🚀 go-pugleaf

"derek / nul" <abuseonly@sgrail.org> wrote in message news:2fu2tvgrd8adpqe9lktc8agh1ji61lrttg@4ax.com...
> On Sat, 6 Dec 2003 16:30:46 +1000, "Cristian Croitoru" <c_croitoru@hotmail.com>
> wrote:
>
> >> Me
> >> >Speaking of speed and not being particularly familiar with the SCSI
> >> >protocol, my common sense tells me that having a stripe on a IDE
> >> >multichannel controller can push the speed quite some way..
> >> >Again my common sense tells me that a SCSI controller can not stream data
> >> >from more than one drive at a time, as only one drive can put data on the
> >> >bus at the same time
> >"derek:
> >> Your common sense failed this time.
> >> With a SCSI ALL disks can read or write at the same time.
> >> That is one of the strengths of scsi
> >>
> >> Of course, at home you will have trouble using this strength.
> >
> >Well, I have tried to understand HOW, how is posibile for multiple devices
> >to send electrical signals on the same wire in the same time.. How is this
> >possible form the electrical point of view?
> >
> >With separate channels you have separate buses, separate electrical paths;
> >the data of the stripe is rebuilt in the controller's logic but essentialy
> >the logical path is 6 times wider.
> >
> >I am not trying my skills in retorics, I am just trying to understand..

> Yes, I understand, and I did not explain myself very well at all.

You mangled the story completely, and are still doing that.

> Let me say that all drives on a controller can be
> reading and writing at the same time, this does not
> mean that they are holding onto the buss at that time.

Utterly mangled all over again. You cant even have
two drives simulaneosly reading at the same time.

> Buss speed 160Mb/s, drive speed 20Mb/s
> therefore 8 devices r/w at the same time.

Wrong. It aint at the same time. Only one drive
can be transferring data at any particular time.

> All drives have buffers (double buffers actually), when a buffer is full, the
> drive raises a flag (data available) to say he wants to transfer some data,
> when the buss is free there will be a buss available signal.

Just as true with IDE.

> In effect we have a multiplexed system happening here,

Thats what ANY bus is.

> unlike an IDE device which holds the buss for the complete transfer.

Utterly mangled all over again.

You've mangling the entirely separate issues of say a
read request being issued, and the drive having to move
the heads to get the data, and hanging onto the bus while
that happens, with the entirely separate question of being
able to read and write simultaneously.

> I hope that sounds better.

Nope, its still completely mangled.

And as far as the original is concerned, IDE with a separate
bus per drive will ALWAYS be able to do much better than
SCSI with a single bus for all the drives.