Slashdot Mirror
Kernel 2.6.31 To Speed Up Linux Desktop
Dan Jones writes "As the Linux community looks forward to another kernel release, the kernel hackers have been working on improving the memory management so that the X desktop responsiveness is doubled under high memory pressure. The result is an improved desktop experience. Benchmarks on memory-tight desktops show clock time and major faults reduced by 50 per cent, and pswpin numbers (memory reads from disk) are reduced to about one-third. Another improvement coming with 2.6.31 is kernel mode-setting support for ATI Radeon graphics cards, enabling faster user switching and a more seamless startup experience. Peripheral developments that will also improve the Linux desktop experience include support for the new USB 3.0 specification and a new Firewire stack. Even minor Linux releases have heaps of new features these days!"
Linus reads XKCD? http://imgs.xkcd.com/comics/supported_features.png
We won't: http://kerneltrap.org/mailarchive/linux-kernel/2008/7/15/2497614
The Break-In process of the new Linux kernel takes time. There is a significant change in Video Performance as the kernel break in. There is still a perception that Linux kernels have a short break in time or worse yet, don't require break in. Some hackers used a second computer to break in the Linux kernel, and transfer the image to their primary computer. This method will not appreciably reduce the break in time required for the kernel. Linux kernel Break-In must be done in the position where you plan to use it.
The System Performance Stages of the kernel are as follows:
* First Stage of Break-In = The system will feel very open, clear and with good detail resolution and dynamics. The greens and lower reds will have elevated intensity levels. The lower output of the blue and green information is due to reduced bandwidth performance at this Stage. In some systems, especially with aluminum or titanium heatsinks, the greens and blues may appear edgy or even fatiguing. The visual stage will appear OK with some lack of Focus. It will take from 5 to 15 hours of break-in for the kernel to reach the Second Stage of Break-In.
* Second Stage of Break-In = The blues and greens will appear less elevated and up front as the monitor intensity level increases. This is followed by the reds starting fill in. The lack of Focus may become more noticeable and the visual stage will start to widen and have more depth. It will take an additional 15 to 35 hours to reach the Third Stage of Break-In.
* Third Stage of Break-In = The system response time will completely flatten out. The presentation will become very clean and less up front. The lack of Focus is disappearing and the imaging will improve as will the low level detail resolution. The Green comes in and it is very tight and you will see lower Red frequencies than your other kernel provided. In effect the visual signature of the kernel will seem to disappear and the X-window presentation will be very real and non-fatiguing. It will take an additional 30 to 50 hours to reach the Final Stage of Kernel Performance.
* Fourth and Final Stage of Kernel Performance = The Visual Stage will be wider than your monitor with excellent depth, height and precise localization of individual icons on the desktop. The hue of the icons will be very accurate over the entire desktop. Symbolic links have excellent referencing ability. The metallic sound of your hard drive is very lifelike. Rhythm, Pace and Dynamics are effortless. Many users find they are now viewing the X-window system at lower Light Levels due to the effortless presentation. You will start to see subtle visual cues like the programmer turning his head while he is programming. You will find you are viewing the Window Manager and forgetting about evaluating your system.
Troll. But I'll bite.
X11 is a whipping boy for anyone who's ever had a complaint about a Unix GUI. No matter whether it's a badly-designed application, an unstable driver, or poor kernel scheduling, or a deranged toolkit drag-and-drop model, people always fault X11. And no matter what the root cause of the problem, the solution is always to throw out the X protocol and design something else. People like you fail to account for the possibility that there's actually very little wrong with X, and that it can certainly be the basis for a modern, functional GUI.
There was a very interesting comment on Slashdot a few years ago by Mike Paquette (who wrote Apple's Quartz) explaining why Apple didn't use X11 for OS X. The funny thing, in retrospect, is that every single feature mentioned in Paquette's post has now been implemented for X11, and that's with volunteer work. If Apple had invested resources into making this happen for X instead of reinventing the wheel, everyone would have been better off. Yet despite these additional features, we still retain full network transparency along with full compatibility stretching back to the 80s.
Don't confuse "newer" and "better". X11's architecture is quite good, and is among one of the better designs for a windowing system ever created. It's clean, extensible, fast, and network-transparent. It defines mechanism, not policy, and does its job extremely well. That it's been extended to support all kinds of modern features is a testament to the strength of its original design.
If it weren't for the soul-crushing stupidity, it'd be hilarious that people claim X is slow. X ran quickly on computers with 1/000 the performance of even a modest desktop system today, but it's slow on these modern computers? That makes no sense. People claim that X's network transparency puts it at a performance disadvantage, but neglect that Unix Sockets, used for local communication, are among the faster IPC mechanisms in existence. Criticism of X as a platform is baseless.