complaining about having a manually flushing toliet in your home, followed by hands that can't wash themselves.
Re:What's the beef with rebooting? (Score:2, Funny)
Lots of good uses for GPUs (Score:4, Interesting)
Re:It doesn't much matter... (Score:2)
Runs at over 60 frames/sec at 1400x1050 with full texturing on a Sony laptop.
Re:Nvidia needs DRI???? (Score:2)
Both sets of drivers have to be compiled from within the source tree of whichever kernel version you're running, but you don't nescesarily have to recompile the whole kernel.
Not quite. (Score:4, Informative)
DRI lets X communicate with the hardware faster. X acceleration works without it, but not as well.
from Gentoo Hardware 3D Acceleration Guide [gentoo.org].
Re:plea for help.... (Score:4, Informative)
I played around with the xfree86.conf (I think that's the name) file, switching "nv" with "nvidia" and back again.
The binary driver is called nvidia. So switch it to that, and leave it.
Read the log file (/var/log/XFree86.0.log), look for lines with (WW) and (EE). This will go a long way to track down your problems.
the DVI output on my card wouldn't work.
Do you mean a second video port? Under Linux the second port (video output) is independent of the first in the XFree86 configuration, so you have to configure it to use it explicitly. Something like 'Screen 1' or 'Option "ConnectedMonitor" "DFP" ' in the Section 'Device' should do it.