How come SATA is actually faster than PATA?

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

This is "I wonder how they do that" kind of post We've just purchased
two SATA drives for use in a RAID configuration. Working a treat. Nice cable


I wondered how come a serial interface can give such a high performance over
parallel? Last time I seriously messed around with UART serial devices was
back in my games writing days when I implemented a simple networking
protocol over the Atari Lynx serial lead. At this time, serial interfaces
were considered poor cousins of parallel. Kind of like why parallel
Centronics interfaces were faster than serial printer leads.

So what key change in technology has there been for serial devices (like
USB) to become so much faster than parallel? Parallel was always supposed to
be faster as you could (say) send down eight bits at the same time compared
to a single bit with serial.

I do have dim memory of reading something about parallel interfaces being
limited in speed due to interference between the data lines. That's why SCSI
ended up with so many grounds?

Just intrigued.

Thanks, Rob.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

> This is "I wonder how they do that" kind of post We've just purchased

Later... Found this article that explains it pretty well:

http://graphics.adaptec.com/pdfs/migration_to_serial_wp_maxtor.pdf

Cheers, Rob.

 

[cjt]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Rob Nicholson wrote:
>>This is "I wonder how they do that" kind of post We've just purchased
>
>
> Later... Found this article that explains it pretty well:
>
> http://graphics.adaptec.com/pdfs/migration_to_serial_wp_maxtor.pdf
>
> Cheers, Rob.
>
>
If it's so much simpler, why isn't it cheaper? ;-)

--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form che...@prodigy.net.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"CJT" <abujlehc@prodigy.net> wrote in message
news:419D2687.4000902@prodigy.net...
> Rob Nicholson wrote:
> >>This is "I wonder how they do that" kind of post We've just
purchased
> >
> >
> > Later... Found this article that explains it pretty well:
> >
> > http://graphics.adaptec.com/pdfs/migration_to_serial_wp_maxtor.pdf
> >
> > Cheers, Rob.
> >
> >
> If it's so much simpler, why isn't it cheaper? ;-)

All new computer prodiucts are more expensive.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Rob Nicholson wrote:

>>This is "I wonder how they do that" kind of post We've just purchased
>
>
> Later... Found this article that explains it pretty well:
>
> http://graphics.adaptec.com/pdfs/migration_to_serial_wp_maxtor.pdf
>
> Cheers, Rob.

That is a well-written article (admission: I know the author) and it
does a nice job of explaining why one particular serial interconnect
is better than another particular parallel interconnect, but it does
not say that serial in general is faster than parallel in general;
indeed, that is not true. Each of the problems Marty attributes to
parallel buses can be and has been solved (sometimes at high cost).

The original question (why is SATA faster than PATA) has a simple
answer: because the "ATA community" wanted to convert from parallel
to serial for its lower cost, and knew that it would be a hard sell
unless the initial SATA was at least as fast as PATA. So, they
stopped development on PATA, and chose a technology for SATA that
has a higher peak datarate than that final PATA.

In spite of how the above paragraph sounds, I believe that the
migration from PATA to SATA is good for everyone: the SATA family
of interconnects is fast enough for HDs and other storage widgets,
it has lower production cost, it has lower development cost, it works
over longer cables, it is more robust due to point-to-point topology,
and its thin cables result in better airflow.
--
Cheers, Bob

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"Bob Willard" <BobwBSGS@TrashThis.comcast.net> wrote in message news:Ailnd.117472$R05.7474@attbi_s53...
> Rob Nicholson wrote:
>
> >>This is "I wonder how they do that" kind of post We've just purchased
> >
> >
> > Later... Found this article that explains it pretty well:
> >
> > http://graphics.adaptec.com/pdfs/migration_to_serial_wp_maxtor.pdf
> >
> > Cheers, Rob.
>
> That is a well-written article (admission: I know the author) and it
> does a nice job of explaining why one particular serial interconnect
> is better than another particular parallel interconnect, but it does
> not say that serial in general is faster than parallel in general;
> indeed, that is not true. Each of the problems Marty attributes to
> parallel buses can be and has been solved (sometimes at high cost).
>
> The original question (why is SATA faster than PATA) has a simple
> answer: because the "ATA community" wanted to convert from parallel
> to serial for its lower cost, and knew that it would be a hard sell
> unless the initial SATA was at least as fast as PATA. So, they
> stopped development on PATA, and chose a technology for SATA that
> has a higher peak datarate than that final PATA.

So what exactly is the peak datarate of P-ATA and S-ATA?

>
> In spite of how the above paragraph sounds, I believe that the
> migration from PATA to SATA is good for everyone: the SATA family
> of interconnects is fast enough for HDs and other storage widgets,
> it has lower production cost, it has lower development cost, it works
> over longer cables, it is more robust due to point-to-point topology,
> and its thin cables result in better airflow.
> --
> Cheers, Bob

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

> The original question (why is SATA faster than PATA) has a simple
> answer: because the "ATA community" wanted to convert from parallel
> to serial for its lower cost, and knew that it would be a hard sell
> unless the initial SATA was at least as fast as PATA. So, they
> stopped development on PATA, and chose a technology for SATA that
> has a higher peak datarate than that final PATA.

I'm sure cost is a big factor in this decision. My take on the article is
that it's more cost effective to concentrate on signal processing techniques
as whilst they might be expensive, this less than the lost of x8 cells for
parallel.

I assume that dual-SATA would be twice as fast

Cheers, Rob.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"Rob Nicholson" <rob.nicholson@NOSPAM_informed-direct.com> wrote in message news:cnnojf$5vr$1@titan.btinternet.com...
> > The original question (why is SATA faster than PATA) has a simple
> > answer: because the "ATA community" wanted to convert from parallel
> > to serial for its lower cost, and knew that it would be a hard sell
> > unless the initial SATA was at least as fast as PATA.

Like trying to sell ATA100 as the successor to ATA133.

> > So, they stopped development on PATA, and chose a technology
> > for SATA that has a higher peak datarate than that final PATA.

Because it is the successor to ATA133, not an alternative.

>
> I'm sure cost is a big factor in this decision. My take on the article is
> that it's more cost effective to concentrate on signal processing techniques
> as whilst they might be expensive, this less than the cost of x8 cells for
> parallel.
>

> I assume that dual-SATA would be twice as fast

Then you assume false, whatever it is that "dual-SATA" is supposed to mean.
The only thing 'twice as fast' will be SATA-300 and only by using concentrators
such as port multiplyers to connect multiple drives to a single channel.

>
> Cheers, Rob.
>
>

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

> If it's so much simpler, why isn't it cheaper? ;-)

The prices now in the UK are very similar. We've just bought a Dabs
(Innovision I think) SATA PCI card for about £15. Two 300GB SATA drives came
it at about another £200. So for around £200, we've got 600GB of secondary
storage.

SCSI is still out in it's own little world We're looking at upgrading
the storage in our primary Dell file server and that's going to come in at
about £1000 for ~300GB. So SCSI is still roughly 10 times as expensive as
SATA. Okay, so that's a RAID-5 array with error correction.

Rob.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

In article <cnnoq0$p54$1@hercules.btinternet.com>, Rob Nicholson
<rob.nicholson@NOSPAM_informed-direct.com> writes

>SCSI is still out in it's own little world We're looking at upgrading
>the storage in our primary Dell file server and that's going to come in at
>about £1000 for ~300GB.

That must be a quote from Dell. Our distie does 147GB 10k rpm 8Mb cache
SCSI drives for 90 quid each.

--
..sigmonster on vacation

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Mike Tomlinson wrote:

> In article <cnnoq0$p54$1@hercules.btinternet.com>, Rob Nicholson
> <rob.nicholson@NOSPAM_informed-direct.com> writes
>
>>SCSI is still out in it's own little world We're looking at upgrading
>>the storage in our primary Dell file server and that's going to come in at
>>about £1000 for ~300GB.
>
> That must be a quote from Dell. Our distie does 147GB 10k rpm 8Mb cache
> SCSI drives for 90 quid each.

New? Froogling "147GB" gets a lowest price of $429.99 for a drive with no
factory warranty.

--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

In article <cnphun02ns1@news3.newsguy.com>, J. Clarke
<jclarke@nospam.invalid> writes

>New? Froogling "147GB" gets a lowest price of $429.99 for a drive with no
>factory warranty.

Yes, brand new and sealed, five year warranty, Fujitsu MAP3367NC. We've
been replacing the 9 and 18Gb HPaq hot-plug drives with them in our
rack-mount servers. Couldn't believe the price.

--
..sigmonster on vacation

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Mike Tomlinson wrote:

> In article <cnphun02ns1@news3.newsguy.com>, J. Clarke
> <jclarke@nospam.invalid> writes
>
>>New? Froogling "147GB" gets a lowest price of $429.99 for a drive with no
>>factory warranty.
>
> Yes, brand new and sealed, five year warranty, Fujitsu MAP3367NC. We've
> been replacing the 9 and 18Gb HPaq hot-plug drives with them in our
> rack-mount servers. Couldn't believe the price.

Are you sure about that number? With Fujitsu drives the last 3 digits of
the number are the capacity and the MAP3367NC is a 36.7GB drive. For that
drive, in the US anyway, 90 GBP would not be a good price--they're
typically going for around 75.

See
<http://www.fcpa.fujitsu.com/products/hard-drives/map-10k-rpm/specifications.html>

The 147GB SCA drives are MAP3147NC.

--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Bob Willard <BobwBSGS@TrashThis.comcast.net> writes:

> In spite of how the above paragraph sounds, I believe that the
> migration from PATA to SATA is good for everyone: the SATA family
> of interconnects is fast enough for HDs and other storage widgets,
> it has lower production cost, it has lower development cost, it works
> over longer cables, it is more robust due to point-to-point topology,
> and its thin cables result in better airflow.
> --
> Cheers, Bob

Fast enough has never been enough to everyone.
Lower production and development costs sounds reasonable but usually they
tend to disappear when someone makes some cool widget to get a little
performance gain, and looses compatability.
PATA also works "long" (ie. 90-120 cm.) cables. Is there longer than 60 cm
PATA-cables in sale?
Robustness in current state is worse because connectors slips out. Too
long leverage by long connectors and too weak attachment clips.
Round SATA-cables are in sale.

Jukka

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

> That must be a quote from Dell. Our distie does 147GB 10k rpm 8Mb cache
> SCSI drives for 90 quid each.

Which distributer is that? And yes, it's the price off the Dell website.
Didn't find a couple of other vendors but they were about the same price.

Cheers, Rob.

PS. If you have the existing caddies on a Dell, can you replace just the
SCSI drive itself?

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

> long leverage by long connectors and too weak attachment clips.

Attachment clips - what are they?

Cheers, Rob.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"Rob Nicholson" <rob.nicholson@NOSPAM_informed-direct.com> writes:

>> long leverage by long connectors and too weak attachment clips.
>
> Attachment clips - what are they?

Basically: (my) bad English.

Connectors which connect SATA-cable to motherboard and to HD.

Too many times had problems which were solved just reconnecting
SATA-cable.

Jukka

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

> Basically: (my) bad English.

No, just my poor attempt at a joke as to why we never get around to cable
tieing the cables in a case

Cheers, Rob.