Linux on a Medion 9626 laptop

Last update : Oct 31st, 2002

Medion laptop picture +Tux the Linux Penguin

This page is here to tell my experience installing Linux on my Medion 9626 laptop.
I will try to provide useful information for future owners of this laptop, or people currently shopping for a Linux-supported one.

I suspect the 9626 was only ever sold in France, but similar machines probably exist under other names elsewhere in Europe, and the information found here can be useful to people with similar VIA686-based laptops anyway...

Oh, of course : I am in no way affiliated to Medion or any other vendor. Please get in touch with their customer service for official support (although they most likely still live on a Windows-only planet...). I'll try and answer non trivial questions but I can only do so much...

What's new (31st October 2002)

  • The MD563X modem chip seems to have a new name at Intel : MD5628D-L-A. Or at least that's what the new driver says, when you send the modem the "ATI3" command. Use the v92ham drivers here.
  • The "xine hang" problem is fixed in version 1.1.25t of the driver.
  • ACPI is getting better with every version : thermal zones now seem to function properly and adjust the fan speed automatically. Sleep mode and CPU throttling sometimes work, but are still unreliable.

Old news (6th April 2002)

  • The IRQ routing patch I used to mention here has been integrated in the ACPI project, so additional patching is no longer necessary.
    The network card in the MD9626 now works out of the box with the latest version (20020404) of the ACPI patch.
    The ACPI section has been updated to reflect this.
  • Power management works much better too, recent patches don't turn off the fan any more.
  • Many people reported that PCMCIA support indeed works very well.
  • The Video section has been updated to mention an annoying problem with Xvideo causing the X server to freeze.

Overview of the Medion MD9626

  • AMD Athlon 1 Ghz
  • 256 MB  SDRAM
  • VIA Apollo Pro KT133 chipset
  • Graphics subsystem : VIA/S3 Prosavage KN133 TwisterK, up to 16Mb shared memory. 14.1" TFT XGA screen
  • Integrated VIA82C686 sound, built-in speaker and microphone
  • RTL-8139C Ethernet 10/100 Mbits controller
  • Creatix modem ( Intel MD563X / MD5628D-L-A chip )
  • Hitachi DK23CA 20 Gb hard disk.
  • Matshita UJDA330 8x4x24x CD-RW
  • PCMCIA : 2 slots, OZ6933 Cardbus Controller
  • 3.5" floppy, 2 USB slots, IrDA, parallel, serial, PS/2, VGA out, TV Out

 Linux hardware support

Summary : With some work, every single peripheral is supported !

ACPI/power management used to be problematic, but it's improving daily as the ACPI project advances.

Overall, this is a quite good Linux machine, despite the nearly total lack of Linux support from Medion (I had to go to the store with a Mandrake install CD to find out about the chips and components by myself...).


The Savage Twister chip is supported in XFree86 4, with the "savage" driver . The chip id is "8d02" (TwisterK).
There is more information, and an utility to enable TV/VGA Out (tested, works ok), at the original driver page.


Supported both in OSS (via82cxxx module) and in ALSA (VIA82XX driver).


Supported by the 8139too kernel module (and Donald Becker's seemingly better drivers ).

However, there is a lack of support for ACPI IRQ routing with the current stable Linux kernels (2.4.19 as of this writing) that prevents the network card from working out of the box on this laptop. See more about ACPI issues below, and how to solve it.

Once again, this issue is unrelated to the network driver/chip, it's an IRQ issue, and I guess it could have affected any other device had the wiring on the motherboard been different.


The really great news is that even the internal modem is supported, something rare enough to be noticed among laptops !

Intel has made a mostly free/open source driver (use v92ham) that supports V.92 with all the bells & whistles (quick connect.
It works ok, although I've experienced short system slowdowns during modem initialization.


It works just fine.


Tested, works well with the standard IrTTY driver.

IRQ and ACPI Issues

ACPI is a specification that gives a standard way to set up device resources (IRQs for instance). It also deals with power management, and as such, it is a replacement for APM.

Many recent systems (especially laptops) need ACPI support from the operating system in order for vital components to work.

IRQ issues

Newer BIOSes, including the Phoenix BIOS on the 9626, rely on ACPI to set up PCI IRQ routing, and don't seem to support the older legacy interfaces.

But the standard 2.4.18 Linux kernel still uses legacy interfaces to set up IRQs, and in the case of the Medion, it fails to assign an IRQ for the network card.
The symptom is this message:

PCI: No IRQ known for interrupt pin A of device 00:09.0. Please try using pci=biosirq.

It will NOT help to use "pci=biosirq" as suggested.

The solution is to use ACPI to set up IRQ routing, which is exactly what the newest ACPI patches do.

The ACPI support included in the standard Linux kernel (2.4.18) is very limited, and not up to date with the latest developments of the ACPI project.
In order to have the latest ACPI code, you must apply their patch to your Linux kernel sources, recompile and install the new kernel.
If you don't know how to do this, see the Kernel HOWTO.

As of this writing, the latest ACPI patch version is acpi-20021024.

When you recompile the kernel with this patch, enable all ACPI options, and you may need to disable *APIC* (not ACPI) support too.
No kernel option is needed, the pci=acpiirq parameter that used to be required is now obsolete.

Power management/thermal issues

The ACPI thermal management code has improved a lot recently:
Unlike previous versions, the current one does not turn off your CPU fan on start-up, and thermal zones seem to work correctly : this means that when the CPU temperature goes over 60°, the fan will automatically go faster to help evacuate the heat.

I've also tested CPU throttling, which gives the possibility to control the CPU speed (and thus save battery life), and Sleep mode. Both have a tendency to crash the machine rather quickly, depending on the ACPI patch version of the day :(

Other ACPI functionality

ACPI can also manage the power button, the open/closed state of the lid, etc, through ACPI events.
You will additional user space packages such as acpid to bind an action (such as going into sleep mode) to an ACPI event.
I have not yet investigated how all this works. Tell me if you have more information.

ACPI Links

Here are some useful links for ACPI information:

Comments, Questions ?

I hope you found this information helpful.
If you have comments, suggestions, or additions, don't hesitate to send them to me .

Last updated  on October 31st, 2002

Back to my home page
I'm looking for a job. Here's my CV.