[WBEL-users] That hard disk - again

[Scott Silva]

Pete Biggs wrote:
> On Mon, 2005-05-02 at 00:40 -0500, Edward Rudd wrote:
> 
>>On Sun, 2005-05-01 at 16:12, Francies Moore wrote:
>>
>>>Hi everyone,
>>>
>>
>>[snip]
>>
>>>The only different thing about this system is that the CD drive is the 
>>>primary master (/dev/hda) - could this setup cause problems?  The failed 
>>>disks were /dev/hdb and /dev/hdc.
>>
>>I always have CDroms as slaves to harddrives. It's better to have the
>>"faster" device the master.. and singles if possible.  
> 
> 
> The critical thing is that the IDE bus will only go as fast as the
> slowest device - so if you have an ATA-33 and an ATA-100 device on the
> same IDE bus, ALL transfers will be at ATA-33 speed regardless of which
> is master or slave.  CDROMs are invariably ATA-33, modern ATA hard disks
> are usually ATA-100 or ATA-133, so mixing CDROM and hard disk on the
> same bus will severely reduce the speed of the hard disk.
> 
> For single drives, have the hard disk on the first IDE bus and CDROM on
> second. If you have two hard disks and simultaneous access to them at
> full speed is not critical, then have them both on the first bus.  If
> speed is critical than either ditch the CD drive or get more IDE buses.
> If speed is really critical, then invest in SCSI!
> 
> P.
> 
I don't think the speed issue is true anymore, especially with the newer
ata specs. If both drives are DMA capable, then There should not be any
speed problem. But it is still true to have the faster drive be the
master, because the controller logic will be newer. Although ATA is
backwards compatible, an older slower device will cause problems.
And don't hang an old PIO zip drive or CDrom on the same channel as a
DMA hard drive. That will almost certainly cause a speed problem.

-- 
"If you have ever eaten crow,
It don't taste like chicken!!"