killed a controller <rant>

[Flasherly]

Had an old SYBA PCI 2-port SATA controller.

Didn't cost much then. MBs didn't always have SATA ports.

So, now I'm running some massive storage HD arrays, along with a
couple interspersed OS partitions.

What's the point, though, of messing with OpSystems in such a
configuration. . .why shouldn't they be elsewhere instead of smack in
the middle, messing with a couple of drives' 3 or 4 terabytes. If I
want to mess with partitions, sizing, shoe-horning in OS's & boot
arbitrators, probably better doing it elsewhere.

So, I pull out this old board equipped with Silicon Image. No luck.
Figures. Those drives are too big, too new.

Why not try and flash the controller for updated compatibility? OK.
I can do that.

I go to Silicon Image with the board# and chip IDs.

There, they say: Silicon Image is a direct wholesale provider and not
a retailer with enduser support. If you have problems contact the
retailer where you bought your equipment. If you don't know what
you're doing you can destroy the board.

Fair enough. I download the BIOS flash utility and appropriate binary
files specified for the board. Boot to DOS and do it.

Dead. POSTs through the system, but won't take any no thing anyway
noway.

Now, overall, they want what a MB costs for these controllers. $10-15
out of a hundred or two, maybe less than a dozen models in discounting
further whatever's a crap review. Most of the better stuff likely
runs $20 and up.

Probably got mine, the old SI, for $7 or such. Back when bluejeans
cost 20cents and a pair of shoes a buck.

Well -- it working great, but I killed it thanks to the
manufacturer.

Bye bye Silicon Image. I'll buy something else, thank you very much
just the same.

Got one of the better equipped ones (some gear with it, cables and
whatnot, in a box, not just a bare board, provisions for external
ports, jumpers, oh - and a PATA port. 3 ports, 2 SATA).

VIA chipped controller in a Rosewill assembly package. $5 off for $15
total.

Totally unnecessary, still. I Never "kill" boards, and I didn't like
doing it one bit. Even if it wouldn't take optics or a HD over 500M
and sat unused in a box until the garbage can, I'd probably still have
bought another updated SI chipset -- just not after this experience
with their flash support.

In a way reminds me of leaving Promise and that nasty taste after
seeing them jack boards up out of normal purchasing prices.

Guess I just don't leaving product names once I get used to running
with something.

<end of rant>

 

[RayLopez99]

Fair enough. I download the BIOS flash utility and appropriate binary

files specified for the board. Boot to DOS and do it.

FLASHerly, you FLASHed the BIOS? How? Seriously I though you need a special ultraviolet EEPROM machine and you wear sunglasses to do that, or maybe I'm 25 years behind the times?

RL

 

[Charlie Hoffpauir]

FLASHerly, you FLASHed the BIOS? How? Seriously I though you need a special ultraviolet EEPROM machine and you wear sunglasses to do that, or maybe I'm 25 years behind the times?

RL

Ray, yes you ARE dating yourself.

 

[Paul]

Had an old SYBA PCI 2-port SATA controller.

Didn't cost much then. MBs didn't always have SATA ports.

So, now I'm running some massive storage HD arrays, along with a
couple interspersed OS partitions.

What's the point, though, of messing with OpSystems in such a
configuration. . .why shouldn't they be elsewhere instead of smack in
the middle, messing with a couple of drives' 3 or 4 terabytes. If I
want to mess with partitions, sizing, shoe-horning in OS's & boot
arbitrators, probably better doing it elsewhere.

So, I pull out this old board equipped with Silicon Image. No luck.
Figures. Those drives are too big, too new.

Why not try and flash the controller for updated compatibility? OK.
I can do that.

I go to Silicon Image with the board# and chip IDs.

There, they say: Silicon Image is a direct wholesale provider and not
a retailer with enduser support. If you have problems contact the
retailer where you bought your equipment. If you don't know what
you're doing you can destroy the board.

Fair enough. I download the BIOS flash utility and appropriate binary
files specified for the board. Boot to DOS and do it.

Dead. POSTs through the system, but won't take any no thing anyway
noway.

Now, overall, they want what a MB costs for these controllers. $10-15
out of a hundred or two, maybe less than a dozen models in discounting
further whatever's a crap review. Most of the better stuff likely
runs $20 and up.

Probably got mine, the old SI, for $7 or such. Back when bluejeans
cost 20cents and a pair of shoes a buck.

Well -- it working great, but I killed it thanks to the
manufacturer.

Bye bye Silicon Image. I'll buy something else, thank you very much
just the same.

Got one of the better equipped ones (some gear with it, cables and
whatnot, in a box, not just a bare board, provisions for external
ports, jumpers, oh - and a PATA port. 3 ports, 2 SATA).

VIA chipped controller in a Rosewill assembly package. $5 off for $15
total.

Totally unnecessary, still. I Never "kill" boards, and I didn't like
doing it one bit. Even if it wouldn't take optics or a HD over 500M
and sat unused in a box until the garbage can, I'd probably still have
bought another updated SI chipset -- just not after this experience
with their flash support.

In a way reminds me of leaving Promise and that nasty taste after
seeing them jack boards up out of normal purchasing prices.

Guess I just don't leaving product names once I get used to running
with something.

<end of rant>

Don't throw the card out. Plug it into another, working
computer, one where the disks still boot off the Southbridge.
Then, find another flasher and try again.

You can try searching the SiliconImage site for SIL3512 instead
of SIL3112. The chips are similar. Maybe there are tools there
you could use.

The flasher could only flash certain brands of EEPROM chip.
A typical brand on a SIL card, might be a "PMC" chip for example.
Only a few cards were made, where they used the wrong kind of
chip. Cards made later on, the manufacturers were pretty good
about putting the proper chip on the card so it could be updated.

One reason for flashing a SIL3112, is to avoid the >500GB bug.
The computer would freeze when the SIL3112 BIOS chip loaded,
and the disk drive was larger than 500GB. I don't recollect
the other bugs it had.

Paul

 

[Flasherly]

Don't throw the card out. Plug it into another, working
computer, one where the disks still boot off the Southbridge.
Then, find another flasher and try again.

You can try searching the SiliconImage site for SIL3512 instead
of SIL3112. The chips are similar. Maybe there are tools there
you could use.

The flasher could only flash certain brands of EEPROM chip.
A typical brand on a SIL card, might be a "PMC" chip for example.
Only a few cards were made, where they used the wrong kind of
chip. Cards made later on, the manufacturers were pretty good
about putting the proper chip on the card so it could be updated.

One reason for flashing a SIL3112, is to avoid the >500GB bug.
The computer would freeze when the SIL3112 BIOS chip loaded,
and the disk drive was larger than 500GB. I don't recollect
the other bugs it had.

Paul

Ah-hah, chip nomenclatures. Correct ones, too.

OK, it's a "global Silicon Image" - Company/Inc., &etc. -versus-
Silicon Image, proper, or headquarter in Silicon Valley, Ca.

Buslink is the actual card, and I did happen to find an actual picture
of it at Spin My Ol' Buddy's hit-&-miss website, located somewhere
alongside the Pacific Ocean:

http://www.geeks.com/details.asp?invtid=PCI-2S-NB

(There's a blow-up for the retail box, and that's how I can tell it's
the Exact Card).

Where I was, [sic] doing the flashing, however, is at Silicon Image's
website - so noted for the actual chip nomenclaure, i.e. 3112 - which
I can say almost certainly would have been a -R designated BIOS file
(for RAID, say as opposed to -B in someway perhaps designating a
"basic board," even though mine is sold for RAID functionality).

BusLogic, by the way, after just looking at, does provide a driver
package only as perfunctory to an operational system support.

In other words -- if one really likes to be where the stuff hits the
fan, goes to their Big Sister's Site - Silicon Image, that's
apparently the best place to be in order to get one's card killed.

Just for you, Paul. I pulled it out of the Garbage Can of Broken
Dreams, where'd it been sitting for the past day. (I've a few even
older PCI boards which do not fit present architecture bus compliance
specs. It can go alongside into one of those boxes). If you haven't
any more last requests, might I mention a seriousness involved with
being a known Pack Rat?

 

[Paul]

Don't throw the card out. Plug it into another, working
computer, one where the disks still boot off the Southbridge.
Then, find another flasher and try again.

You can try searching the SiliconImage site for SIL3512 instead
of SIL3112. The chips are similar. Maybe there are tools there
you could use.

The flasher could only flash certain brands of EEPROM chip.
A typical brand on a SIL card, might be a "PMC" chip for example.
Only a few cards were made, where they used the wrong kind of
chip. Cards made later on, the manufacturers were pretty good
about putting the proper chip on the card so it could be updated.

One reason for flashing a SIL3112, is to avoid the >500GB bug.
The computer would freeze when the SIL3112 BIOS chip loaded,
and the disk drive was larger than 500GB. I don't recollect
the other bugs it had.

Paul

Ah-hah, chip nomenclatures. Correct ones, too.

OK, it's a "global Silicon Image" - Company/Inc., &etc. -versus-
Silicon Image, proper, or headquarter in Silicon Valley, Ca.

Buslink is the actual card, and I did happen to find an actual picture
of it at Spin My Ol' Buddy's hit-&-miss website, located somewhere
alongside the Pacific Ocean:

http://www.geeks.com/details.asp?invtid=PCI-2S-NB

(There's a blow-up for the retail box, and that's how I can tell it's
the Exact Card).

Where I was, [sic] doing the flashing, however, is at Silicon Image's
website - so noted for the actual chip nomenclaure, i.e. 3112 - which
I can say almost certainly would have been a -R designated BIOS file
(for RAID, say as opposed to -B in someway perhaps designating a
"basic board," even though mine is sold for RAID functionality).

BusLogic, by the way, after just looking at, does provide a driver
package only as perfunctory to an operational system support.

In other words -- if one really likes to be where the stuff hits the
fan, goes to their Big Sister's Site - Silicon Image, that's
apparently the best place to be in order to get one's card killed.

Just for you, Paul. I pulled it out of the Garbage Can of Broken
Dreams, where'd it been sitting for the past day. (I've a few even
older PCI boards which do not fit present architecture bus compliance
specs. It can go alongside into one of those boxes). If you haven't
any more last requests, might I mention a seriousness involved with
being a known Pack Rat?

OK, try this page.

http://www.siliconimage.com/support/searchresults.aspx?pid=63&cat=15&ctid=2&os=0&

Paul

 

[Michael Black]

Ray, yes you ARE dating yourself.

No.

Eproms needed to be erased with ultraviolet light.

EEproms are "electronically erasable ROM".

They aren't a new thing, though they are more recent than mere eproms.

But they've been around in small sizes for a while, used to keep settings
in tv sets and the like, small devices with small amounts of space, the
early ones were often interfaced serially.

Having eeproms that are big enough for a bios and such is more recent, but
not "new".

Michael

 

[Flasherly]

OK, try this page.

http://www.siliconimage.com/support/searchresults.aspx?pid=63&cat=15&...

Paul

Yeppers, that's where I killed it. A DOS flash (appropriate to the
chipset) - includes three files, two included -R (raid) -B (non-raid/
basic), one file for onboard (mainboard/motherboard) chipsets. I
tried the raid. It'll boot -through-, I think, with nothing hooked
up. Hook one up an the same old HD, it did work with, and it now
won't. First flash looked to have taken without anything unusual.
Any further attempt, however, to reflash always results in a time-out
now, into what appears a couple of hexadecimal addresses before having
to hit the reset. Moral of the story - may as well have some fun with
the new VIA one coming in (it'll work, supposedly, with newer 2T
drives, and I won't I won't have to use an ink-eraser to clean edge
contacts before plugging it in).

 

[Loren Pechtel]

FLASHerly, you FLASHed the BIOS? How? Seriously I though you need a special ultraviolet EEPROM machine and you wear sunglasses to do that, or maybe I'm 25 years behind the times?

Maybe 15 years.

These days the BIOS is done with flash memory, the same as in a thumb
drive.

 

[RayLopez99]

On Monday, December 10, 2012 5:39:06 PM UTC+2, Michael Black wrote:

Thanks Charlie and Michael. So how do you 'flash' a BIOS that has NAND RAM? I guess it has a command of some sort like "look for BIOS update from any connected USB thumb drive" in the BIOS menu?

RL

 

[larrymoencurly]

Had an old SYBA PCI 2-port SATA controller.

So, I pull out this old board equipped with Silicon Image.
No luck. Figures. Those drives are too big, too new. Why
not try and flash the controller for updated compatibility?
OK. I can do that. I go to Silicon Image with the board#
and chip IDs.

I download the BIOS flash utility and appropriate binary
files specified for the board. Boot to DOS and do it.
Dead. POSTs through the system, but won't take any no
thing anyway noway.

Bye bye Silicon Image. I'll buy something else, thank you
very much just the same. Got one of the better equipped
ones (some gear with it, cables and whatnot, in a box, not
just a bare board, provisions for external ports, jumpers,
oh - and a PATA port. 3 ports, 2 SATA). VIA chipped
controller in a Rosewill assembly package. $5 off for $15
total. Totally unnecessary, still. I Never "kill" boards,
and I didn't like doing it one bit.

Are you saying the Silicon Image card doesn't even recognized
the disk drives? That would indicate a bad flash. OTOH if
the disks are recognized but the controller can't read them
and maybe causes a NTLDR message, then you might have to
switch between a RAID and non-RAID version of the BIOS.

I've flashed at least a dozen Silicon Image controllers but
never killed one by doing that, even when I used EEPROM chips
taken from video cards and motherboards. SI's Windows flasher
never worked, except for backing up BIOSes to disk, but their
DOS flasher always did. I also flashed one of those SI cards
with either Uniflash or FlashROM. These SI cards were all
no-name from Ebay, $2.50 - $5, delivered.

I can't say that the SI cards have run flawlessly, and I've had
problems using them and the motherboard's controller together
(Windows XP, 98SE). But the same has been true with VIA VT6421A
cards and even the original Promise TX150 SATA/PATA cards.

 

[larrymoencurly]

FLASHerly, you FLASHed the BIOS? How? Seriously I though you need
a special ultraviolet EEPROM machine and you wear sunglasses to do
that, or maybe I'm 25 years behind the times? RL

Some really cheap controller cards were delivered with UV EPROMs, as
recently as 3 years ago, and they can't be erased because they have
no clear window to let in the sunshine. For chips with a window, at
least 90% of the bits would be erased in a day in the Arizona summer
sun, 100% in 2 days, so I'd leave them out for 3-5 days.

 

[DevilsPGD]

In the last episode of

Thanks Charlie and Michael. So how do you 'flash' a BIOS that has
NAND RAM? I guess it has a command of some sort like "look for
BIOS update from any connected USB thumb drive" in the BIOS menu?

Or it can even be done from within the OS in many cases. While this can
be somewhat less reliable, if your hardware has a secondary BIOS or
other fail-safe mechanism, this can work just fine too (although I still
wouldn't recommend it as a primary choice)

 

[Charlie Hoffpauir]

On Monday, December 10, 2012 5:39:06 PM UTC+2, Michael Black wrote:

Thanks Charlie and Michael. So how do you 'flash' a BIOS that has NAND RAM? I guess it has a command of some sort like "look for BIOS update from any connected USB thumb drive" in the BIOS menu?

RL

Remember Ray, although I've been fooling with computers since the
early 70's when I put together a 6800 kit, I don't really know
electronics. The BIOS I've flashed that I can remember were for my
Motherboard, and the best I can recall I had to download software to
do it from the MB mfg's site to a floppy, then boot from the floppy.
From there it was simply follow the directions. How it was done I have
no idea.

Now back in the days whan we used EPROMS I actually had an EPROM
burner card that plugged into my computer, and software to burn
instructions to the EEPROM. Erasing was as you remembered by using a
UV light. The best I can recall it took several minutes, so you put
the chip and light in a room, turned it on, and left the room closed
off.

I remember on my old 6800 computer there was an EPROM with the "boot
loader" burned into it. I don't remember how much memory the chip
held, but some of us actually wrote machine language instructions to
add to the boot loader, and burned them into a chip with twice the
memory, thus vastly increasing the usability of the computer. At that
time, 8K memory boards were a luxury, selling for about $200. I can
only imagine the cost that would be in today's dollars.

I also seem to remember that there were "write once" or PROMs
available, but I'm not sure about that. To change the programming,
you'd need to get a new chip.

 

[Flasherly]

Are you saying the Silicon Image card doesn't even recognized
the disk drives? That would indicate a bad flash. OTOH if
the disks are recognized but the controller can't read them
and maybe causes a NTLDR message, then you might have to
switch between a RAID and non-RAID version of the BIOS.

I've flashed at least a dozen Silicon Image controllers but
never killed one by doing that, even when I used EEPROM chips
taken from video cards and motherboards. SI's Windows flasher
never worked, except for backing up BIOSes to disk, but their
DOS flasher always did. I also flashed one of those SI cards
with either Uniflash or FlashROM. These SI cards were all
no-name from Ebay, $2.50 - $5, delivered.

I can't say that the SI cards have run flawlessly, and I've had
problems using them and the motherboard's controller together
(Windows XP, 98SE). But the same has been true with VIA VT6421A
cards and even the original Promise TX150 SATA/PATA cards.

Thought you'd show up, or someone like you, who's run with them
without a hitch.

Wish I could say the same, as I'm in the same camp - always flash with
DOS and know about driver conflicts and resolving them. Disks aren't
being recognized after flashing the raid version binary with FlashROM.
Any attempts to re-flash after result in a system hang.

No complaints, either. Up until now, it's been a good little backup
board. Everybody gets a fluke now and then, pulls the joker. Guess
it was my turn.

 

[Franc Zabkar]

I can't say that the SI cards have run flawlessly, and I've had
problems using them and the motherboard's controller together
(Windows XP, 98SE). But the same has been true with VIA VT6421A
cards ...

https://ata.wiki.kernel.org/index.php/Sata_via

"Certain WD SATA II 3 Gbps disks on VT6420 and VT6421 will trigger
error conditions under which the SATA controller FIFO overflows during
controller-disk communication because these disks do not fully conform
to the SATA specification."

I know that Rosewill uses this chip on at least some of their cards. I
recall one thread where a WD user was experiencing thousands of UDMA
CRC Errors.

- Franc Zabkar

 

[Flasherly]

https://ata.wiki.kernel.org/index.php/Sata_via

"Certain WD SATA II 3 Gbps disks on VT6420 and VT6421 will trigger
error conditions under which the SATA controller FIFO overflows during
controller-disk communication because these disks do not fully conform
to the SATA specification."

I know that Rosewill uses this chip on at least some of their cards. I
recall one thread where a WD user was experiencing thousands of UDMA
CRC Errors.

Sounds like déjà vu and early PCI from ISA compliance issues on W95.
That OS was a witch with hardware.

Got mine the other day --

Front and back PCB circuit board traces are first painted with some
kind of plastic paint before silk-screening in the circuit components.

VT6421A VIA and Winbond chips do all the wizardry for 3 functional
ports, 5 total assignable (gold ol' skewl pin jumpers on the gnarly
block), over 2 SATA and 1 PATA.

After a week of screwing with and diagnosing the best config for dual
T-byte storage drives and a small beater OS drive for bring them up --
I'm pooped.

Send me a monkey to finish it off, please.