Slashdot Mirror


WYSE uses Linux for thin clients

Denovous writes "WYSE is apparently producing a new Linux thin client based upon the Slackware distribution, instead of Java as originally planned. The client can be started without a network, as the OS resides in 8.5 MB of memory. Check the article out here "

4 of 62 comments (clear)

  1. Re:Flash memory? How? by mmontour · · Score: 2

    I don't know how Wyse does it, but the Netwinder boots directly from a 1M flash chip, mapped as ROM rather than as a disk drive. The flash contains a small (~64k) boot loader + debugger and a compressed Linux kernel, and can also contain a compressed initrd image (handy for some 'rescue' situations where you've trashed your hard drive; with 1M flash there's only room for a couple of utilities).

  2. exploits by cxreg · · Score: 2

    The one thing that concerns me about embedded linux boxes like this is what happens when 3 months down the road someone finds a DoS for the kernel thats installed on these boxes. Even if its flashable, chances are there will be more boxes left un-upgraded than there will be maintainted boxes. I hope this doesnt turn a feature (open source) of linux into a fault.

  3. Re:quake over thin clients WILL work by BadlandZ · · Score: 2
    Maybe, maybe it will work.

    First of all I don't know anything about games, I don't play them. But that doesn't mean you can't run a big game on a thin client.

    You run the game on the server you connect to, not the thin client. The client only runs X. And OpenGL might be in XFree86 4.0, so if they wait for that, and used it, it would probably really rock.

    My work on thin clients has shown me that there are even some preformance gains IF you configure your thin client right. The thin client's CPU and RAM only runs X, and you want a rocking video card (From my experiance doing molecular modeling and such). But, your bottle neck is bandwidth, not video or ram. All the background calculations to tell the video unit what to render are done on the server, if the server is fast, that will be fast. If the thin client has a good video subsystem, it will render the graphics fast. Since the thin client has it's own CPU, and if the video unit is fast, it has the potential to blow away a "complete system" because one box does video, and one does the application, and your not bogging down any one system by doing both on it.

    The problems are, 1) you piss off other users that are on that server. 2) you frequently find the bottleneck is the data transfer rate between the server and the thin client (espically on 10baseT or less) 3) Most thin clients just don't have really rocking video subsystems.

    Now, at home, I have a Linux box that I call a "server" in a generic sence, that is basically my workstation. I also have a lesser linux box (good video, good monitor, 32M RAM, MilleniumII vid card), that I run just as a thin client off the server. I have found that some applications (particularly that use a lot of CPU time for the application, AND a lot of CPU time for X) are much faster on the thin client.

    Best thing to do, is just build your own. Pick up an P60 through P200, slap in a good video card and 8 to 32M ram, and just hang that puppy of a super fast server, and you'll be much happier, and probably pay less, than buying anyones "Thin Client."

  4. quake over thin clients by Mazurbul · · Score: 2

    Well, this looks like a really good system, because from the looks of this article, it looks like it is possible to play quake over this network, so I'm all in favour of it :) No 3dfx, but you can't have everything.

    I know I always make every article relevant to quake, but in the big picture, thats what really matters.