Tag Archives: linux desktop

The Impact Of KDE On 3D Gaming

Being discussed following the Ubuntu 12.04 Desktops Impact Performance, Power Consumption was the impact that KDE’s KWin compositing window manager (and others that don’t redirect fullscreen windows by default) has on the OpenGL gaming performance.

Depending upon the driver it can potentially cause a hit as shown in Wednesday’s comparison of Unity, Unity 2D, GNOME Shell, KDE, Xfce, LXDE, and Openbox. All of the desktop environments were tested in their “out of the box” / stock configurations on Ubuntu 12.04. The KDE aspect is being discussed in this forum thread where the usual items are brought up.

There’s interest in more KDE KWin directed vs. undirected fullscreen OpenGL rendering performance, so that will come in a separate article looking specifically at that item rather than mixing default and non-default test parameters.

However, in the meantime until the next set of results are complete, an independent Phoronix Test Suite user has conducted his own KDE graphics benchmark results and shared them on OpenBenchmarking.org.

Hit up 1204109-BY-KDE3DDESK93 for a number of gaming (Nexuiz) results under KDE with 3D on and off. Testing was under Fedora 16 with an AMD E-350 Fusion system and its Radeon HD 6310 graphics. The individual also used the Phoronix Test Suite’s system monitoring module to collect various system vitals while Nexuiz was being run.

More Phoronix.com benchmarks are on the way, but until then there’s other Phoronix Test Suite users sharing their data on OpenBenchmarking.org and taking advantage of new features.

by Michael Larabel

The Linux Desktop Debate: Setting Things Straight

By Tony Bradley, PCWorld

Over the weekend I wrote a post called “Why Linux on the Desktop Is Dead” which seems to have sparked some controversy among the Linux crowd. One of my PCWorld peers felt compelled to put together a rebuttal piece in defense of Linux, and point out that the Linux desktop death knell has been heard before at PCWorld, but people are still using Linux on the desktop…so there.


I’m not knocking Linux itself—I’m just saying it will never be a mainstream desktop OS.

After reading Katherine Noyes response to my post, and sifting through the various comments on both posts, I felt compelled to set a few things straight. There are some issues I have with the arguments in favor of the vitality of Linux on the desktop, and a clarification that needs to be made related to my article.

You’re Making My Point

What Ms. Noyes, and many of the commenters seem to not understand is that their arguments make my case for me. Commenters who talk about how it can’t be dead because they use it, or Ms. Noyes arguing that its premature to call it dead because it’s in use by millions miss the point.

Just because you’re one of the one percent—the hackers and hobbyists who love the Linux OS—doesn’t mean that it has any realistic hope of ever seeing mainstream adoption. It won’t.

Most people would agree that the Commodore Amiga, or the Betamax video recorder, or the Sega Dreamcast gaming console are dead. Saying that Linux on the desktop can’t be dead because people still use it is like saying that these things aren’t dead because there are still hobbyists and user groups that hack, tweak, and find creative and innovative ways to use them still.

One Percent…Still

I had almost forgotten about the 2010 post by Robert Strohmeyer. Ms. Noyes pointed it out in an attempt to illustrate how he was wrong then, and reinforce that my declaration of Linux on the desktop must be wrong now. But, what it did for me was emphasize my point.

In October of 2010 when that article was written, Linux had 0.85 percent desktop OS market share according to Net Applications. 15 months later, Linux has 1.16 percent desktop OS market share.

Ms. Noyes puts forth a claim that the Linux stats aren’t accurate because its an open source operating system downloaded for free—so we have no sales figures to go by. I would counter that sales figures are utterly meaningless, and that the methodology used by Net Applications is what really counts—how many actual people are actually using the OS to actually do things?

She also cites a ridiculously high Linux share stat from “The H”. I had to click the link because I’ve never heard of “The H” and had no idea what it was about. When I did, I discovered why it’s reporting an astronomical 25.36 percent share for Linux—it’s a site dedicated to open source, and it’s reporting share based on the operating systems people use to visit the site.

Counting Linux market share on a website dedicated to open source software is like declaring that Ford has 99 percent market share of the automobile industry by counting the number of Ford vehicles at a Ford dealership.

Now, if we want to start dicing up stats we can argue that Linux desktop market share has leaped by 35 percent since Strohmeyer wrote his article in October of 2010. Of course, by that same math we would have to note that it also dropped precipitously by nearly 26 percent in one month just between January and February of this year.

But, all of that would be silly really because it’s still the same basic one percent it always has been.

I Wasn’t Talking To You

If you’re part of the one percent who uses and loves Linux as a desktop OS, I wasn’t really talking to you. Obviously, you’re using Linux just fine. I even went out of my way to preemptively single out and congratulate you.

The article was written with businesses and IT admins in mind. Linux may be a great desktop OS for you, but for the vast majority of businesses out there—virtually all of them—deploying Linux as the mainstream desktop OS would be a tragically horrible idea.

First, it takes a fair amount of reconditioning if you’re already used to Microsoft Windows—which almost everyone is. I’m not saying Windows is better or worse, but right, wrong, or indifferent there would be a learning curve during which productivity would decrease, and support calls would increase.

Second, Linux is a poor desktop OS choice for business for many of the same reasons that Mac OS X is a poor desktop OS choice for business—it lacks the tools that businesses and IT admins need. I’m sure there’s some duct tape and chewing gum solution out there, or someone can argue that you could just develop tools in-house, but all of that is like swimming upstream when you’re used to having endless options.

Why Linux on the Desktop Is Dead

In the discussion that ensued on my Google+ page as a result of my article, one commenter made the point that Linux must be simple because his grandfather uses Ubuntu Linux as a desktop OS just fine. If he can do it, anyone can.


Linux is a capable OS, but too much effort for most businesses to adopt as the mainstream desktop OS.

When I dug a little deeper, though, I discovered it wasn’t quite that simple. I asked how his grandfather researched the diverse array of Linux distributions to choose the one that would best meet his needs, and whether it was truly easy enough for him to download, install, and configure the OS, and install the software he needed to make the system functional.

The answer was that the commenter—part of the one percent crowd of Linux hackers and hobbyists who love the OS and know their way around it—did all the dirty work and set up the system for him. In fairness, he pointed out that when you buy a Windows or Mac OS X system it comes with the OS pre-installed and probably some tools and apps ready to go out of the box.

I agree with that point. But, not everyone has a grandson who is part of the one percent to wade through the Linux distro chaos and choose one, then choose a specific release that’s deemed stable enough, then install it and configure it.

Even if they did, at that point it sort of becomes its own little “walled garden”. You can use the software and tools that have been provided for you, but to really use it as a PC, and make it do what you want you have to call your grandson back, or go through the learning curve to figure it out yourself.

If you try to seek out help on Linux forums, good luck. There are many Linux users who are patient with newer users, and generous with their time and knowledge. Unfortunately, they’re outnumbered five to one by indignant, self-righteous jerks who’d rather belittle you for not being a Linux guru.

Do you know what works? Buying a mainstream operating system that works with all of the mainstream hardware, and connects with all of the mainstream services, and uses all of the mainstream software.

I realize that is a bit circular—Linux can’t be a mainstream desktop operating system because its not a mainstream desktop operating system. But, that is the harsh reality.

Steven J. Vaughan-Nichols a technology of writer of ZDNet usually has some rather intriguing insights into the nature of linux and android, etc. This time he gives a more accurate account of his experience with Windows 8, sharing some comparisons to the linux desktop (Gnome 3, Cinnamon, KDE). What’s compelling is his initial summary, “I’m a Linux user, but I kind of like Windows XP, and I can get along fine with Windows 7, but Windows 8? Argh!” LOL Click on the title link and read on.

A Linux desktop and tablet user and Windows 8

Plugging X.Org GPU Hot-Plugging Into Mainline

Earlier today was the first round of comments by David Airlie regarding the finishing and up-streaming of his X.Org GPU hot-plugging support. This allows for new GPUs to be dynamically added to a running X.Org Server environment.

Airlie has been working on the GPU hot-plugging support for a number of months, but recently it’s finally come together. In particular, he’s been playing with the USB-based DisplayLink graphics adapters. David has written a DisplayLink KMS driver to top off DisplayLink’s other code and open-source support. This could also be adapted to work with dynamically hot-plugging/unplugging other graphics processors too.

It was in October of last year when the first phase of GPU hot-plugging was working, but the code for the most part wasn’t in mainline trees. The support is still out-of-tree, but now that the support has moved along and is in a working state, the agenda has turned to making this feature a reality for the Linux desktop.

In this mailing list post is where David Airlie outlines his current hot-plug plans. “So I’ve fleshed out most of a working solution for hotplugging and gpu switching but I’ve started wondering how to move ahead with upstreaming and finishing it off.”

Due to how David designed the hot-plug support on top of this old architecture, he’s asking for ideas from other developers how to merge the invasive changes. This is what’s currently being discussed as whether to take the routes of being disruptive, duplicating lots of code and renaming, or trying to share as much code as possible.

It will still be a while before you’ll find GPU hot-plugging being supported under Linux cleanly in an “out of the box” manner (aside from Fedora possibly carrying the bits early), but at least it’s moving along.

X.Org Server 1.13 is now next with the 1.12 release just days ago, but it’s not known yet whether this major graphics feature will make the 1.13 merge window. There’s still work on PRIME and other areas to make this desktop feature very compelling for the Linux desktop.

Additional details can be found in this earlier posting, along with a GPU hot-plug demo video under Linux.

Where is the Linux desktop going, and where should it go? This is a hot topic, and an important one. Unfortunately the discussion usually starts from the wrong premise, that the Linux desktop has only recently achieved parity with its Mac OS X and Windows cousins. Not so! The Linux desktop has been superior since its early days, and would have to go backwards to achieve parity.

Is the Linux Desktop “On Par” With Mac and Windows? No Way! ??

%d bloggers like this: