Can't enable SMART on RAID0 Raptors on Asus P5W DH Deluxe

[Synapse Syndrome]

I can't seem to get SMART monitoring working on these Raptors.

They are connected to the orange SATA connectros for the Silicon Image
controller for EX-Bazkup, using RAID 0. I have changed the SMART settings
for the single striped HDD entry in teh BIOS from Auto to Enable, but SMART
is still not working.

Does this controller not supposrt SMART?

Cheers

ss.

 

[Paul]

I can't seem to get SMART monitoring working on these Raptors.

They are connected to the orange SATA connectros for the Silicon Image
controller for EX-Bazkup, using RAID 0. I have changed the SMART settings
for the single striped HDD entry in teh BIOS from Auto to Enable, but SMART
is still not working.

Does this controller not supposrt SMART?

Cheers

ss.

In a Port Multiplier context, I can see a SMART command
sent from an originating port, being encapsulated in
a FIS and sent to the correct disk. So as long as the
command can be passed through the hardware, it would
work. Each disk is addressable on a Port Multiplier, so the
hardware is pretty transparent.

In the case of the 4723 operating in hardware RAID mode,
one of the things the array should do, is "information hiding".
That is why the Asus manual tells you that the identity of
the disks in the array will not be visible in the BIOS. The
array of two disks is just "array" as far as the BIOS knows.
The 4723 has no bus interface, and there are no registers
that the BIOS can query.

Now, the 4723 Product Brief here, claims SMART is supported.

http://www.siliconimage.com/docs/SiI 4723 Prod Brief_FINAL.pdf

At least "Mandatory Commands" are supported. But, if two
disks are in an array, how can this work ? To the Southbridge,
it appears as a single volume, and a SMART command sent to
the array, will be handled how ? The SMART command would have
to be handled on a "per disk" basis, and the only agent that
can possibly know that the thing is a RAID, is the Silicon Image
management software for Windows. Maybe it has some option to
tunnel through the RAID, and address the individual disks. If
you were using some third party SMART software, I don't know how
that software would figure out that a tunneling scheme was
required.

Maybe you could connect one disk to the 4723, effectively
nullifying it. Then set up two arrays on the Southbridge,
and stand a better chance of getting support on what
the Southbridge thinks are its four ports.

Hub_bus
|
|
Southbridge ------ PRI_IDE
| | | \
Sata Sata Sata \ Using one disk on the 4723
1 3 4 \ should make it transparent,
Silicon so that any software that works
Image with the Southbridge, will be
4723 unaware the 4723 is in the path.
| |
Sata (No drive)
2

BTW - I noticed in the Product Brief that performance is estimated:

"Max. performance greater than 110MBps (RAID 0)"

So there is really no reason for the 4723 to be on this motherboard.
The Southbridge can do better than that, using software RAID.

HTH,
Paul

 

[old man]

SMART never worked on my Sil controler, nor does it work on my Adaptec
controler card, though the Adaptec utility seems to show Smart enabled

 

[Synapse Syndrome]

SMART never worked on my Sil controler, nor does it work on my Adaptec
controler card, though the Adaptec utility seems to show Smart enabled

As far as I know, SMART data cannot go through the pseudo-SCSI interface of
PCI disk controller cards. I'm sure Paul could clarify this.

ss.

 

[Synapse Syndrome]

In a Port Multiplier context, I can see a SMART command
sent from an originating port, being encapsulated in
a FIS and sent to the correct disk. So as long as the
command can be passed through the hardware, it would
work. Each disk is addressable on a Port Multiplier, so the
hardware is pretty transparent.

In the case of the 4723 operating in hardware RAID mode,
one of the things the array should do, is "information hiding".
That is why the Asus manual tells you that the identity of
the disks in the array will not be visible in the BIOS. The
array of two disks is just "array" as far as the BIOS knows.
The 4723 has no bus interface, and there are no registers
that the BIOS can query.

Now, the 4723 Product Brief here, claims SMART is supported.

http://www.siliconimage.com/docs/SiI 4723 Prod Brief_FINAL.pdf

At least "Mandatory Commands" are supported. But, if two
disks are in an array, how can this work ? To the Southbridge,
it appears as a single volume, and a SMART command sent to
the array, will be handled how ? The SMART command would have
to be handled on a "per disk" basis, and the only agent that
can possibly know that the thing is a RAID, is the Silicon Image
management software for Windows. Maybe it has some option to
tunnel through the RAID, and address the individual disks. If
you were using some third party SMART software, I don't know how
that software would figure out that a tunneling scheme was
required.

Maybe you could connect one disk to the 4723, effectively
nullifying it. Then set up two arrays on the Southbridge,
and stand a better chance of getting support on what
the Southbridge thinks are its four ports.

Hub_bus
|
|
Southbridge ------ PRI_IDE
| | | \
Sata Sata Sata \ Using one disk on the 4723
1 3 4 \ should make it transparent,
Silicon so that any software that works
Image with the Southbridge, will be
4723 unaware the 4723 is in the path.
| |
Sata (No drive)
2

BTW - I noticed in the Product Brief that performance is estimated:

"Max. performance greater than 110MBps (RAID 0)"

So there is really no reason for the 4723 to be on this motherboard.
The Southbridge can do better than that, using software RAID.

Thanks Paul

The Silicon Image controller must have been added for this easy driverless
RAID, and for the extra port. So you are saying that there is no possible
speed advantage of using it? That would confirm what another person said
about this controller, and I'll use the southbridge controller when I
re-install, if you are right about being able to use one of the orange ports
so that I still have four disks running, two RAID 0 and two RAID 1.
Hopefully SMART will work as well.

Cheers

ss.

 

[Paul]

Thanks Paul

The Silicon Image controller must have been added for this easy driverless
RAID, and for the extra port. So you are saying that there is no possible
speed advantage of using it? That would confirm what another person said
about this controller, and I'll use the southbridge controller when I
re-install, if you are right about being able to use one of the orange ports
so that I still have four disks running, two RAID 0 and two RAID 1.
Hopefully SMART will work as well.

Cheers

ss.

I suppose I should clarify that a bit. The chip would be
transparent if in Port Multiplier mode. In RAID mode, even
with one disk, it will still show in the BIOS as "array",
and will be doing its "information hiding" thing. So the
RAID mode adds a layer, that I don't see an easy way of
penetrating. Maybe it is wishful thinking on my part,
that the Silicon Image guys would recognize a one drive
RAID array as simply a single drive for the purposes of
the SMART comnands.

With the above diagram, what you should get, is two (two drive)
arrays with better performance. I cannot guarantee that SMART
data will be accessable for "Sata 2", since I think all the
Asus jumper options are considered to be RAID configurations.
_Probably_ the chip would need to be in Port Multiplier mode,
in which case the Intel Southbridge might be the issue. I
don't see a way for me to guess at the functionality from
looking at specs, since at least some of it relies on
drivers/software to work.

This is the SATA I spec. The old link I had was no longer
valid, so this archived copy is all I could get. This failed
to unzip for me, until I opened the file in a hex editor,
and added "00" hex (one extra byte) to the very end of the file.

http://web.archive.org/web/20030213....org/collateral/zipdownloads/serialata10a.ZIP

"FIS = Frame Information Structure
The user payload of a frame, does not include
the SOF, CRC, and EOF delimiters."

So nothing unique there. I guess every SATA packet has an FIS.
With Port Multipliers, they refer to "FIS based switching",
which I guess the FIS is addressed to a particular drive or
something.

"9.6 Non-data command protocol
...
- SMART DISABLE OPERATION
- SMART ENABLE/DISABLE AUTOSAVE
- SMART ENABLE OPERATION
- SMART EXECUTE OFFLINE IMMEDIATE
- SMART RETURN STATUS

9.7 PIO data-in command protocol
...
- SMART READ DATA
- SMART READ LOG SECTOR"

I take it those are the manditory commands. But for the
4723, most likely to work in Port Multiplier mode.

I wish a datasheet for the 4723 was available for download.
I'm real curious as to whether there is a fourth jumper
for Port Multiplier mode, or it is a matter of leaving
the other three jumpers open, to get to Port Multiplier
mode. But without a datasheet, there is no way to get a
pinout for the chip.

Paul

 

[kony]

As far as I know, SMART data cannot go through the pseudo-SCSI interface of
PCI disk controller cards. I'm sure Paul could clarify this.

ss.

If the feature were there, it could, but it isn't so if you
want Smart status you load the SiI Manager software to see
it, it is not handled through motherboard bios.

It is indeed slower, any controller integrated into the
chipset will tend to be faster. Will it matter? For
storage probably not but for main OS drive(s) I'd avoid it.

 

[Synapse Syndrome]

I suppose I should clarify that a bit. The chip would be
transparent if in Port Multiplier mode. In RAID mode, even
with one disk, it will still show in the BIOS as "array",
and will be doing its "information hiding" thing. So the
RAID mode adds a layer, that I don't see an easy way of
penetrating. Maybe it is wishful thinking on my part,
that the Silicon Image guys would recognize a one drive
RAID array as simply a single drive for the purposes of
the SMART comnands.

With the above diagram, what you should get, is two (two drive)
arrays with better performance. I cannot guarantee that SMART
data will be accessable for "Sata 2", since I think all the
Asus jumper options are considered to be RAID configurations.
_Probably_ the chip would need to be in Port Multiplier mode,
in which case the Intel Southbridge might be the issue. I
don't see a way for me to guess at the functionality from
looking at specs, since at least some of it relies on
drivers/software to work.

This is the SATA I spec. The old link I had was no longer
valid, so this archived copy is all I could get. This failed
to unzip for me, until I opened the file in a hex editor,
and added "00" hex (one extra byte) to the very end of the file.

http://web.archive.org/web/20030213....org/collateral/zipdownloads/serialata10a.ZIP

"FIS = Frame Information Structure
The user payload of a frame, does not include
the SOF, CRC, and EOF delimiters."

So nothing unique there. I guess every SATA packet has an FIS.
With Port Multipliers, they refer to "FIS based switching",
which I guess the FIS is addressed to a particular drive or
something.

"9.6 Non-data command protocol
...
- SMART DISABLE OPERATION
- SMART ENABLE/DISABLE AUTOSAVE
- SMART ENABLE OPERATION
- SMART EXECUTE OFFLINE IMMEDIATE
- SMART RETURN STATUS

9.7 PIO data-in command protocol
...
- SMART READ DATA
- SMART READ LOG SECTOR"

I take it those are the manditory commands. But for the
4723, most likely to work in Port Multiplier mode.

I wish a datasheet for the 4723 was available for download.
I'm real curious as to whether there is a fourth jumper
for Port Multiplier mode, or it is a matter of leaving
the other three jumpers open, to get to Port Multiplier
mode. But without a datasheet, there is no way to get a
pinout for the chip.

In much simpler terms:

Shall I just install the Intel RAID drivers onto the OS RAID 0 disks, and
unplug plug them both from the the Sil controller and then plug them into
the Intel southbridge controller for better speed and SMART support?

I can then put one of the RAID 1 HDDs on the remaining southbridge port and
the other in one of the orange Sil ports. This orange port will then
effectively be the fourth southbridge port.

So I will have RAID 0 and RAID 1 working from the Intel Southbridge, and I
can forget about the Sil controller completely? And SMART will probably
work on all four disks?

And we aren't sure what the motherboard jumpers for the Sil controllers
should be set at?

Is that the summary of the situation here?

Thanks again.

ss.

 

[Paul]

In much simpler terms:

Shall I just install the Intel RAID drivers onto the OS RAID 0 disks, and
unplug plug them both from the the Sil controller and then plug them into
the Intel southbridge controller for better speed and SMART support?

I can then put one of the RAID 1 HDDs on the remaining southbridge port and
the other in one of the orange Sil ports. This orange port will then
effectively be the fourth southbridge port.

So I will have RAID 0 and RAID 1 working from the Intel Southbridge, and I
can forget about the Sil controller completely? And SMART will probably
work on all four disks?

And we aren't sure what the motherboard jumpers for the Sil controllers
should be set at?

Is that the summary of the situation here?

Thanks again.

ss.

Hub_bus
|
|
Southbridge ------ PRI_IDE
| | | \
Sata Sata Sata \
1 3 4 \
Silicon
Image
4723
| |
Sata (No drive)
2

SATA 3 and SATA 4 would be your RAID 0 OS array.
SATA 1 and SATA 2 would be your RAID 1 data mirror (with SMART
support to be proven).

1) Enable Intel RAID in BIOS.
2) Configure array(s) in RAID BIOS. You could do the boot
array first, finish the Windows install, then come back
and configure the second array. This avoids any problems
with Windows install doing something to the wrong volume.
3) Install Intel RAID driver. If either array holds the boot volume
then the RAID driver has to be installed via F6 at the beginning
of Windows install (or repair install if applicable)

There are three jumpers shown in the Asus manual for the 4723.
Asus labels them as:

RAID_SEL 2x3 header
Jumper Position

1 2 3
X X RAID 1 (mirror)
X Spanning
X RAID 0 (stripe)

The manual says to use RAID 1 jumper setting for the single drive
on the SIL4723, in the above four drive Intel configuration.
I don't see how SMART is going to work on the single drive
on the 4723, as long as it is using one of its non
port multiplier modes.

The jumper block has five other jumper combinations whose
function is unknown. One of those combinations could enable
Port Multiplier mode, and since Asus has been careful not to
document that mode, my presumption is it doesn't work with the
Intel RAID software. If it did, I am pretty sure that the
manditory SMART features would work. When the 4723 is in one
of the three RAID modes shown in the above table, there is not
a transparent way to get SMART commands into the "array"
on the 4723.

If you are willing to experiment, I have a feeling that
jumper position 2 doesn't do anything, and only position
1 and 3 are real. The fourth combination would be no
jumpers present, and should be "JBOD" as mentioned here:

http://www.prnewswire.com/cgi-bin/stories.pl?ACCT=104&STORY=/www/story/01-04-2006/0004242496&EDATE=

You could try it with no jumpers, a drive on SATA 1 and
a single drive on the 4723, and see if the Intel RAID
driver freaks when it sees the 4723 in JBOD/Port_Multiplier
mode. If you can see the brand name and model number
of the drive on the 4723 SATA port while in the BIOS, when
no jumpers are present on the RAID_SEL jumper block, then
the drive on the 4723 is then the equal of the three other
drives. If the drive doesn't appear at all in the BIOS, when
there are no jumpers on the RAID_SEL header, then you have to go
back to the RAID 1 jumper setting (makes a single drive mirror
using the one drive on the 4723).

Paul

 

[Synapse Syndrome]

You could try it with no jumpers, a drive on SATA 1 and
a single drive on the 4723, and see if the Intel RAID
driver freaks when it sees the 4723 in JBOD/Port_Multiplier
mode. If you can see the brand name and model number
of the drive on the 4723 SATA port while in the BIOS, when
no jumpers are present on the RAID_SEL jumper block, then
the drive on the 4723 is then the equal of the three other
drives. If the drive doesn't appear at all in the BIOS, when
there are no jumpers on the RAID_SEL header, then you have to go
back to the RAID 1 jumper setting (makes a single drive mirror
using the one drive on the 4723).

Cheers Paul. I still haven't been able to try this out as I can;t seem to
install the Intel RAID software. It won't install until I have the
southbridge controller in RAID mode. When it is in RAID mode the drives
don;t boot in the Sil controller anymore. Is the only way to try this out
going to be a fresh OS installation, or repair installation?

ss.

 

[Mitch Crane]

(e-mail address removed) (Paul) wrote in 192.168.1.178:

There are three jumpers shown in the Asus manual for the 4723.
Asus labels them as:

RAID_SEL 2x3 header
Jumper Position

1 2 3
X X RAID 1 (mirror)
X Spanning
X RAID 0 (stripe)

The manual says to use RAID 1 jumper setting for the single drive
on the SIL4723, in the above four drive Intel configuration.
I don't see how SMART is going to work on the single drive
on the 4723, as long as it is using one of its non
port multiplier modes.

The jumper block has five other jumper combinations whose
function is unknown. One of those combinations could enable
Port Multiplier mode, and since Asus has been careful not to
document that mode, my presumption is it doesn't work with the
Intel RAID software. If it did, I am pretty sure that the
manditory SMART features would work. When the 4723 is in one
of the three RAID modes shown in the above table, there is not
a transparent way to get SMART commands into the "array"
on the 4723.

If you are willing to experiment, I have a feeling that
jumper position 2 doesn't do anything, and only position
1 and 3 are real. The fourth combination would be no
jumpers present, and should be "JBOD" as mentioned here:

I know I'm replying to a relatively old post, but this info may still be
relavent.

It isn't officially documented that I know of, but users on vip.asus.com
and www.xtremesystems.org have pointed out that you can remove all
jumpers and use the second EZ-Backup port as the missing SATA 2 port.
I've verified this, myself. You get SMART and NCQ on drives connected
this way.