Asus P5Q Deluxe ESATA port hard drive size limit.

My esata port works great with my 1tb drives but will only show my 2tb
drive when the drive is powered up at boot. The 1tb drives can be turned
off without a problem. If I boot with the 2tb powered it will show but if I
turn it off then back on it will not. It is not an enclosure problem as I
have tried a couple. Any suggestions? I search Google and cannot find any
info on an esata port drive size limit so I am puzzled. Thank you.

 

Consider the hardware path being used.

I think it's a Marvell 88SE6121 (pretty ordinary)
followed by a SIL5723 (not so ordinary). Something
along these lines ?

--- 88SE6121 --- 5723 ---- drive
---- drive

http://www.siliconimage.com/products/product.aspx?pid=103

http://www.siliconimage.com/images/products/5723_front.jpg

Trouble is, there are no statements that I can find, about
device capacity. There is a compatibility table, with a list
of disk drives, but it isn't up to date (doesn't have
the drives you can buy today listed). The largest drive in
the table appears to be 1TB, but that could simply be
a function of the date of publication, rather than a limit.
In fact, a few pretty small drives, are listed as not
compatible. So the table is useless for predicting success.

If this was just a port on your Southbridge, it probably
would have been a different story. I think there is an
actual processor and firmware inside the 5723.

http://www.siliconimage.com/docs/sii5723_pb_FINAL_11-28-06.pdf

This isn't the first time this has happened. Asus put one of those
damn things on another motherboard, and there were no end of
questions about that one too. You'd think Asus would have
learned their lesson and stopped using those things. They're a waste
of silicon, on a motherboard. There are plenty of better things
they could use instead. Even a SIL3132 would have been better.
I'd rather have a SIL3132 than a 5723, because it supports
port expansion if you can find an expander for a decent price.

Paul

 

I don't see a reason why you can't leave the drivers in place. It's probably
not going to hurt anything.

On some of my motherboards, I turn off hardware in the BIOS, in the interest
of speeding up the boot. If you notice the computer wasting time checking
for drives on those ports, then that may give you an incentive to turn
that stuff off. Otherwise, the lazy approach, is to just leave it

Paul

 

Thank you for the reply Paul. BTW, I am using XP 64 bit and can't find the
size limit for either version for the esata so I assume it's probably 1TB
because my other drives work fine. If I stop using the marvel and use usb
only would you recommend I disable it in the bios or uninstall the esata
driver? Thx.

 

Paul.

I agree with you 100%. My external drives are used for backup only and if
usb 2.0 is a bit slower than esata it won't matter. I like usb because the
lights on the external drives work showing activity and when esata the
machine shows the activity which can be condusing if I want to shut the
external down but not sure if it's still in use or the main machine drive
is in use.

Would you recommend configuring the externals for quick removal or the
higher performance option? Thank you for all the help, it's appreciated.

Robert

 

If your computer is on a UPS, and the computer is connected via a USB or
serial cable to the UPS (controlled shutdown interface), then I'd select
the High Performance option, if that's what you want. My UPS is set up
that way right now - it sends a message to the computer, when there is
limited time left on battery (after a power failure), so that Windows
can shut down cleanly, flush the cache on the disk drive and so on.

The Quick Removal thing might be more appropriate if the computer was
connected directly to wall power, with no UPS. Since the computer
has very little time in the event of a power failure there, I doubt
any cache could be flushed. Quick Removal then would mean less
damage to the file system. NTFS is journaled, so the next startup
could help resolve any fragments left over. FAT is more exposed to
incidents like that.

I presume there are rules for what options you'll be offered, and
you might not always have both options available to you.

There is a table here, for example. And some accompanying text to
describe what options may be offered, and ways to override them.

http://www.uwe-sieber.de/usbstick_e.html

I haven't honestly looked at this when using my USB hard drive. I slap
a drive into the enclosure and use it occasionally, then remove it later
(because normally the enclosure has a DVD writer in it). And don't really
leave the thing unattended. If I was to leave it connected all the time
with a hard drive in it, I should be paying more attention

Paul

 

Excellent info Paul, thanks much. I also want to get your opinion of
converting my P5q Deluxe to AHCI. There is plenty of info regarding how to
do it without having to reinstall windows. There are .reg files etc
involved and I have no problem with that but what I don't know if I would
get any performance increase or decrease with AHCI? I have heard stories of
longer boot times etc. Would it be worth converting in the long run or
would you just leave it be? Thank you.

Robert

 

AHCI is good for two things.

1) Supporting hot plug of drives. Useful for an ESATA device for example.
Or if you use one of those SATA dock devices. I'm pretty careful with
my hard drives, and like to only handle them with the power off, so
there is less chance of mechanical shock to the drive while it is
spinning. If your setup gives you good control over that (only
handle drive when it is not spinning), then Hot Plug support could
be a useful thing.

2) NCQ support. This is useful for server loads, where multiple pieces of
software are sending commands to the drive at the same time. With NCQ,
the drive has the ability to re-order the commands, for less head
movement. If you're a single user on a desktop system, this might be
less of a benefit. AHCI might actually add a second or two to some
operations, as there is some overhead involved. It's a positive thing,
if you have lots of queue buildup in the disk queue (pending commands
building up in queue, drive can't keep up, multiple storage threads).

If you wanted one ESATA port, the easiest way to get it might be to plug in
an add-in card (like a card with a SIL3132), and then install an AHCI driver
for that. So that, there is less disturbance to any choices you'd already
made when installing the OS (i.e. changing Southbridge mode).

Yes, there are undoubtedly recipes out there, for dynamically switching.
In particular, Windows 7 makes it dead easy to change to AHCI (re-arm via
registry change). The previous OSes are more of a nuisance, and the recipe
is a lot more complicated. Certainly, you can do a "Repair Install" on an
older OS, offer an F6 driver early in the process, and change to AHCI that
way, but then you'd have to do all your Service Packs and Security Updates
over again. It works best, if you prepare a new installer CD for yourself,
using NLite, and slipstream in the Service Pack level you're currently using
on the OS. (So the installer CD matches the service pack level of the current
OS installation.) That eliminates one step, during your Repair Install.

"Integrate a Service Pack"
http://www.nliteos.com/guide/part1.html

If doing a Repair Install, read up on the caveats with respect to Internet
Explorer. The latest versions are not tolerant of repair installs. You're
supposed to uninstall things like IE8 and try and return the OS to the
level of Internet Explorer it was using when it was installed. That
wasn't always the case, as there was a time where the repair install
would take care of IE for you. But now, you can muck up IE if you leave
an advanced version present, and Repair Install over top of it.

Now, that advice doesn't explain, how a person repairing a "broken OS" is
supposed to remove IE, before doing a Repair Install. And that just shows
you how dumb Microsoft is. If the OS is broken, there might be no opportunity
to remove IE, and then there could be trouble when you try to Repair Install
your way out of the situation.

Repair Install doesn't change your installed applications, or delete your
email. It does affect the Service Pack level, and remove the Security
Updates, such that a visit to Windows Update afterwards, and an hour
of downloading is still required.

As always, do a complete backup before you push the button, so you
have an escape route if the Repair Install doesn't work out. The same
would be true for one of those recipes that involves extensive registry
changes.

I just shut down the PC, connect my SATA device, and use the disk that
way (with no hot plug). For a lazy guy, that's less work than the above.

Paul

 

Excellent advice Paul. I ahven;t decided if I want to try an AHCI recipe as
yet because hings are working including thr esata port provide the drive is
not larger than 1tb because my 2tb would not work on esata port.

I gorgot to ask you before. I have a couple new WD 1tb drives and the one
that I was attempting to connect to esata before getting the driver issues
fixed shows many Ultra DMA CRC errors with HD Tune. If I understand
corectly it is not a drive problem but a problem that the controller had
communicating with the drive, correct? CRC errors displayed in SMART data
could be due to a bad usb cable etc. Am I correct regarding this? If a
drive has a problem connecting to the controller because a cable is faulty
a CRC error can occur. I am hoping that it's not a drive issue as the drive
is brand new and not a refurb. Nothing to worry about I hope? Thanks.

 

I found some info here. SMART is apparently documented in the ATA spec,
according to the article.

http://en.wikipedia.org/wiki/S.M.A.R.T

"0x01 Read Error Rate ... is often not meaningful as a decimal number"

Now this looks more interesting. Mine reads out zero in the data field value.

"0xC7 UltraDMA CRC Error Count

The count of errors in data transfer via the interface cable as
determined by ICRC (Interface Cyclic Redundancy Check).

My two SATA drives, have quite large stats for 0x01, but they show zero
for the 0xC7 UltraDMA error count.

Both ends of the SATA cable, need to check for errors, like this. The SATA
cable has a separate diff pair for each direction, and the recipient has to
check for errors. CRC is calculated over the data in the packet. It's similar
to say, an Ethernet packet concept. I don't know how re-transmissions are handled,
or whether the protocol is "reliable" or has acks piggybacked on data packets
or the like. The SMART stat, would be a measure of the counter on the
right hand end of this diagram. I don't know where the counter for the left
hand end is kept (it would be hard for SMART to do it).

+-------------+ +-------+
| Motherboard | ------------------> (error check) | disk |
| | (error check) <-------------------- | |
+-------------+ +-------+

Based on that skimpy info, if your 0xC7 data value is non-zero, it implies
a cable (or chip on either end of the cable) issue. A pinched SATA cable
could do it. Running the SATA cable next to a source of high amplitude interference
could do it (next to the flyback on an old TV set, next to the ignition wires
on your car, etc).

HTH,
Paul

 

My Seagate 1.5tb boot drive has one Ultra DMA CRC error and I am not sure
how long it's been there because I didn't check smart data until recently.
I am not going to worry about one but my two external 1tb WD Green Drives
had lots of them and I wonder if they were caused by trying those drives as
esata when it wasn't working correctly yet? I have exchanged those two
drive for Seatgate 2t green drives and they seem perfect. I am thinking
that my external enclosures may to more ompatible with certain drives but
who knows. I will keep tabs on the specs from now on. Do you think my boot
drive with just one error is something to worry about? I plan on upgrading
it to a faster drive someday anyway. Thank you for the very useful
information.

 

The 0xC7 statistic Ultra DMA CRC Errors , seems to be associated by other
people with a cabling problem.

Remember that, the reach budget (cable length) is a function of the
devices on either end of the cable. The chip on the motherboard has
to "launch" at ESATA amplitude. The device receiving the signal, has
to "sense" at ESATA thresholds. If both those requirements are met,
the cable has "perfect impedance" and isn't pinched or kinked, then
you could get up to two meters (six feet) of reach and see zero errors
accumulate.

Departures from that, eat into the budget. If either end is cheating
on the ESATA electrical spec, then you're probably allowed to have one
meter (three feet) of cable.

It could be as well, that if an extension cable is used (motherboard to faceplate
with one cable, external connector on faceplate to ESATA enclosure), that adds
an additional discontinuity to the electrical signals (one additional male/female
connector pair in path).

That's why, unless you have good specs for the devices in question,
to verify that stuff, you can assume you really have three feet of
cable to work with. And if an extension cable is used inside the
computer, then you might be left with 18" of budget for the part outside
the computer.

Now, if you use a PCI Express SIL3132 card with an ESATA connector on
the faceplate, then PCB material (FR4) carries the signals from the
chip on the card, to the faceplate. There is no pesky cable to eat budget.
The distance the copper tracks travel on the PCI Express card, is
extremely short. That leaves more of your budget for external cable.

The same could be said of a motherboard chip. The motherboard chip has
diff pair copper tracks running to the back panel ESATA connector stack.
That still uses perhaps one foot of reach, but in terms of impedance,
they can do a good job of getting the signals to the connector with
minimum disturbance.

*******

As near as I can tell, the 0xC7 statistic doesn't "reset" itself.
So if you counted 328 errors today, then fixed the cable, the drive
could well read 328 for the rest of its life. It is up to you to write
down when last the count incremented, for future reference when trying
to detect a problem.

I had no luck at all, from a standards point of view, in tracking that
thing down. I tried a search against t13.org, but still didn't manage
to find what I was looking for.

Paul

 

So Paul, they way I understand it is that if lets say I had an sata cable
connected to the boot drive that has that one error and the cable wasn't
completely on the connector then that could cause an error? I was using
the cables that came with the board but bought some that were just long
enough to reach from the mobo to the drive. If my cables are too short I
would have more errors than just one I would think. I am not going to worry
about that drive with one Ultra DMA CRC error. I will check it often and
if I see more then I will take action. I would say that seeing bad sectors
increasing woud be a big warning. Diskinfo and HD Tune both claim the disk
is good.

BTW, DiskInfo lists the one CRC error but doesn't attach a yellow warning
over it. DiskInfo shows no warnings. I have read that other users have
just one of those errors like mine and they can't figure out why either.

My old friend told me this and I stick to it today "Back it up or cry".
With the old tape drives of the 80's he also said "Back it up and cry",
because most of the tapes had major problems when trying to restore data.
Mine was the type that fit in a drive bay. Really popular then and tapes
were expensive.

Thanks for everything Paul. I enjoy chatting with you.

Robert

 

I'll say this about your example - the other examples I could find, always
had a larger count than you've got. I didn't see any examples with a low
quantity. Either zero or hundreds or low thousands.

Those numbers are just too low to be reasonable. If you've worked with
error counters in a lab before, if something runs amok, you get *much*
higher counts than that. I have a feeling that number is massaged somehow,
and isn't an actual error count.

I bet there is some significance to only seeing one error, but explaining
what is going on, is going to require finding whatever standard defines
how that count is collected. The ATA/ATAPI specs I could find so far,
do mention SMART, but only a specific section of SMART. Not the
statistics collection part.

In the Wikipedia article on S.M.A.R.T. they mention

"SMART Attributes were included in some drafts of the ATA standard,
but were removed before the standard became final."

which means the manufacturers wanted to wing it. Companies play games
in standards forums, and one of the games is introducing a feature
that causes your competitor to be "non-standard". Removing that
section of the spec is a polite way of saying, each company already
had a feature like that in place, and didn't want to change it.
And that means, we'll need to find a "guru sitting on a mountain"
somewhere, to discover the definition

Paul

 

HD Tune "C7" Ultra DMA Error Count displays under "Current= 200, under
"Worst" is 200, under "threshold" is 0, and "data" shows 1.

You lost me a bit but I am trying to understand. You think that "data"
showing 1 could be normal? I did read about many people having just 1 and
they didn't know why. I guess it's nothing to worry about because data
transfers are flawless and if I swapped cables the figure of 1 won't go
away.

BTW, so you know which 1.5tb drive is the fastest these days? The Seagate
7200 RPM like the one I have was the fastest supposedly but heard that the
WD caviar black is fastest now. I heard that Seagate has a new fast 2tb
drive, not the green model bit the other. Thanks.

Robert

 

Xbitlabs does some drive testing.

http://www.xbitlabs.com/articles/storage/display/hdd-2010_7.html#sect1

Check their "Storage" section, for more articles of that sort.

http://www.xbitlabs.com/articles/storage/

It looks like the Advanced Format drives, still have some
bad looking results in some of the graphs (4KB sectors).

There are some 1TB and 2TB drives in this review. I'm surprised
they're as close as they are. The big drives sometimes have
to compromise on performance, in order to work.

http://www.xbitlabs.com/articles/storage/display/1tb-2tb-roundup-2_8.html#sect0

I presume how a 1.5TB drive stands out, is the platter type it is
made from. Are those 750GB platters perhaps ? Maybe such a drive
represents a switch from three to two platters. In some of those
reviews, the extra platters help performance.

The Storagereview.com site used to keep a database of their test
results, but it looks like they're not updating it any more.

I'd say you probably wouldn't go badly wrong getting the WD one.
But check the Newegg reviews, to see how many of them are
failing early or arrive DOA. I'd sooner buy a drive with slightly
better quality numbers, than one with a better transfer rate.
If it's fast, but dead tomorrow, it isn't of much use to me. If
I have to RMA it, it's costing me money, and then it is no longer
a bargain. On the Newegg reviews, one person went through three
drives, before getting a good one. And that is pretty bad, in terms
of purchasing experience. That kind of thing takes all the fun out
of shopping for a drive.

Paul

 

Paul,

I think what I will do is email or call Seagate support and inquire about
it. I never take advantage of these companies free support but will at
least ask them about this. Will let you know. Thanks again for all the
info.

Robert

 

I appreciate the info and those links and will check them out. I also agree
with you regarding speed versus reliability. The Seagate 1.5t has been a
good drive for me and was rated excellent when the first appeared. And fast
too. When I was trying to get the esata going last week with the help of
the group I had two WD 1tb caviar green drives and somehow both got Ultra
DMA errors in the hundreds in just a few days. I used them via usb and
esata but the error figure was real high and that worried me. I contacted
WD via email and explained the situation and they actually took the time to
call me. I thought companies calling people by phone was unheard of these
days. They didn't help me much so I had the receipts and picked up two 2tb
Seagate barracuda green drives and they are working great. They were on
sale so I spent $16.00 and went from 1tb to 2tb! The first WD Green I
bought a while back acted weird and I checked warranty status online and
the system said it was out of warranty. What happened was a local store
sold me a refurb drive and somehow the drive sticker did not get marked as
such and I informed WD and when I showed their emails to the store they
replaced it. I went online and checked the serial and that drive was new.
You must be careful these days. I made a mistake of buying a couple refurb
drives from Unity Electronics and they crapped out just after their 30 day
return period. Steer clear of that place because all their drives are
refurbs but could just be used ones, who knows. Thanks again and take care.

Robert

 

Paul,

I have an idea why the 2tb drive wouldn't work esata. I read that those
drives can be used with 4k clusters or 512. I am thinking that the drives
may have worked if it was formatted with 4k instead of 512. It is possible
that the system will allow those drives to operate with 512 on a usb port
and not esata?

I formatted both of the two 2tb drives with one extended partition each and
used windows xp x64 default 512 cluster six.

Just wanted to run it by you, Thx.

Robert

 

If you read this article, I don't think it works that way. You can "align"
a partition to 4K boundaries - if the drive is actually 4K, then you should
notice an improvement in performance. I don't think the user actually
has the ability to "soft sector" the drive, and change it from using 4K sectors
to 512 byte sectors. All you can change, is what sector the partition
starts at, and consequently how often the drive must do read-modify-write
operations to handle the size mismatch on fractional chunks.

http://www.anandtech.com/show/2888/2

Paul