SATA and ATA

[Folkert Rienstra]

He's just being his usual overly-pedantic self.

Yup, none of J. Clarke's eloquent 'Ignore Eric--he's in "full of crap" mode'
type of courtesy from me.

I can see him in a college somewhere making life Holy Hell for some poor
lot of hapless students who don't have the guts to give him a blanket party.

Weird fantasies you have. Too much bar cruising?

And what's with the siding with some nym-shifting troll by the name of "John
Doe", calling people troll at will, posting full headers with reply addresses in
them, recently trying to redo a subject that last time it was discussed resulted
in a thread with a massive 180 posts?

It's limited by the standard, not by any theoretical consideration. If
cross-talk was a hard physical limitation on parallel interfaces then U320
SCSI would not be possible.

And what's with the completely broken newsreader thingy, it makes quite a mess.

newssvr19.news.prodigy.com!newsdbm06.news.prodigy.com!newsdst02.news.prodigy.com!newsmst01a.news.prodigy.com!prodigy.com!newscon06.n
ews.prodigy.com!prodigy.net!border1.nntp.dca.giganews.com!nntp.giganews.com!green.octanews.net!news-out.octanews.net!authen.white.re
adfreenews.net.POSTED!not-for-mail

Reply-To: "Folkert Rienstra" <******************@******.**l> From:
"Folkert Rienstra" <see_reply-to myweb.nl> Newsgroups:
comp.sys.ibm.pc.hardware.storage References:
<[email protected]>
<[email protected]>
<[email protected]>
<[email protected]> Subject: Re: SATA and ATA Date:
Sun, 10 Apr 2005 01:19:25 +0200 MIME-Version: 1.0 Content-Type:
text/plain; charset="Windows-1252" Content-Transfer-Encoding: 7bit
X-Priority: 3 X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Lines: 20
Message-ID: <42588e2c$0$6058$892e7fe2 authen.white.readfreenews.net>
Organization: Read Free News
NNTP-Posting-Date: 09 Apr 2005 21:23:41 CDT
NNTP-Posting-Host: 14b168b3.authen.white.readfreenews.net
X-Trace:

DXC=OGh2mEEiNaMlSHlB]=`R^Ab_[jUcf=dBGPH=X44`;MPI[m3>;UVXR3OjiVWP31XC\@6>oZm1@7WSHUY
[HT0[D2BdcRG[dhc8PG[QUOTE]
X-Complaints-To: (e-mail address removed) Xref: newsmst01a.news.prodigy.com
comp.sys.ibm.pc.hardware.storage:348675

"Derek Baker" (e-mail address removed)> wrote in message "John Doe" (e-mail address removed)> wrote in message a.h. wrote:

What technical differences are between SATA and ATA?

SATA sends data one bit at a time down a single pair and
receives it the same way, vs 8 bits in parallel for PATA. The
SATA interface has a transfer rate of 150 or 300 MB/sec vs 133
for the fastest version of PATA,

Is that because of shielding on the serial cables? Otherwise I
can't imagine why the parallel interface would be limited.

[snipped]

The PATA interface is limited by crosstalk

Nope.

[/QUOTE]

 

[David Dyer-Bennet]

And you defragment daily of course.

No, but often enough that at least the software being loaded should
all be contiguous. And most of the old images; nearly anything older
than a month or two.

 

[Carlos Moreno]

Fraid not.

Care to give some reasons to support your argument?

From what I've read, what Nick was claiming is true. From
the fact that PCB layout gets much too complicated at high-
speed parallel transmission to the fact that electro-magnetic
interference between bits is a problem, and also to other
circuits, mainly when several of the (parallel) bits change
from the 0 to 1 or from 1 to 0 simultaneously (making the
radiated EM pulse much stronger)... It all seems to point
in the direction that Nick said.

I'm eager to hear arguments against or in favor of the above,
as I myself have been curious about this issue, and recently
found myself reading some technical literature on it (I got
to it from http://www.serialata.org, IIRC)

Carlos
--

 

[Carlos Moreno]

If I'm the one who wrote that I apologize--must not have had my coffee or
something--what I meant to write obviously was "One can buy 10,000 RPM
drives with SATA interfaces but not with PATA."

And I'm surprised that you're the only one that caught it.

Well, not the only one... I had already begun writing a message
saying "I don't know what surprises me more, what you wrote, or
the fact that no-one else has said anything" ... Then, I figured
it had to be my newsreader that wasn't showing me all the messages

(well, it wasn't -- it was me this time, apparently, who hadn't
had their coffee -- it was right there in front of me all the
time )

Carlos
--

 

[J. Clarke]

Care to give some reasons to support your argument?

From what I've read, what Nick was claiming is true. From
the fact that PCB layout gets much too complicated at high-
speed parallel transmission to the fact that electro-magnetic
interference between bits is a problem, and also to other
circuits, mainly when several of the (parallel) bits change
from the 0 to 1 or from 1 to 0 simultaneously (making the
radiated EM pulse much stronger)... It all seems to point
in the direction that Nick said.

I'm eager to hear arguments against or in favor of the above,
as I myself have been curious about this issue, and recently
found myself reading some technical literature on it (I got
to it from http://www.serialata.org, IIRC)

The simple argument can be summed up in one acronym. "SCSI".

The transfer rate on the SCSI bus, despite allowing 15 devices and an order
of magnitude longer cabling, is still 20 MB/sec higher than that of
SATA-II.

Sorry, but SATA is faster than PATA for one and only one reason. The people
who write the specs decided that they wanted it that way.

And I do wish that that particular bunch of idiots had given some real
thought to the connector design before they finalized the spec and stuck us
with one of the most fragile connectors ever made.

 

[Rod Speed]

Rod Speed wrote:

Care to give some reasons to support your argument?

Nick made the claim, so Nick gets to do that. Thats how it works.

From what I've read, what Nick was claiming is true.
Nope.

From the fact that PCB layout gets much too complicated at high-speed parallel
transmission to the fact that electro-
magnetic interference between bits is a problem,

Trivial to deal with that.

and also to other circuits, mainly when several of the (parallel) bits change
from the 0 to 1 or from 1 to 0 simultaneously (making the radiated EM pulse
much stronger)...

Its much more complicated than that when the bits
are just as likely to be opposed transition wise.

It all seems to point in the direction that Nick said.

Fraid not.

 

[Folkert Rienstra]

Not in the words he said it.
The potential for SATA 'to be' faster than PATA (in being the next
faster interface), is higher. PATA is at it's end, SATA is at it's start.
SATA-150 and PATA-133 are probably equally fast data wise after
all overhead is corrected for.

Not to speak of maintaining compatability with older modes of transfer.

Yeah, it may blow up the whole computer.

The simple argument can be summed up in one acronym. "SCSI".

The transfer rate on the SCSI bus, despite allowing 15 devices and an order
of magnitude longer cabling, is still 20 MB/sec higher than that of SATA-II.

Utterly clueless allover again.
Fast-320 transfer rate is 340MB/s higher than that of SATA-II.
And that type of comparison of course is completely flawed in multiple ways.
SATA-II 's clockrate is 'magnitudes' higher than Fast-320's.

Sorry, but SATA is faster than PATA for one and only one reason.

No it's not. You just can't compare the two just on clockrate.

The people who write the specs decided that they wanted it that way.

And sensible people they were, not to try defining another (parallel)
kludge but cutting the rope completely and go for serial.

And I do wish that that particular bunch of idiots had given some real
thought to the connector design before they finalized the spec and stuck us
with one of the most fragile connectors ever made.

The connector is probably fine for what it was designed for: Backplanes.
They should have gone for an alternative for drives used with cabling.

 

[Derek Baker]

The simple argument can be summed up in one acronym. "SCSI".

The transfer rate on the SCSI bus, despite allowing 15 devices and an
order
of magnitude longer cabling, is still 20 MB/sec higher than that of
SATA-II.

Sorry, but SATA is faster than PATA for one and only one reason. The
people
who write the specs decided that they wanted it that way.

More likely they decided it was easier - read cheaper - to make SATA go
faster.

 

[J. Clarke]

More likely they decided it was easier - read cheaper - to make SATA go
faster.

Unlikely given that the necessary technology was several generations old at
the time. I suspect that the decision was driven more by fashion and
marketing considerations than by any technological issue.

 

[Carlos Moreno]

Nick made the claim, so Nick gets to do that.

I wasn't talking about Nick's argument -- I was talking
about your argument -- you simply said "Fraid not", and
then repeated it. Saying it twice doesn't make it any
more convincing, so I was hoping that you could throw
some arguments to support your "fraid not" argument.

Carlos
--

 

[Rod Speed]

Rod Speed wrote:

I wasn't talking about Nick's argument -- I was talking about your argument --
you simply said "Fraid not",

Because his claim is just plain wrong
and he gets to substantiate his claim.

and then repeated it. Saying it twice doesn't make it any more convincing,

It wasnt intended to be convincing.

so I was hoping that you could throw some arguments to support your "fraid
not" argument.

It isnt an 'argument', its just pointing out
that Nick's claim is just plain wrong.

The most obvious problem with his claim is that
SCSI parallel does better than SATA speed wise.

There are certainly some real advantages with serial,
particularly in the cable and connector required, but
speed isnt one of them at hard drive speeds.

And while serial can certainly be faster than parallel,
particularly with optical serial, its rather silly to claim
that optical serial is less complicated than parallel.

 

[Nick]

It isnt an 'argument', its just pointing out
that Nick's claim is just plain wrong.

The most obvious problem with his claim is that
SCSI parallel does better than SATA speed wise.

And FC does fairly good compared to SCSI, even copper FC
http://custom.lab.unb.br/pub/x/docs/t10/t11/document.03/03-025v2.pdf

And while serial can certainly be faster than parallel,
particularly with optical serial, its rather silly to claim
that optical serial is less complicated than parallel.

Sure.

Nick

 

[Rod Speed]

And FC does fairly good compared to SCSI, even copper FC
http://custom.lab.unb.br/pub/x/docs/t10/t11/document.03/03-025v2.pdf

Sure, but there clearly isnt anything too
COMPLICATED about the fastest SCSI.

Yes, serial does have real advantages in the cable and connectors.

 

[J. Clarke]

And FC does fairly good compared to SCSI, even copper FC
http://custom.lab.unb.br/pub/x/docs/t10/t11/document.03/03-025v2.pdf

And when SATA starts running at fibre-channel speeds, then maybe that will
have some relevance. But in point of fact it's running slower than
parallel-SCSI speeds, which clearly demonstrates that it was not needed for
any technological reason.

 

[Carlos Moreno]

The most obvious problem with his claim is that
SCSI parallel does better than SATA speed wise.

This is certainly a good argument. But then again, perhaps
an equally relevant question would be: "all else being
equal, is serial simpler and/or potentially faster than
parallel?"

I guess there would be a considerable level of difficulty
in finding a sensible definition for "all else equal"...
Obviously, all else IS NOT equal as soon as we say that
we compare SATA with Parallel ATA...

My comment goes along these lines: while reading some
semi-technical literature on SATA, they claim that the
parallel technique, both for IDE and for SCSI, are about
to hit a "performance ceiling"... They talk about various
issues that sounded moderately convincing to me -- who
otherwise had always followed the "common sense" notion
that parallel *by definition* has to be potentially
faster than serial -- all else being equal, that is.
I mean, take 8 SATA cables and send them all to the
same hard drive -- it has to be 8 times faster, isn't
it? The embedded sync as part of the data can be used
to counter the alleged difficulty with parallel interfaces
that all the bits do not necessarily arrive at the very
exact moment. For that matter, send 8 or 16 fiber-optic
wires of the exact same length and there you go, EM-free
you-name-the-speed parallel connection.

I was just hoping to hear some convincing arguments in
favor of the parallel interface (well, or arguments
in disagreement with the notion that serial can be
potentially faster than parallel in this context).

Carlos
--

 

[J. Clarke]

This is certainly a good argument. But then again, perhaps
an equally relevant question would be: "all else being
equal, is serial simpler and/or potentially faster than
parallel?"

Why is that relevant to the issue of the reasons for adoption of the SATA
standard?

We are not talking about abstractions here but about a specific
implementation.

I guess there would be a considerable level of difficulty
in finding a sensible definition for "all else equal"...
Obviously, all else IS NOT equal as soon as we say that
we compare SATA with Parallel ATA...

My comment goes along these lines: while reading some
semi-technical literature on SATA, they claim that the
parallel technique, both for IDE and for SCSI, are about
to hit a "performance ceiling"...

This may or may not be the case. If it is, it is going to be a problem for
SCSI, which supports 15 devices on a bus and for which faster devices are
available, long before it becomes a problem for ATA, which, even before
SATA, only supported 2 devices on a bus.

They talk about various
issues that sounded moderately convincing to me

This is called, politely, "bafflegab" or less politely "burying 'em in
bullshit". Until they produce numbers all they are doing is waving their
arms and jumping up and down and hoping that someone will believe them.

-- who
otherwise had always followed the "common sense" notion
that parallel *by definition* has to be potentially
faster than serial -- all else being equal, that is.
I mean, take 8 SATA cables and send them all to the
same hard drive -- it has to be 8 times faster, isn't
it? The embedded sync as part of the data can be used
to counter the alleged difficulty with parallel interfaces
that all the bits do not necessarily arrive at the very
exact moment. For that matter, send 8 or 16 fiber-optic
wires of the exact same length and there you go, EM-free
you-name-the-speed parallel connection.

In point of fact this is exactly how so-called "PCI Express" is implemented.
It is serial only in the sense that the multiple data paths do not share a
single timing pulse.

I was just hoping to hear some convincing arguments in
favor of the parallel interface (well, or arguments
in disagreement with the notion that serial can be
potentially faster than parallel in this context).

To some extent it depends on how you define "parallel". Is PCI Express x16,
with 16 data lines, each independently clocked, serial or parallel? A
poing is reached where timing does become an issue with parallel
architectures. The crosstalk etc arguments are clearly bogus as
demonstrated by PCI Express, which with multiple data lanes is going to be
no more free of crosstalk than any conventional parallel implementation.

 

[Peter]

The most obvious problem with his claim is that

This is certainly a good argument. But then again, perhaps
an equally relevant question would be: "all else being
equal, is serial simpler and/or potentially faster than
parallel?"

I guess there would be a considerable level of difficulty
in finding a sensible definition for "all else equal"...
Obviously, all else IS NOT equal as soon as we say that
we compare SATA with Parallel ATA...

My comment goes along these lines: while reading some
semi-technical literature on SATA, they claim that the
parallel technique, both for IDE and for SCSI, are about
to hit a "performance ceiling"... They talk about various
issues that sounded moderately convincing to me -- who
otherwise had always followed the "common sense" notion
that parallel *by definition* has to be potentially
faster than serial -- all else being equal, that is.
I mean, take 8 SATA cables and send them all to the
same hard drive -- it has to be 8 times faster, isn't
it? The embedded sync as part of the data can be used
to counter the alleged difficulty with parallel interfaces
that all the bits do not necessarily arrive at the very
exact moment. For that matter, send 8 or 16 fiber-optic
wires of the exact same length and there you go, EM-free
you-name-the-speed parallel connection.

I was just hoping to hear some convincing arguments in
favor of the parallel interface (well, or arguments
in disagreement with the notion that serial can be
potentially faster than parallel in this context).

Try to read about a signal skew:
http://www.powerdesign365.com/Power_interconnects/Article29003.aspx

 

[Rod Speed]

Rod Speed wrote

This is certainly a good argument. But then again, perhaps
an equally relevant question would be: "all else being equal, is serial
simpler and/or potentially faster than parallel?"

That wasnt even being discussed. His original claim was
JUST that parallel is MUCH MORE COMPLICATED and
that that is the reason SATA is faster. Fraid not.

I guess there would be a considerable level of difficulty
in finding a sensible definition for "all else equal"...

Not really with a hard drive interface standard on that
question being discussed, whether parallel is actually
MUCH MORE COMPLICATED than serial.

Obviously, all else IS NOT equal as soon as we say that we compare SATA with
Parallel ATA...

Sure, but thats an entirely separate issue to
Nick's original claim that I chose to comment on.

My comment goes along these lines: while reading some semi-technical
literature on SATA, they claim that the parallel technique, both for IDE and
for SCSI, are about to hit a "performance ceiling"...

And that claim can certainly be substantiated.

It is no however the claim that NICK
made that I chose to comment on.

They talk about various issues that sounded moderately convincing to me -- who
otherwise had always followed the "common sense" notion that parallel *by
definition* has to be potentially faster than serial -- all else being equal,
that is.

Thats just plain wrong too. Parallel will always have
the problem that keeping the bits in synch becomes
more of a problem as the speed is increased.

I mean, take 8 SATA cables and send them all to the
same hard drive -- it has to be 8 times faster, isn't it?

All irrelevant when just one is quite a bit faster
than the drive physical characteristics set by
the rotation rate and the sectors per track.

The embedded sync as part of the data can be used to counter the alleged
difficulty with parallel interfaces that all the bits do not necessarily
arrive at the very exact moment.

Its more complicated than that with parallel.

For that matter, send 8 or 16 fiber-optic wires of the exact same length and
there you go, EM-free
you-name-the-speed parallel connection.

Sure, but a single fibre optic channel is far faster
than the speed determined by the drive physical
characteristics so its all completely pointless.

I was just hoping to hear some convincing arguments in favor of the parallel
interface

There arent any really with hard drives.

(well, or arguments in disagreement with the notion that serial can be
potentially faster than parallel in this context).

That is nothing like Nicks original claim.

 

[Peter]

Why is that relevant to the issue of the reasons for adoption of the SATA
standard?

We are not talking about abstractions here but about a specific
implementation.


This may or may not be the case. If it is, it is going to be a problem for
SCSI, which supports 15 devices on a bus and for which faster devices are
available, long before it becomes a problem for ATA, which, even before
SATA, only supported 2 devices on a bus.

SCSI has already extended to Fiber Channel, now going SAS.

This is called, politely, "bafflegab" or less politely "burying 'em in
bullshit". Until they produce numbers all they are doing is waving their
arms and jumping up and down and hoping that someone will believe them.


In point of fact this is exactly how so-called "PCI Express" is implemented.
It is serial only in the sense that the multiple data paths do not share a
single timing pulse.

Oh, short lived PCI-X.

To some extent it depends on how you define "parallel". Is PCI Express x16,
with 16 data lines, each independently clocked, serial or parallel? A
poing is reached where timing does become an issue with parallel
architectures. The crosstalk etc arguments are clearly bogus as
demonstrated by PCI Express, which with multiple data lanes is going to be
no more free of crosstalk than any conventional parallel implementation.

So on reassembly of multiple (serialy propagated) data lines there is no
signal
skew between them? How come?

Crosstalk was greatly reduced by a special signal coding technigue to reduce
noise generation.

But the biggest change is in the concept. Less hard wired signals, more
virtual wires which deliver information not by direct electrical state but
by
content of data passed through them. Message Signaled Interrupt is an
example.

All of those techiques called for a new standard(s). But serial
communication
is the real engine in all of them (PCI Express, SAS, SATA).

 

[Folkert Rienstra]

Why is that relevant to the issue of the reasons for adoption of the SATA
standard?

Duh! Because it is a future standard, maybe?
The one to follow-up ATA?

We are not talking about abstractions here but about a specific
implementation.

No, we're not.

This may or may not be the case.

If it is, it is going to be a problem for SCSI,

It is, but SCSI was well thought out from the onset.
IDE is a kludge.

which supports 15 devices on a bus

Not speed wise. That's only 3 or 4.

and for which faster devices are available,

long before it becomes a problem for ATA,

That's assuming ATA is upgradable in the same sense as SCSI.
That may not be so.

which, even before SATA, only supported 2 devices

on a bus.
Interface.


This is called, politely, "bafflegab" or less politely "burying 'em in bullshit".

Excuse John C. He got his education at the "school of hard knocks".

Until they produce numbers all they are doing is waving their arms
and jumping up and down and hoping that someone will believe them.

Hmm, isn't that exactly what you are doing?

I think that is how PCIe works.
Need more bandwidth, use more 'lanes'.

In point of fact this is exactly how so-called "PCI Express" is implemented.

Uhm, no.

It is serial only in the sense that the multiple data paths do not share a
single timing pulse.
Nonsense.


To some extent it depends on how you define "parallel".

Is PCI Express x16,
with 16 data lines, each independently clocked, serial or parallel?

Serial, definetely.

A point is reached where timing does become an issue with parallel architectures.

The crosstalk etc arguments are clearly bogus as demonstrated by PCI Express,

Nope.
PCIe(xpress) is serial which is more than just sending bits one after the other.
The data is coded such that minimal crosstalk is generated.

which with multiple data lanes is going to be
no more free of crosstalk than any conventional parallel implementation.

Nope.
The difference is in "conventional" i.e. no data modification to reduce crosstalk