-= Ratz O. Fratzo =- said:
Drives on separate channels (controller 0 and controller 1) can be
accessed at the same time, where two drives on controller 0 (master
and slave) can not be accessed at the same time. It's a case of one
or the other. So in theory, it should be faster to copy from channel
to channel than from master to slave.
I believe this to be true to a point. If two devices are sharing a bus and
each device is capable of saturating that bus, then a slow-down will occur
if trying to access both devices to their fullest capabilities
simultaneously.
However, this isn't usually the case with optical devices as, even using two
on one controller, they are far from capable of saturating a modern IDE
interface (ATA66 or higher) and only really applies to HDDs. Most HDDs, even
if they are rated at ATA133, are incapable of saturating an ATA66 bus on
their own, other than for the milliseconds it takes to empty the drive's
on-board cache. However, two fast HDDs on one controller can result in
slow-downs if copying from one to the other or if both are being accessed
simultaneously at their fastest through-put. It is even concievable that a
fast HDD and a fast optical drive, both being utilised to their fullest (a
*very*, *very* rare occurance, once again cache related and lasting
milliseconds) could cause slow-downs on one controller.
What really irks me is the myth, continually perpetuated by people who can
sound credible, that puting a slow device and a fast device on the same
controller will reduce that controller to the lowest common denominator.
How many times a week do you see someone posting in one of the hardware
groups saying "Don't put your CD-ROM on the same cable as your HDD as it
will slow your HDD down."? Utter twaddle, and I get sick of refuting these
totally erronous statements. This hasn't been the case since around the dawn
of the 486 CPU era when PIO 1 (Programed Input/Output, version one) mode was
the bee's knees. These days, when virtually any drive/controller made this
century is capable of using DMA mode (Direct Memory Access, non-CPU
dependant), it matters not what goes where.
In the case of PCs (as opposed to servers), in 99.999% of the cases, it
doesn't matter in the slightest what drive goes where on what cable as the
situation where long, sustained full (device-capable) speed read/write
events occur is extremely rare. About the only time this occurs is when
copying large anounts of data from one HDD to another on the same channel.
(Note the 'HDD' part of that, optical devices, even two of them working
together, are incapable of saturating an ATA66 or better IDE channel) *And*
this only applies to straight copying, when encoding video or audio the
bottle-neck will always (in the forseeable fututre) be the processing time,
not the drive through-put.
The upshot of all this is, as long as your boot drive is master on channel 1
and your IDE controller is ATA66 or better, cable the drives however you
like.
Seriously, normally the primary IDE controller is the blue one