Slashdot Mirror
Xig Ad Campaign Slamming Xfree?
San Mehat wrote in to point us to a full-paged Accelerated-X ad that has taken to some old fashioned mudslinging.
The most incriminating quotes are
"Buckle Up. If you're still using that free X server that came with your linux distribution, well hazardous conditions lie ahead" and
"When the X server 'falls over'--crashes--the entire operating system goes down and usually the user unfairly blames Linux itself'.
What do you think?
OK, I think we all have better things to do than complain about a lame ad. This post should really have been "What's better? Commercial X servers or XFree?" Let's just take a deep breath and realize that what the post is about is just tride and true commercial marketing. I think we're all smart enough to see through this fluff.
So after all those posts saying that how Linux is so much more stable then NT, it is now that we can surmise that well, yeah, the box is more stable, but only if you have a second console to be able to telnet into with in the case that X locks up the display...
You can also change to a virtual terminal to kill the X server, provided your keyboard isn't locked out, even if Ctrl-Alt-Backspace doesn't seem to work.
Let's not forget that LOTS of people have networks at home now. Telnetting in isn't impossible...you don't even need a second Linux box. Although, you are correct in stating that if X crashes AND locks out your keyboard AND you can't telnet in, you're pretty well hosed.
However, on my box, which is AMD K62 450 with a Riva 128 board, X quite rarely crashes. I had a few problems in the beginning when I didn't make a big enough swap partition, but after that, its smooth sailing. The only applications that can totally hose X on my system are Netscape and StarOffice, and in that case, I tend to place the blame on those applications, which are severely bloated and buggy.
X doesn't use all the available memory on some video cards
Well, in the case of XFree86, this is mostly due to the fact that good specs aren't available to the writers of the drivers, which often have to reverse engineer things to get a driver, or if there is no specific driver, the user is forced to use the SVGA server, which is pretty generic.
However, to play devil's advocate here, I'll make the case that you should really pick hardware that works well with your chosen platform, not the other way around. Since software is the reason you're using the computer in the first place, the software must dictate what hardware you will run, and not the other way around. Macintosh fans can scream 'til their blue in the face that their hardware platform is somehow "superior" to my generic Intel-based box, but since the software I need doesn't run on Macintoshes, they'll never persuade me to change my hardware.
The same holds true for peripherals. If you're using Linux, you're obviously not going to buy a winmodem. You can't fault Linux for not having winmodem support, you have to say well, since Linux doesn't support winmodems, I simply will choose not to buy one. If you follow the logic, then you'll have to say the same thing about video cards: if Linux (or in this case XFree86 in particular) doesn't support a given video card fully then it becomes obvious that you shouldn't buy that card until such support becomes available.
X is a lot more pickey about what monitor it runs on.
Not at all. If you're monitor isn't directly supported in the config files or by your favorite X configuration utility, you can always program the refresh rates yourself. X will work with ANY monitor your video card supports.
X crashes leave Linux in an unuseable state
Not always. Most of the time, if X crashes, I'm able to fix it.
Kernel prevents X from accessing memory it needs in order to run..
Well, if thats true, then that statement would apply to any X server, not just XFree86. However, a blanket statement such as this one is most certainly false. While I could forsee that in some instances, the kernel might not allow X to allocate memory, if this happened all the time X would be unusable. The fact that the majority of Linux users have no problem using X on standard configurations would seem to point to the contrary.
My journal has hot
I've been on the XFree86 developer's list for quite some time, and apparently this is nothing new.
Basically what's been said on the issue (since it's been brought up several times) is that the guy who started Xig (Thomas Roell I think it was?) is the one who wrote X386, which is the basis for XFree86 today. Apparently he is a bit miffed at the success of the project, and the apparent lack of success in getting ahead of it in his own project.
This generally tends to inflame people who are working on free software that is generally of higher quality than the corresponding commercial software, but they are bashed for their free work. XFree86 4.0 will be worlds above Accel-X in performance, modularity, and features. The thing is, XFree knows this, and so does Xig. The XFree policy is not to sling mud back, and not to post any kind of benchmarks (which often don't make sense anyway), but to simply let the consumer decide what they want to use. We're writing XFree86 for ourselves, and if other people get a good bit of use out of it, then we've more than served our purpose.
The main thing that companies like Xig and Metro-X have above XFree is that they can write proprietary drivers for cards where the manufacturers are too stingy with their specs to let the open source/free software people at it. At least Metro-X knows this, and they contribute code back and forth freely with XFree (including the new module loading system in XFree 4.0, which is pretty awesome).
Don't get all hyped up about this. It's nothing new. It's sad that it's happening in the way it's happening, but just wait for XFree86 4.0 to come out, and there won't be much mud left to sling except that tired old FUD that most people try to use in commercial vs free software. As Linus Torvalds said, "talk is cheap".
Cryptic Allusion - New Mac and Dreamcast Games!
Maybe what he's refering to there is that stock Linux kernels aren't capable of allocating large (like, 4-16 megs) contiguous physical chunks of memory.
X doesn't need this, the gui doesn't need it generally speaking.
It is however pretty helpful for some sorts of video streaming setups.
Currently there's a BigPhysArea patch that allows you to remove a large physical area from the pool that the kernel draws from, which you can then use in a userland driver.
It's certianly possible to advance the kernel to make this sort of thing possible, Linux isn't heavily video-oriented at the moment, But if you want bragging rights, compare X to the windowing system used in BeOS.
The only reason NT's gui gets everything it wants is because it's inserted into the kernel. That's a gargantuan trade-off in terms of robustness, and certianly nothing to brag about.
This is just like television, only you can see much further.
So after all those posts saying that how Linux is so much more stable then NT, it is now that we can surmise that well, yeah, the box is more stable, but only if you have a second console to be able to telnet into with in the case that X locks up the display...
Really! Just a little while ago (in the Gartner thread) there were all these posts saying things like: "Linux kicks NT's as in terms of stability", "Linux can dominate the desktop"... Face it, if X crashes, the system may as well be crashed if you're an end user.
Lets' just continue the list a little more, shall we?
X doesn't use all the available memory on some video cards.
X is a lot more pickey about what monitor it runs on.
X crashes leave Linux in an unuseable state
Kernel prevents X from accessing memory it needs in order to run...
And you wonder why Gartner says that Linux isn't ready for the desktop? Joe user (and me) doesn't want to worry about a kernel not wanting to give my video card the memory it needs to run. I just want to plug it in, maybe load some drivers and have it work...
-----------------
Go ahead, moderate me down... I've got karma to burn!
You mean, two competing products are.... competing????
Here is the text of the gigantic image:
Bumpty Ride
Buckle up. If you're still using that "free" X server that came with your Linux distribution, well, hazardous conditions lie ahead.
The X Server is the graphics sub-system in a Linux or UNIX installation. It is more than twice the size of the Linux kernel and much, much busier. Critical communications, fonts, drawing, windowing, mouse, keyboard, memory functions, and more all depend on the X server.
When the X server "falls over" - crashes - the entire operating system goes down. And usually, the user unfairly blames Linux itself.
To make your graphical Linux all that it can be, you need a commercial quaklity X server that's proven itself in thousands of mission-critical applications. An X server that delivers the full power of your graphics hardware to your LCD or monitor in the form of crisp, clean, and fast images. For all that, you need Accelerated-X.
Unmatched stability. Lightning-fast graphics. Superior performance. You'll find accelerated-X is like a fresh set of tires on brand-new blacktop. Want a test drive? Steer your browser to our website.
-konstant
-konstant
Yes! We are all individuals! I'm not!
Well, they are only seperate in name. The X server runs as root and has full access to the hardware on the system, it can even disable and enable interrupts whenever it likes. Thus, a bug in its low level driver code would be the same as a bug in the kernel. It could happily walk all over kernel data structures. Or even if the driver was ok and the code that called it was buggy and told it to DMA a chunk of video memory into kernel space, there would be nothing to stop it.
In my opinion, the major advantage Xi Graphics, Inc. has over the Free Software Community is their willingness to sign non-disclosure agreements which, though they effectively tie the product to a closed-source model, allow for a fuller featured and more hardware-specific X server, not a more stable X server.
Yes: it's corporate mudslinging at its finest, a vain attempt to differentiate themselves from the free "competition" by including terms that the public can identify with (like "stable", "secure", and "fast") that have very little technical merit. A page explaining NDA agreements and the politics of the situation wouldn't make a real convincing ad, and would prove that Xi is doing little more than attempting to translate a political advantage into an economic one.
My 2e-2 cents...
I have AccelX 5.0.2 for Linux.
It crashes on the SiS 5598 chipset at random.
XFree86 3.3.5 at least stays up.
I would not mind their add compain if it was
completely, but I have examples of the contrary.
XawTV has fits starting up under AccelX running
on S3Virge/DX, something about their
implementation of DGA is screwy. And the PS/2
MS IntelliMouse on my server works for only
minutes at a time, then the mouse pointer shifts
to the corner and any mouse movement results in
random pointer movements and button presses. I had
to go back to XFree86.
Click on the link for the ~371k jpg.
--
...as of AccelX 4.1.2 they still had a "dissappearing/corrupting" mouse cursor problem reminiscent of a MIT-SHM bug(very annoying when using Blender or the Gimp), couldn't use DGA extensions, and couldn't drop below 640x480 resolution. The only reason I ordered it in the first place was to support my i740 card, and when the XBF project produced an XF86 driver two weeks after I ordered it..
It _is_ faster than XF86, and you do get a nice splash screen, but XF86 is more feature-rich, IMHO.
Also, I could be mistaken, but it seems that the version of Accel-X I had didn't install any X source. When I compiled xanim for the first time, I had an error, and the author of xanim pointed out the problem--in the XF86 libraries that were still installed on my machine.
--
Unfortunately for xig, I bought MetroX, but I will vouch for Xfree86 "falling over". It happens people. It may not happen to you but it does happen.
When I was running Xfree on my Matrox Mystique card, I would have daily system lockups where the keyboard, mouse, and network all went down. Interestingly, linux really didn't, as proven by my still-running cron jobs, but the system was all but unusable (a solipsistic turing machine). It wasn't until I saw xig's ad that I put two and two together. Once I installed MetroX, I had uptimes of a month or more. Just goes to show you that you really don't want a buggy X server, whoever makes it.
I now have a different graphics card, and I'm back on XF86 (those vmware accellerations are hard to beat), but I wouldn't hesitate to go to a commercial X server if I started having problems again. Nor should you.
I've used AcceleratedX for 7 months now since they offered support for a particular laptop graphics chip that Xfree didn't. The installation/configuration process is nothing like Xfree. I had a working server config in about 30 seconds....whereas Xf86setup can still sometimes take an hour if i switch to a different monitor. ....
XiG tends to have support for chips 3-6 months before Xfree, and the server itself is much faster. Sure, $150 was a lot to pay, but if I waited for the open source version, I would have been carrying around a $2500 vt100
--BlueLines "The cost of living hasn't affected it's popularity." -anonymous
The XFree project's goal is to produce "a freely redistributable implementation of the X Window System that runs on UNIX(R) and UNIX-like operating systems (and OS/2)." Slamming them in an ad campaign is kind of a cheap shot -- they're volunteers producing a product because they think it's "The Right Thing To Do", not to compete in a commercial marketplace...
I think the point of the submission is that XiG is slamming a product that's provided gratis, free-of-charge, fo' no money, etc, and representing it like some third-rate software package when, for all experiences and reports, it's actually quite stable and useful.
And keep in mind, the XFree project has almost certainly been a driving force behind Linux' acceptance and popularity. Think about it: would as many people have been so likely to start using Linux on a regular basis if the only X server available was commercial and cost at least $100? I know I wouldn't have.
I wonder if XiG is worried that XFree 4.0 will come out and give them a real run for their money, for no money.
-=-=-=-=-
-=-=-=-=-
My mom's going to kick you in the face!
Just in case you wonder who these people are, their web page is at http://www.xigraphics.com.
We use xig's AcceleratedX 5.x to provide multi-headed support. However, it is not at all stable with Red Hat 6.1 (though it is fine with Red Hat 6.0 and Debian 1.2). If you want an example, try running kpackage, then slide the scroll bar for the package tree up and down. You'll be logging in and rebooting remotely in no time!
We use AcceleratedX because it works with our multi-headed hardware today, but we are keeping a very close eye on XFree and anxiously awaiting 4.0 and xenerama as a possible replacement down the road, with plans to switch when XFree 4.0 is stable and well tested.
AcceleratedX's strengths are its early access to hardware specs (providing support for hardware XFree doesn't) and the features it offers today which XFree doesn't yet (multiheaded support being the critical one for us). It is IMHO very foolish of xig to sling mud like this. Their market is comprised of Linux users, many of whom (myself included) take offense at that kind of ridicule against an Open Source project, especially when the ridicule is nothing more than FUD of the worst kind -- something Linux users and administrators are very good at seeing through.
The Future of Human Evolution: Autonomy
Competition by negative campaigning often backfires and, even when successful, can lead to a more apathetic audience, which can actually lead to fewer sales in the long run. (Voting figures are a good example of this. The years of negative campaigns run by politicians have carved away from the number of people who can be bothered to vote. Why bother? Much the same happened in the UK, during the Thatcher Years.)
The only way to be successful, and KNOW you're selling a good product, is to sell on merit, not deficit. If people buy your product, because they believe it does what they want, and keep buying, because it DID what they want, you have a good, long-term future. On the other hand, if you DO get people to buy, because they believe everything else is so much carp, you stand to lose your entire customer base when the competition shatters the illusion. They only have to do so once.
A company is never stronger than it's foundation. A foundation of bullshit and FUD doesn't offer much security.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
I won a copy of this program from Linux Today. It turned out to not be the laptop version (which I needed at the time)(Darn Neo-Magic) but it was the desktop version instead. All of my Linux machines have Matrox Millenium II's inside, and I was actually noticing more crashes on the machine with the XIG drivers. I noticed 2 crashes in the first month. (!) The machine was still accessible over the network (only for a reboot), but was unresponsive to any of the std. escape sequences. For comparison, I maybe crashed Xfree twice in a year previously, and the process was always killable with an escape seq. or a top over the network. After two months their CD was being used as a drink coaster. Add in the inconvenience of a less than intuitive setup, and no real performance benefit that was noticeable, and now I only reccommend their package for cards that are unsupported by the base distro.
I would also like to add that I did at some point in the proces need tech support, and I never got through on the telephone, and I never recieved an E-mail that was not an auto-responder. Now a newbie needing install support would not be able to get it from his Distro provider, and no response from XIG?? That sounds like a lose-lose situation.
In the interest of fairness, the Matrox acceleration is the best among the Xfree servers, and other people using ATI or such might experience different results. This experience that I relate happened ~1 1/2 years ago, so the state of their customer service and/or their drivers might be quite different. (But the the Xfree SVGA server has gotten better too...)
Other points of interest: Since Xfree SVGA is open source, you can diagnose and fix any driver problems yourself if you know how. But then I hardly think that this ad targets folks with know how... This is just another case of FUD.
Jason Maggard
hamnrye@mindspring.com
"I went mad for a while, It did me no end of good..."
-Z. Beeblebrox
I started with Accel-X when I upgraded to a new machine with a Matrox Mystique card. At the time, the XFree86 SVGA server was somewhat unstable with the Mystique and it lacked a lot of features and acceleration. I still have Accel-X 4.X with a Matrox G100 on my main computer, but I switched to XFree86 on my other box when I got a TNT card for it. Only Accel-X 5.0 supports the TNT and I haven't been compelled to spend another $50 for an upgrade.
Anyway, here are my impressions of the product:
- Accel-X is easier to install and configure than XFree86.
- It is significantly faster than XFree86 on most cards, and extremely fast with Matrox cards.
- It seems to be a little better at garbage collection, or is more memory efficient, or both.
- Accel-X is very stable with my cards. Of course, so is XFree.
Accel-X 5.0 also has an integrated TTF server which saves a lot of pain. Overall, I'd have to say that I've found Accel-X to be a better X server than XFree86. So if there was no cost involved, and no hassle of ordering and waiting for them to ship it to you, I'd easily choose Accel-X. However, XFree86 has improved quickly. Two or three years ago, they were so far behind in speed, features, and hardware support that I wouldn't consider anything but Accel-X. Now, I find that the advantages of Accel-X are not enough to compel me to upgrade.
Anyway, I wouldn't worry too much about Xig going under due to XFree86. They still have a performance edge, and much of their market comes from supplying X servers to other operating systems. Before Linux, they got into business selling replacements for the abysmal X server in SCO UNIX. Metro-X, on the other hand, is really going to dissapear quick. Accel-X was always much, much better than Metro-X and now even XFree86 is probably better.
This phenomenon seems to pop up with amusing regularity in the 'nix world. Every once in a while, some marketing genius comes up with the brilliant idea to jump on the popularity of the open source bandwagon by.......showing how their product is proprietary and and thus superior. It's really difficult to say if they are just clueless or somewhat thick.
Free hint to marketers - if you're trying to target free/open software customers, DON'T TRY TO IMPRESS THEM BY SAYING YOU'RE THE OPPOSITE.
Look at the above statement and think real hard. Repeat till done.
L.
One thing you may not be aware of re: bloat is that the mmapped video memory is included in the various reported sizes of the XFree86 process. I don't recall right off hand how much is allocated on the i740, but on the ET6000 16 MB is allocated, even though the chipset is physically limitted to a 4 MB frame buffer. This is because of the way that the hardware works.
Many i740 cards have 8 MB frame buffers, and there is probably some slop in there for MMIO. So that is responsible for a lot of the apparent bloat.
-sw