SIL 3112a Controller Issues....

  • Thread starter Thread starter Dave Onex
  • Start date Start date
D

Dave Onex

Hi Folks;

I have a SIL3112a RAID controller in my mainboard (I know, it's old) and I
have the latest SIL BIOS for it and the latest driver (matched pair).

I'm running MCE 2005 and I see a lot of disk error entries in the event
viewer and also controller entries. I know this controller was problematic -
does anyone recall if there was a solution of any sort? There's quite a few
posts about them on the 'net albeit mostly old ones.

Thanks!
Dave
 
Did you obtain driver from Sil, direct?
Are you using a mobo with 'active amour'?
What is the mobo?
Has your mobo bios been updated?
 
Dave Onex said:
Hi Folks;

I have a SIL3112a RAID controller in my mainboard (I know, it's old) and I
have the latest SIL BIOS for it and the latest driver (matched pair).

I'm running MCE 2005 and I see a lot of disk error entries in the event
viewer and also controller entries. I know this controller was
problematic - does anyone recall if there was a solution of any sort?
There's quite a few posts about them on the 'net albeit mostly old ones.

Thanks!
Dave

Have you got the appropriate, or latest, bios on it, and correct driver
installed ? ...which you can download here :-
http://www.siliconimage.com/PRODUCTS/product.aspx?pid=63

I've got a PCI-e Silicon Image SiI 3132 SATAlink Controller :-
http://www.google.co.uk/search?hl=en&q=3112a+silicon+image&meta=&aq=9&oq=3112a
....and the driver installation *.inf file was poorly crafted, and, (if I
remember correctly), ...it took two or three attempts to get the correct
driver, correctly installed/configured.
i.e. ....I think "it," and my Asus motherboard was looking for a RAID array
every boot, until I finally got the correct driver installed / configured.
I've found my cheapy SiI 3132 pci-e card to be quite reliable.

....oddly, perhaps of interest, last night, after waiting for Norton Ghost
14.0 to backup two SATA hd's C:\ (XP boot ) and D:\ to SiI 3132 E:\, (all
internal hd's),
....Ghost was also set to make an "off-site" copy to external hd USB drive
F:\ ...and after it had backed up C:\ and D:\ onto E:\ it copied 58 gb's
to USB F:\ and I thought it was never going to finish ! ...did a quick
1gb file copy with "DiskBench," out to ext. USB hd first, then internal to
internal :-

Copy File Bench started...

Copy E:\M3N78\X2-6000_C_Drive001_s01.v2i to
F:\temp\X2-6000_C_Drive001_s01.v2i
Size: 1073485444
Time: 50516 ms
Transfer Rate: 20.266 MB/s

Copy File Bench ended

Copy File Bench started...

Copy E:\M3N78\X2-6000_C_Drive001_s01.v2i to
D:\Temp\X2-6000_C_Drive001_s01.v2i
Size: 1073485444
Time: 18625 ms
Transfer Rate: 54.967 MB/s

Copy File Bench ended

....so I think I shall fork out for an eSATA external hard-disk !

regards, Richard
 
ooops !
Have you got the appropriate, or latest, bios on it, and correct driver
installed ? ...which you can download here :-
http://www.siliconimage.com/PRODUCTS/product.aspx?pid=63

I've got a PCI-e Silicon Image SiI 3132 SATAlink Controller :-
I meant:-
http://www.siliconimage.com/products/product.aspx?pid=32
...and the driver installation *.inf file was poorly crafted, and, (if I
remember correctly), ...it took two or three attempts to get the correct
driver, correctly installed/configured.
i.e. ....I think "it," and my Asus motherboard was looking for a RAID
array every boot, until I finally got the correct driver installed /
configured. I've found my cheapy SiI 3132 pci-e card to be quite reliable.
regards, Richard
 
Dave said:
Hi Folks;

I have a SIL3112a RAID controller in my mainboard (I know, it's old) and I
have the latest SIL BIOS for it and the latest driver (matched pair).

I'm running MCE 2005 and I see a lot of disk error entries in the event
viewer and also controller entries. I know this controller was problematic -
does anyone recall if there was a solution of any sort? There's quite a few
posts about them on the 'net albeit mostly old ones.

Thanks!
Dave

A SIL3112 on a motherboard, is handled a bit differently than a SIL3112
on a PCI card.

On the PCI card, there is a separate BIOS chip, which holds the
Silicon Image BIOS code. That BIOS has the extended INT 0x13
routine, for allowing the computer to boot from the PCI card.
The chip can be flashed, and by using the correct file from the
Silicon Image site, you can upgrade your PCI card. That makes it
relatively easy to update a card, as long as the flash chip
on the card, is one that the Silicon Image flasher program is
designed to support. (One brand is a PMC chip.)

On a motherboard, the necessary BIOS code module is stored in the main BIOS.
The motherboard manufacturer may choose to upgrade the module,
or not. (My Asus motherboard had about 10 different versions
of the BIOS, and I verified that the same, old, stale module
was present in all of them.) If you want to handle changing the
module yourself, it is not without risks. Back at a time
when this was popular, a person might own a BIOS Savior
duplicate BIOS chip for the motherboard, so that flashing
a bad homemade BIOS file, would not brick the motherboard.
If you don't have a BIOS Savior, and you make a mistake, you
may have to purchase a replacement chip from a site like
badflash.com .

http://www.ben.pope.name/a7n8x_faq.html#SATA_Update

Silicon Image denies there are any problems with their chip.
The chip is relatively simple in design, and most of the
intelligence comes from the firmware or system level software.
One of their firmware bugs, was the system BIOS freezing when
a 1TB drive was plugged into the SIL3112. Apparently that
can be fixed, but to fix it, you'd need to prepare a new
motherboard BIOS, as sketched out in the above FAQ.

You should examine the SATA cables, and make sure that they're
seated properly. First generation cables have virtually no
retention properties. If a cable is bent or pinched, that
can also cause problems. While it is unlike to be your
cables, you should still take a quick look at them.

There have been some other obscure bugs, just to annoy users.
One person, set up a RAID 1 mirror on a SIL3112. One of the
disks died. Upon examining the "mirror" copy on the other
disk, the data on the disk was three months stale. It appeared
that the array had silently stopped mirroring some time ago,
leaving the owner with old information on the degraded
array. Pretty hard to check for stuff like that, if
no error messages are showing at the time.

Paul
 
I experienced data files being 'out of sync' in a mirror array with this
chip, it was on an asus mobo, Tech support confirmed that there were know
issues, but though it had been cured with a mobo bios update
 
Hi guys - thanks for the replies!

The mainboard BIOS is the latest and the SIL BIOS (that's integrated into
the mainboard BIOS) is the latest release from SIL - so, both BIOS's are the
latest. I did have to use a special system BIOS that included the latest SIL
one 4.2.x I believe (it's the latest on their site).

I also installed the latest driver from SIL for that specific BIOS. The
mainboard is an Asus A7N8X-E and I'm running (2) 500 gig drives in a RAID 0
Array (for 1 TB total).

These are SATA II drives with the jumpers set to SATA I mode. The problem
I'm seeing is that under heavy disk I/O errors appear and, at times, the
controller does not respond within the timeout period. This only happens
under heavy disk I/O.

The cables never gave an issue under the old O/S and they haven't been
disturbed since. Again, it only happens under heavy disk I/O - I know others
posted about similar issues on the 'net but I was hoping someone would have
a magic bullet. I've even ensured that the onboard SIL controller is using
it's own dedicated IRQ (hoping that would fix it).

Any more ideas? They're all appreciated :-)

Best;
Dave
 
Dave said:
Hi guys - thanks for the replies!

The mainboard BIOS is the latest and the SIL BIOS (that's integrated into
the mainboard BIOS) is the latest release from SIL - so, both BIOS's are the
latest. I did have to use a special system BIOS that included the latest SIL
one 4.2.x I believe (it's the latest on their site).

I also installed the latest driver from SIL for that specific BIOS. The
mainboard is an Asus A7N8X-E and I'm running (2) 500 gig drives in a RAID 0
Array (for 1 TB total).

These are SATA II drives with the jumpers set to SATA I mode. The problem
I'm seeing is that under heavy disk I/O errors appear and, at times, the
controller does not respond within the timeout period. This only happens
under heavy disk I/O.

The cables never gave an issue under the old O/S and they haven't been
disturbed since. Again, it only happens under heavy disk I/O - I know others
posted about similar issues on the 'net but I was hoping someone would have
a magic bullet. I've even ensured that the onboard SIL controller is using
it's own dedicated IRQ (hoping that would fix it).

Any more ideas? They're all appreciated :-)

Best;
Dave

I tried a search of the nforcershq site, and don't see anything solid to go on.
Plenty of people substituting newer SIL BIOS modules in their BIOS, but
nobody able to say whether it is fixing anything or not.

I'm kinda curious as to whether you're using the Microsoft driver for your IDE
ports, or the Nvidia driver. "Chipset maker" drivers, tend to include their
idea of caching, which is good most of the time, but occasionally has compatibility
issues. As a consequence, I try to use the Microsoft Windows driver if I can
get it installed instead. Then, when using the Nvidia chipset package, I
untick the IDE driver so it won't be installed. (It's been quite a few years
since I've done that, so the details are foggy.) I still use stuff like the
AGP driver, because the rest of it is OK. But for Southbridge IDE, I prefer
the default Windows driver, even if the performance level isn't quite the
same.

On my A7N8X-E Deluxe, I never had any issues with disks, but I also wasn't using
the SIL3112. I got my first two SATA drives just this year, and never had any SATA
around while I was using the A7N8X-E Deluxe.

You can also take a look through the threads here.

http://vip.asus.com/forum/topic.aspx?board_id=1&model=A7N8X-E+Deluxe&SLanguage=en-us

This one isn't quite the same as yours.

"Drive timouts and errors with SATA RAID 0 and Seagate Drives"

http://vip.asus.com/forum/view.aspx...1&model=A7N8X-E+Deluxe&page=1&SLanguage=en-us

Paul
 
I had an A7N8(X?) delux, Asus stated there were known issues with the
onboard Sil sata controler, which they thought had been cured by a mobo bios
update - untill I spoke to them
I immediately changed the mobo and fitted a high end Adaptec sata pci card
controller

There are known issues with the earlier nvidea chip that uses Active Armour
(Not relevent in your case)
 
Great replies - thanks very much!

I'm not using the SIL IDE driver (I thought it might be an issue so I
removed it long ago).
I do have Seagate drives though.

I'm starting to think it's a problematic product (SIL) and I have thought
about moving over to a more proper RAID controller (such as Adaptec) but
when you get right into the nitty-gritty I actually might as well build a
new system.

Here's an interesting question - anyone here have experience with Intel's
ICH10R in a RAID configuration? I'm thinking that Intel programmers (both
BIOS and drivers) would be far superior to SIL

Best & Thanks!
Dave
 
Dave said:
Great replies - thanks very much!

I'm not using the SIL IDE driver (I thought it might be an issue so I
removed it long ago).
I do have Seagate drives though.

I'm starting to think it's a problematic product (SIL) and I have thought
about moving over to a more proper RAID controller (such as Adaptec) but
when you get right into the nitty-gritty I actually might as well build a
new system.

Here's an interesting question - anyone here have experience with Intel's
ICH10R in a RAID configuration? I'm thinking that Intel programmers (both
BIOS and drivers) would be far superior to SIL

Best & Thanks!
Dave

Proving a product works, seems to be a hard thing to do.

The advantage of a separate plugin card, is that is allows
removing the product, if it doesn't work. Having the SIL3112
soldered to the motherboard, means you cannot remove it.
(You can always disable and ignore it like I did.)

I don't recollect any complaints about Intel stuff, but
it is still another instance of a "softraid". It still
relies on a BIOS module at startup, and drivers while the
OS is running. So there is the same potential for
issues, and an issue could be introduced by an
innocent driver upgrade.

RAID controllers span quite a price range. Some of the
more expensive stuff, is tested and used by people on the
2cpu.com forums. So if you want comparisons of some of
the RAID controllers, that might be a place to look.

Part of the "RAID experience" doesn't show up, until there
is trouble. Some RAID products appear to be great the first
day you get them, but then, when a disk fails, and you
actually see the interface and try to swap in a spare
drive, your opinion may differ. This is why, a proper
review of a RAID product, cannot be a casual one. It takes
a few months to see what they're really made of. For example,
one user finds an array rebuild takes an entire week to
complete. Those are things you might not discover in
a single day evaluation.

As for what options you have, note that sometimes those
SIL3112 chips do get used in branded RAID cards. I think
there is at least one Adaptec card that uses a SIL3112.
They might have reworked the code though, so it might
bear little resemblance to the generic Taiwan SIL3112
$20 card.

Paul
 
A quality raid controller card is going to cost you approx 4 x price of an
half decent mobo, the Adaptec I used retailed at some $500, and that some 4
years ago.
And see my last response re your mobo
 
Thanks for the contiuned help!

I guess the conclusion is that there are known issues with the controller,
I'm running into some of them, and there's no work-around :-) That's pretty
much what I figured - I just wanted to make sure by contacting some
knowledgeable people before making my decision.

I have several servers here with 'real' SCSI RAID controllers so I have some
experience with them. I didn't really expect that a 'built-in' home user
type solution would have the same robustness to it but on the other hand I'm
only running 2 drives in a RAID 0 configuration so I figured it couldn't be
that hard :-)

I'm still kind of thinking that the Intel controller _should_ be OK given my
past experience with the company and the rapidity with which they update
their drivers & firmware. I'll take a look at the 2cpu site mentioned and
see what people who are actually using it think of it.

Thanks again!
Dave
 
Back
Top