On Thu, 30 Dec 2004, Robert Epprecht wrote:

> >> > /dev/hda   hd
> >> > /dev/hdb   dvd writer
> >> > /dev/hdc   hd
> >> > /dev/hdd   dvd rom

bad idea .. esp if the drives are NOT the same ata-33, ata-66 or ata-100
or ata-133 speeds
        ==
        == you cannot mix devices with different ata speeds ( language )
        ==      huh, what?? what did you just say 3 clocks ago ??
        ==

both dvd drives should be on its own ide cable or at least on the
3rd ide cable probably connected to the extra pci controller or the
unused raid port on the motherboard
        - (slow) dvd drives should NTO be connected on the same cable
        as (fast) hd

> >> I have found mixing hd's & cd'd on the same IDE  controller to cause 
> >> issues. I have even had issues with the position of the drives on the 
> >> ribbon cable, depending on the UDMA number of the device. I would start 
> >> with hd's on one cable  & cd/dvd's on the other. Make sure your chipset 
> >> is supported by kernel. This may be old info as I have been running 
> >> scsi mostly.
> 
> I'm still hoping someone tells me this *is* old info...

it's 10 yr old info .. but it still holds ... clock pulses hasn't changed 

> > But then performance will take a hit.
> 
> That's the point. This computer is used as an audio workstation, so
> hd performance is very important. I don't care about performance
> of the optical drives, that's why I did put the hd's as masters
> on both channels.

if "hd performance" is important ...
        - is it tuned with hdparm ?
        ( like enabling U1 will usually increase performance by about 5%
        ( which is verifiable by having 2 identical system side by side,
        ( and the only difference being -U1 or -U0

        - you will get the fastest disk transfers with one disk per ide
        cable

        - you should also have tagged IDE turned on in the kernel

        - you can get a 50% - 100% performance thruput differences by
        using different motherboards ... 2x faster with some motherboards 

        - mirroring than stripping might make reading faster, but
        will be slower for writing

        - lots of fun tuning things

> It always did for me, only this time I can't make it work.

something changed ... :-)

        - like 2.4 to 2.6 kernels ...

> Do you mix hds and cd/dvd drives on the same ide channel?

sometimes it works .. sometimes it dies ... 
        - just depends on the drives, controllers and motherboards

> How can I tell the kernel/module that hdd should *not* be under ide-scsi?

in 2.4 .... append="ide-scsi=off" ??
in 2.6 .... i don't think you have a choice ?? unless oyu disable it to
        start .. donno

> Or how can I tell ide-scsi that hdd is not the same drive as hdb?

it already knows what kind of drive is connected to which ide ports

- if you force user defined options to the wrong ports, things will die

        in /etc/fstab .. make sure you have the right devices listed as
                /dev/cdrom

        in lilo.conf ... make sure you have the right options for /dev/hdb
                and /dev/hdd ..

        for /dev/cdrom .. make sure you link to the right devices
        for /dev/dvd..... ?? guess what ?? ( same as above )

        .. etc ..

        if dvd drive was on /dev/hdc and you moved it around... 
        you'd need to change all the links ....

c ya
alvin


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to