Economics of SATA hard drive

[kony]

As in every major chip manufacturer has one. I don't count the rebrands.
There may not be many rebrands in the professional market that PCIe is.
There never were with PCI-X either.

Let's put it into a real world context.
Newegg, world's largest online seller of retail hardware.

Look for PCI Express sound cards,
http://www.newegg.com/ProductSort/SubCategory.asp?SubCategory=57
45 PCI
7 USB
PCI Express? Nada

Newegg's selection of PCI Express NICs includes only 3.

Tuner/capture cards? 2

Regardless of what chips exist, what products are seen is a
reflection of the immaturity of the PCI Express addon card
market.

Nope, 75MB/s burstrate suffices (is sufficient) for a drive
with an STR below that to run without reduced performance.

That was only the very first link... sustained transfer
rates are also lower. I hope you didn't conclude it was
high enough performance based on only the one link because
plenty of people did have issues using PCI controller cards
in that era. Google for "Via PCI latency" or "Via PCI
controller", or here's another page, merely using the Intel
compatible version of the chipset.

http://www.tecchannel.de/ueberblick/archiv/401770/index7.html

With a single drive the issue was not exceeding the
available bandwidth of the PCI bus, it was the Via chipset.
It makes MOST single drives slower, they need not have a
sustained performance pushing the (theoretical max) PCI bus
limits. Also keep in mind that these benches were ONLY
focusing on HDD performance without concurrent use of any
other PCI devices. That will make matters worse.

Burstrate is the maximum obtainable STR between the drive and the
host interface, so your observations clash with the report you showed.

Call it a clash, it doesn't change the fact that in a real
world test the chipset and PCI card did have a marked
reduction in performance. The result was reproduced by MANY
other people at the time (contemporary KT266A users).

I have to therefor conclude that you just made that up.

Why would I care if you conclude it? If you had bothered to
search you would find the evidence and I already KNOW having
done the benchmark.

 

[Rod Speed]

Since any kind of testing would be of a synthetic
bench or real world app, not booting or running the
OS, it would not matter if the boot drive or not.

What matters is how the system is to use, not some benchmark.

The difference is there.

What matters is whether you can actually pick it in a proper
double blind trial. If you cant, its there but irrelevant.

Some may perceive it and others may not,

And some differences are so small no one can pick them in a
proper double blind trial, so the difference is entirely academic.

but some won't perceive the difference between the CPU they paid
for and the next cheaper one so does that really validate perceptions?

No it does not.

Yes it does. If you cant pick between the alternatives in a proper double
blind trial, you are wasting your money on the more expensive system.

A system is comprised of many subsystems. Each taken
alone may fall within a threshold of inperception but additively
each minor change will result in a system performance
increase large enough to notice by practically anyone.

Irrelevant when deciding whether a PCI SATA card will work
fine in that particular elderly system and so whether it makes
more sense for the OP to buy a SATA drive so that he gets no
constraints with the new system and can buy whatever is best value.

In short, paying more to use a SATA card
when it ends up SLOWER is madness.

That is just plain wrong if you cant even pick the slower in a
proper double blind trial. It makes much more sense to get
the SATA drive so your choice of new system isnt constrained.

 

[Rod Speed]

It is in fact slower. Were you paying attention to the
details provided in the thread?

Simple scenario:

System 1
KT266A motherboard
PCI SATA controller card
Typical budget grade HDD, 160GB Seagate SATA

System 2
KT266A motherboard (both systems same beyond drive and PCI card)
Southbridge integral PATA
Typical budget grade HDD, 160GB Seagate PATA

System 2 will bench faster, more than a single digit %
difference if the disk subsystem is a significant bottleneck
in whatever-the-test.

And if you cant even pick between those two configs in a proper
double blind trial, it makes much more sense to go the SATA drive
route SO YOUR CHOICE OF NEW SYSTEM ISNT CONSTRAINED
and so the new system will be completely bog standard with the
SATA drive as the boot drive in the new system and you dont have
to find and implement a PCI Express PATA card for the PATA boot drive.

 

[Rod Speed]

Rod Speed wrote

SATA DVD-writers are available from several
makers, certainly from Plextor, Samsung, and MSI.

Yes, but most of us prefer to go the ATAPI route with those,
because that give you lots more choice and much better value.

They seem to be typically around 75% more
expensive than the equivalent PATA device;

And there is no point in paying that.

but that will change, I'm sure.

But not necessarily by the time he wants to buy the new system.

In a hypothetical future system with only one
PATA connection the DVD is likely to be SATA,

You dont know that, or when that will be the case either.

so that PATA connection will be free.

You dont know that either. In spades if he decides
he wants a pair of optical drives for whatever reason.

DVD burners are now so cheap that it can make a lot of
sense to have more than one for convenient copying of
DVDs and when there is still quite a bit of variation in
how well particular burners cope with particular media.
It can make sense to have more than one just so you
can use any media that turns up with no hassles etc.

Nobody is claiming that SATA doesn't have a slight technical
advantage, but there is a price penalty for adopting SATA
today, and no persuasive argument to prefer it.

Wrong when you buy the SATA drive SO YOUR CHOICES
WITH THE NEW SYSTEM ARENT CONTRAINED AT ALL.

If I were building a new system (with a motherboards that
supported it) I'd fit a SATA drive (and damn the expense),

There is **** all expense difference now.

but for upgrading an old system with no SATA interface
I wouldn't think twice about buying PATA -- I'd just do it.

More fool you. Some of us prefer to have more choice
with the new system and going SATA provides that.

The chances are that that drive will have died,

Oh bullshit.

or its size will seem to laughably small
that there's no point in reusing it,

More bullshit. Depends entirely on how soon he will be getting the new system.

before PATA interfaces become so rare as to be a problem.

They dont have to be 'so rare' to be a nuisance, there are already
motherboards with just one PATA interface and that is a significant
limitation on what can be done conveniently on that system drive wise.

 

[Rod Speed]

Rod Speed wrote

You ignore his point that cables are quite manageable.

No I didnt. And I was JUST commenting on how likely it is
that anyone WILL BOTHER TO DEVELOPE A PCI Express
RAID card for PATA drives due to the cable mess WITH THOSE.

Period. Even with flat PATA cables.

Wrong, as always.

 

[Rod Speed]

There are a few, but "plenty"? I don't think so. Having a
select few cards for a given function is hardly a market
saturation. I am confident there will be multiple times as
many PCI Express cards available in the next few years.

Bet there will be **** all PCI Express RAID cards for PATA drives.

The market for PCI Express RAID cards will be quite small just
because most decent motherboards have optional RAID and there
are significant advantages with having that functionality built in, so
only a fool would bother to design a PCI Express RAID cards for
PATA drives when the main market for addon PCI Express RAID
cards will be those who need more than the motherboard RAIDs
provide. And they wont be interested in PATA drives because
of the cable mess alone, let alone the lack of hot swap etc.

It wouldnt surprise me if no one bothers.

The market tries to make $ in individual
cases, there will be cards. Wait and see.

Not interested in waiting, I'll go for the SATA drive now
and am absolutely guaranteed to be able to use it as the
boot drive in the new system with no farting around at all.

They're directed at systems exactly like the one the OP has.

Nope. The single PATA channel is for the
optical drives which are mostly still PATA.

I'm still in disbelief that this thread even exists, that
people are trying to make such a simple thing as
buying the drive type supported by the system,
an order of magnitude more difficult in the end.

Doesnt surprise me at all that dills like you are silly enough
to constrain the choices with the new system by getting a
PATA drive now when they have passed their useby date.

You cant even buy the most recently released Samsung
400G in PATA format and if you want one of their drives
because they are nice and quiet, you are stuck with the 300G.

yes it is EXACTLY what sufficient means,

Nope, that is not what sufficient means.

everyone does not have the same criteria.

No one ever said they do.

What is fast enough for one user may not
be for another, or another use/same user.

If you cant even pick the difference with a proper
double blind trial, its clearly sufficient for everyone.

Actually I've benched drives on KT2666/333 chipsets for
sustained rate too. Same drive is noticably slower on a PCI
IDE card (in this case it was a Promise FastTrack100).

And I bet the user wont even be able to pick it in
a proper double blind trial if it aint the boot drive.

This was a while back but vaguely it was a Maxtor Plus 8 or
9 and the figures were something like 35 MB/s on the PCI
card and 52MB/s on the motherboard's southbridge integral
controller. This was before even trying to do anything else
significant on the PCI bus like network transfers or audio,
with the latter known to be effected as well.

And I bet the user wont even be able to pick it in
a proper double blind trial if it aint the boot drive.

 

[Ed Light]

So with a pci card could he still make 20 Mb/S? That would be ok until he
upgraded his motherboard.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

 

[Ed Light]

You ignore his point that cables are quite manageable. Period. Even
with flat PATA cables.

Except flat cables can impede air flow and round ones are stiff and
sometimes have to be tied away from contact with heatsinks, etc.

But still manageable.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

 

[Ed Light]

Rug Spud doesn't have any more to say. His statements contradict
published objective testing results using real-world systems. He even
contradicts hiimself.

He's lost this round and proven he doesn't have any clue about hard drive
performance. He's probably just some 14-year-old kid.

It's a shame because this newsgroup has been very neat for years. Perhaps
he'll get totally filtered out.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

 

[Folkert Rienstra]

It is in fact slower.

Were you paying attention to the details provided in the thread?

What, the part that you made up?

Simple scenario:

System 1
KT266A motherboard
PCI SATA controller card
Typical budget grade HDD, 160GB Seagate SATA

System 2
KT266A motherboard (both systems same beyond drive and PCI
card)
Southbridge integral PATA
Typical budget grade HDD, 160GB Seagate PATA

System 2 will bench faster, more than a single digit %
difference if the disk subsystem is a significant bottleneck
in whatever-the-test.

What a load of gibberish.

 

[kony]

And if you cant even pick between those two configs in a proper
double blind trial,

Who said "you" couldn't pick?
That has not been established. So far the ONLY evidence
presented are the links I have provided that clearly show it
WILL BE SLOWER.

This is leap #1 you need to make the point.

it makes much more sense to go the SATA drive
route

Actually there is only one possible scenario where the SATA
drive makes the most sense...

IF:

1) User doesn't care about the performance, AND

2) User doesn't mind paying more for lower performance, AND

3) User does eventually install the drive in the next
system, actually DOES it, not just thinks about it, AND

4) Next system did only have one IDE channel instead of 2,
so only 2 device support, AND

5) User also wanted more than one other PATA device
installed, which would have to assume the first device were
a PATA ATAPI drive?? OR there is already yet another PATA
device being transferred, AND YET there is still the one
PATA position open even on the least # of PATA channels, 1.
Now consider that user's next board may have 2 channels, or
he may transfer an existing PATA PCI card and then will
have PLENTY of PATA channels.

So far we already have 5 specific conditions necessary to
benefit even the slightest bit eventually, but having to
suffer the performance penalty now no matter what. Then
there's PAYING to do that... Most people would want to be
PAID to do that.

SO YOUR CHOICE OF NEW SYSTEM ISNT CONSTRAINED
and so the new system will be completely bog standard with the
SATA drive as the boot drive in the new system and you dont have
to find and implement a PCI Express PATA card for the PATA boot drive.

You're out of your mind.
For all we know that next system will have a new drive
bought at that time and the current drive being considered
isn't used at all or in a secondary role.

The only thing known for certain right now is:

- Drive connected to the southbridge controller will be
faster, all else being equal (choosing same relative product
tiered drive in PATA vs SATA interface).

- System natively supports one but not the other and by
your own terms is a "dinosaur" which makes less and less
sense to be buying more cards for, particularly when
unnecessary.

I see you just don't understand, and I'm ok with that. I
have presented one option and leave it to the OP to decide
since we will clearly not come to an agreement on this.

 

[kony]

What matters is how the system is to use, not some benchmark.

YES!!

That is EXACTLY why the PCI SATA card is such a horrible
idea. In anything but the synthetic benchmark, the PCI bus
will be playing a more significant role in performance
degradation. The synthetic bench where only the drive was
being accessed is just the tip of the iceburg. By adding
the SATA card the OP could easily have OTHER devices that
are currently working fine, begin misbehaving in addition to
the slower but more costly new drive inclusion.

That is just plain wrong if you cant even pick the slower in a
proper double blind trial. It makes much more sense to get
the SATA drive so your choice of new system isnt constrained.

Nobody every concluded it wouldn't be possible. It "might"
not be possible, I did play devil's advocate for a moment
previously, but ultimately the only evidence is that it WILL
be slower, and that plenty of people DID NOTICE. There are
hundreds of thousands of search engine hits... not just
some random keywords that happened to show up on the same
page, but specficially about the Via PCI issue and it's
manifestation when using addon PCI cards like drive
controllers.

WHERE IN THE WORLD WERE YOU DURING THIS ERA??
It was common knowledge.

 

[kony]

So with a pci card could he still make 20 Mb/S? That would be ok until he
upgraded his motherboard.

IF he didn't mind paying more for lower performance, ok.
It still doesn't resolve the other issues though, for
example his sound may not work properly anymore, or some
other device on the PCI bus. "Maybe" some PCI latency
changes can fix it, but WHY would someone pay more to do
this?

Are we certain he is planning on pulling this drive out at
some later date and using it as the main boot drive on the
next new system? It could help to know if this is the top
priority above all else, and how long it will be until that
next system purchase.

If the next system purchase wouldn't be far off, it could
make more sense to go ahead and buy it rather than spending
on a PCI SATA card. if the next system purchase is a couple
years or more away, it may not make sense to reuse the drive
as the main boot/os drive later as right now we have a big
shift in the industry to perpendicular recording and that
may change performance a lot in the next couple years,
especially combined with the next-generation of drives with
giant flash memory caches on them.

 

[kony]

Bet there will be **** all PCI Express RAID cards for PATA drives.

We don't know that for certain, even if you don't assume as
I do that there will be some.

What you have also ignored is that boards STILL HAVE PCI
SLOTS. All the conditions you presuppose in order to make
your argument for SATA, when combined together become
incredibly unrealistic.

 

[chrisv]

No I didnt. And I was JUST commenting on how likely it is
that anyone WILL BOTHER TO DEVELOPE A PCI Express
RAID card for PATA drives due to the cable mess WITH THOSE.

Even you should be able to bullshit your way out of your predicament
better than that, Ron. The above shows that you clearly ignored the
real issue, that PATA cables are quite manageable, and proceeded to
rant about round PATA cables.

Wrong, as always.

Ridiculous and false claim noted. People have been managing PATA
cables for years, Ron.

 

[Timothy Daniels]

It's a shame because this newsgroup has been very neat for years.
Perhaps he'll get totally filtered out.

You'll eventually realize that most of the argumentive postings
in this one thread are by sock puppets, four of them so far, several
of them animated by the same guy. It's funny and amazing how sick
people can get.

*TimDaniels*

 

[Rod Speed]

Who said "you" couldn't pick?

No one.

That has not been established.

Thats why I repeatedly used the word BET.

So far the ONLY evidence presented are the links I
have provided that clearly show it WILL BE SLOWER.

Irrelevant if you cant even pick between them with a proper double blind trial.

This is leap #1 you need to make the point.

Wrong, as always.

Actually there is only one possible scenario
where the SATA drive makes the most sense...

Wrong, as always.

IF:

1) User doesn't care about the performance, AND

Wrong, as always. What matters is if the user can pick
between the two configs in a proper double blind trial.

2) User doesn't mind paying more for lower performance, AND

You aint established that the user does pay more
when you consider the total cost of the old and new
systems because buying a SATA drive now gives
much more choice with the new system and may
well allow a cheaper new motherboard to be used.

3) User does eventually install the drive in the next
system, actually DOES it, not just thinks about it, AND

Rather unlikely that he wouldnt given
that its the best drive available.

4) Next system did only have one IDE channel
instead of 2, so only 2 device support, AND

Thats the only one you did actually manage to get right.

5) User also wanted more than one other PATA device
installed, which would have to assume the first device were
a PATA ATAPI drive?? OR there is already yet another PATA
device being transferred, AND YET there is still the one
PATA position open even on the least # of PATA channels, 1.
Now consider that user's next board may have 2 channels,

He STILL gets more choice if he has a SATA drive.

or he may transfer an existing PATA PCI card
and then will have PLENTY of PATA channels.

Pity that the user has already said that he doesnt like
the fact that booting off a drive on one of those is a mess.

So far we already have 5 specific conditions necessary
to benefit even the slightest bit eventually,

No we havent, and they are certainly not ALL required either.

but having to suffer the performance penalty now no matter what.

But a 'performance penalty' that the user
cant even pick in a proper double blind trial.

Then there's PAYING to do that...

You aint even established that that happens either when
the total cost of the new and old systems is considered.

Most people would want to be PAID to do that.

Most people are completely irrelevant, what matters is the OP.

You're out of your mind.

You couldnt bullshit your way out of a wet paper bag.

For all we know that next system will have a
new drive bought at that time and the current
drive being considered isn't used at all

Wrong. The OP CLEARLY said that he wants
to be able to move that drive to the new system.

or in a secondary role.

Why would he with the new system ?

The only thing known for certain right now is:

Wrong, as always.

- Drive connected to the southbridge controller will
be faster, all else being equal (choosing same relative
product tiered drive in PATA vs SATA interface).

Irrelevant if the user cant even pick that in a proper double blind trial.

- System natively supports one but not the other and by your
own terms is a "dinosaur" which makes less and less sense
to be buying more cards for, particularly when unnecessary.

THE WHOLE POINT OF THE PCI SATA CARD IS THAT IT
GIVES YOU A LOT MORE CHOICE WITH THE NEW SYSTEM.

I see you just don't understand, and I'm ok with that.

Never ever could bullshit its way out of a wet paper bag.

I have presented one option and leave it to the OP to decide
since we will clearly not come to an agreement on this.

Yep, you've never ever had a ****ing clue on the basics.

 

[Rod Speed]

That is EXACTLY why the PCI SATA card is such a horrible idea.

You aint established that the user will even notice any speed difference.

In anything but the synthetic benchmark, the PCI bus will be
playing a more significant role in performance degradation.

You aint established that the user will even notice any speed difference.

The synthetic bench where only the drive was being
accessed is just the tip of the iceburg. By adding the
SATA card the OP could easily have OTHER devices
that are currently working fine, begin misbehaving in
addition to the slower but more costly new drive inclusion.

And the user may well find it makes no difference.

Nobody every concluded it wouldn't be possible.

No one ever said that.

It "might" not be possible, I did play devil's advocate for a
moment previously, but ultimately the only evidence is that
it WILL be slower, and that plenty of people DID NOTICE.

NOT WITH THE SPECIFIC CONFIG BEING DISCUSSED THEY DIDNT.

There are hundreds of thousands of search engine hits... not
just some random keywords that happened to show up on the
same page, but specficially about the Via PCI issue and it's
manifestation when using addon PCI cards like drive controllers.

WHERE IN THE WORLD WERE YOU DURING THIS ERA??
It was common knowledge.

AND YOU WILL FIND THAT I MADE SOME
COMMENTS ON THAT DURING THAT ERA.

 

[Rod Speed]

Rod Speed wrote

Even you should be able to bullshit your way
out of your predicament better than that, Ron.

Cant even manage its own lines, or anything else at all, either.

The above shows that you clearly ignored the real
issue, that PATA cables are quite manageable,

Like hell it does ON THE LENGTH OF THE CABLE
AND STILL STAY WITHIN THE ATA STANDARD.

and proceeded to rant about round PATA cables.

I clearly didnt, you pathological liar.

Ridiculous and false claim noted.

Never ever could bullshit its way out of a wet paper bag.

People have been managing PATA cables for years, Ron.

And plenty with a clue prefer to use SATA standard cables
instead of PATA cables that flout the ATA standard, ****wit.

 

[Rod Speed]

IF he didn't mind paying more for lower performance, ok.

Paying peanuts more in fact.

It still doesn't resolve the other issues though,
for example his sound may not work properly
anymore, or some other device on the PCI bus.

And may work fine too.

"Maybe" some PCI latency changes can fix it,
but WHY would someone pay more to do this?

SO THEIR CHOICES ARENT CONTRAINED IN ANY WAY
WHEN THEY COME TO BUY THE NEW SYSTEM, STUPID.

Are we certain he is planning on pulling this drive out at some later
date and using it as the main boot drive on the next new system?

Yes, he said that in his original post.

It could help to know if this is the top priority above all
else, and how long it will be until that next system purchase.

He wont necessarily know how long, depends
on what turns up thats good value etc.

If the next system purchase wouldn't be far off,
it could make more sense to go ahead and buy
it rather than spending on a PCI SATA card.

He's clearly decided not to do that.

And a PCI SATA card costs peanuts from the right place.

if the next system purchase is a couple years or more
away, it may not make sense to reuse the drive as the
main boot/os drive later as right now we have a big
shift in the industry to perpendicular recording and that
may change performance a lot in the next couple years,

Bet it wont. And it would make a lot more sense to
spend peanuts on the PCI SATA card in case it doesnt.

especially combined with the next-generation of
drives with giant flash memory caches on them.

You dont even know that he is interested in that class of drive either.