LSI Logic 21040 card won't POST

[Lady Margaret Thatcher]

I have an LSI (Symbios) 21040-33 card. Dual channel SCSI. 64-bit
wide, 133 Mhz. This card won't finish POST, although the its "welcome
banner" does appear. However, I can't get control-C to work to get
the card to start its BIOS config menus.

I was hoping that a BIOS flash might "unstick" the card, but since the
card doesn't POST, then my system can't POST, so I can't do a BIOS
flash.

Two years ago, I bought two of these cards used, along with two Asus
dual AMD CPU motherboards. At that time, I built up one system,
including this model card, and it all runs quite nicely. I don't
recall having any problems with initial configuration of this card.

I've just started to build up the second system. I've R'ed TFM, but
there is no information about trouble-shooting this problem.

I always build up a new system one step at a time. Without the LSI
card, the skeletal system consists of the motherboard, both AMD
processors installed and running, and a Matrox VGA card. That boots
up fine.

When I installed the LSI card and encountered this POST problem, there
were no drives connected except a floppy drive for BIOS flashing
purposes only. As noted, without the LSI card in the system, it posts
perfectly

Both the LSI card in my first system and this one are at SMDS 4.16,
and both cards have identical V4.16.00 stickers on chip U17, near the
lower right corner of the card.

What to do?

Thanks.

 

[wolfgang schneider]

What to do?

Thanks.

if you waited ~5 minutes and the ha doesn't react , you can flash its
bios . if it doesn't allow this it is dead . nothing to help .

 

[Arno Wagner]

I have an LSI (Symbios) 21040-33 card. Dual channel SCSI. 64-bit
wide, 133 Mhz. This card won't finish POST, although the its "welcome
banner" does appear. However, I can't get control-C to work to get
the card to start its BIOS config menus.

I was hoping that a BIOS flash might "unstick" the card, but since the
card doesn't POST, then my system can't POST, so I can't do a BIOS
flash.

Two years ago, I bought two of these cards used, along with two Asus
dual AMD CPU motherboards. At that time, I built up one system,
including this model card, and it all runs quite nicely. I don't
recall having any problems with initial configuration of this card.

I've just started to build up the second system. I've R'ed TFM, but
there is no information about trouble-shooting this problem.

I always build up a new system one step at a time. Without the LSI
card, the skeletal system consists of the motherboard, both AMD
processors installed and running, and a Matrox VGA card. That boots
up fine.

When I installed the LSI card and encountered this POST problem, there
were no drives connected except a floppy drive for BIOS flashing
purposes only. As noted, without the LSI card in the system, it posts
perfectly

Both the LSI card in my first system and this one are at SMDS 4.16,
and both cards have identical V4.16.00 stickers on chip U17, near the
lower right corner of the card.

What to do?

Might indeed be a broken flash memory.

If the BIOS chip is removable, have it re-flashed or have
a new one flashed at one of the outfits offering this service.
Not expensive usually so it is worth a try.

Arno

 

[Trevor Hemsley]

I have an LSI (Symbios) 21040-33 card. Dual channel SCSI. 64-bit
wide, 133 Mhz. This card won't finish POST, although the its "welcome
banner" does appear. However, I can't get control-C to work to get
the card to start its BIOS config menus.

Have you tried it in a different PCI slot? Or a different motherboard? With only
one CPU installed?

 

[Folkert Rienstra]

I have an LSI (Symbios) 21040-33 card. Dual channel SCSI. 64-bit
wide, 133 Mhz. This card won't finish POST, although the its "welcome
banner" does appear. However, I can't get control-C to work to get
the card to start its BIOS config menus.

I was hoping that a BIOS flash might "unstick" the card, but since the
card doesn't POST, then my system can't POST, so I can't do a BIOS
flash.

Two years ago, I bought two of these cards used, along with two Asus
dual AMD CPU motherboards. At that time, I built up one system,
including this model card, and it all runs quite nicely. I don't
recall having any problems with initial configuration of this card.

I've just started to build up the second system. I've R'ed TFM, but
there is no information about trouble-shooting this problem.

I always build up a new system one step at a time. Without the LSI
card, the skeletal system consists of the motherboard, both AMD
processors installed and running, and a Matrox VGA card. That boots
up fine.

When I installed the LSI card and encountered this POST problem, there
were no drives connected except a floppy drive for BIOS flashing
purposes only. As noted, without the LSI card in the system, it posts perfectly.

Both the LSI card in my first system and this one are at SMDS 4.16,
and both cards have identical V4.16.00 stickers on chip U17, near the
lower right corner of the card.

What to do?

So you have 2 identical cards and 2 identical systems and you don't know
what to do.

Gee.

What a dilemma.

 

[Folkert Rienstra]

Gee, what a yob you are. Out of four responses, yours was the only
one that wasn't constructive.

Yes, the bad LSI card doesn't POST in either system. See my other
responses, and then try to think of something else I could do.

You're obviously very smart in this area,

Right, which obviously suggests that you already know what to do but
want it personally rehashed for you.

but what is your problem?

I have no problem. You are having one.

Are you incapable of actually helping out here?

Nope, not really.

Prove me wrong. I challenge you.

Too late. I smelled a troll which you now just confirmed.
I had a response ready to Rob's answer but I have now deleted it.

 

[Lady Margaret Thatcher]

Right, which obviously suggests that you already know what to do but
want it personally rehashed for you.

No, I'm not that knowledgeable. I'm just good at recognizing when
other people have significant knowledge.

I have no problem. You are having one.

See below.

Nope, not really.


Too late. I smelled a troll which you now just confirmed.
I had a response ready to Rob's answer but I have now deleted it.

No troll, Heer R (or Herr R). If "I" have a problem, it is with your
supercilious, arrogant attitude. Honestly, I think you were bs'ing
here. You don't have any more suggestions.

Go ahead. Prove me wrong. If you don't have any new suggestions,
that is prima facie evidence that it is YOU who are the Troller of all
Trolls. The überTroller.

-thatcher--

 

[The Moojit]

If the motherboard manufacturer has a later BIOS. I would remove the card
and upgrade the system BIOS to see if you get lucky. If you still can't
POST, try breaking into the system BIOS menu and assigning a different IRQ
to this card. It's a hack but worth a try.

The Moojit

 

[Lady Margaret Thatcher]

On Mon, 22 Aug 2005 22:46:06 -0700, Lady Margaret Thatcher


OK, so shoot me for top-posting. Here is a status report.

The only way I could get the "bad" board to get through a post was to
put it into the same system as the "good" board. When I first plugged
the good board into my system--in-progress, it too reported invalid
configuration. But it was no problem to enter the setup menu and
reset parameters.

With both boards in the same system, I could start up the
configuration menu from the good board, and that menu would also try
to configure the bad board. When I selected one of the two channels
(these are dual-channel boards) in the bad board, the system hung,
just as it hung because the bad board wouldn't POST.

When I restarted my system, both boards went through POST and board
boards searched for SCSI devices. Since the new system is still being
built up, there were no SCSI devices present. What is interesting,
though, is that the good board would scan SCSI IDs at about 1 per
second, but the bad board took about 30 seconds per ID. And at 15 x 2
IDs (dual channel boards, remember), that's a long time.

When I pulled the good board out of the system, the bad board refused
to POST, just as before.

I should also mention that once I got the system to post with both
boards, the first thing I did was back up the BIOS from the good
board, at rev 4.16 and use that BIOS to flash the bad board, which was
at 4.19. All the above behavior about configuration and SCSI ID
scanning, was with the bad board already at 4.16.

So in the end, I have a board that may be useful as a host adapter,
but can't be configured. So, it goes into my "dud parts" pile.

In the meantime, I just checked out eBay, and for about the same
$80-$100 I paid for this board 2 or 2 1/2 years ago, I can now get an
LSI SCSI 320 board. Since some of my drives (which I also buy used on
ebay - never had any problems) are SCSI 320, that seems like the way
to go.

--lady margaret thatcher--

 

[Lady Margaret Thatcher]

I haven't been following PCI developments too closely. It __seems
that__ the two are identical, but perhaps not. I have already RTF web
sites.

Can anyone enlighten me.

maggie

 

[daytripper]

I haven't been following PCI developments too closely. It __seems
that__ the two are identical, but perhaps not. I have already RTF web
sites.

Can anyone enlighten me.

Regardless of the operating frequency, PCI-X Mode 1 uses a slightly different
bus-level protocol than PCI (all signals get a full bus tick to make
transfers, no need for "sneak clocks" in the io ring to make timing back to
the bus from hot-off-the-bus inputs).

The comparitively rare PCI-X Mode 2 is quite different electrically; PCI-X
Mode 1 and PCI use single-ended signaling while PCI-X Mode 2 uses differential
signaling, includes ECC at the bus level and is more packet oriented...

/daytripper

 

[Folkert Rienstra]

OK, so shoot me for top-posting. Here is a status report.

The only way I could get the "bad" board to get through a post was to
put it into the same system as the "good" board. When I first plugged
the good board into my system--in-progress, it too reported invalid
configuration.

That is normal for the LSI bios. They all do that.

But it was no problem to enter the setup menu and reset parameters.

With both boards in the same system, I could start up the
configuration menu from the good board,

That is one of my recommendations for getting a bad-flash board through post.

and that menu would also try to configure the bad board.

When I selected one of the two channels (these are dual-channel boards) in the
bad board, the system hung, just as it hung because the bad board wouldn't POST.

When I restarted my system, both boards went through POST and both
boards searched for SCSI devices. Since the new system is still being
built up, there were no SCSI devices present. What is interesting,
though, is that the good board would scan SCSI IDs at about 1 per
second, but the bad board took about 30 seconds per ID. And at
15 x 2 IDs (dual channel boards, remember), that's a long time.

When I pulled the good board out of the system, the bad board
refused to POST, just as before.

I should also mention that once I got the system to post with both
boards, the first thing I did was back up the BIOS from the good
board, at rev 4.16 and use that BIOS to flash the bad board, which
was at 4.19.

All the above behavior about configuration and SCSI ID scanning,
was with the bad board already at 4.16.

I'll take that as 'after you flashed it with the other boards image'.

So in the end, I have a board that may be useful as a host adapter,
but can't be configured.

That may be it's problem. A bad CMOS chip or CMOS image.
I have had that happen with Compaq boards.
That particular Compaq boards had a smaller CMOS chip and the
full blown LSI bios couldn't run with that (configuration error).
It would run with the small bios from LSI though.
Next I managed to kill one of those boards when I ran a flash program
from TeKram. Just starting the program already killed it.
The configuration errors now would appear with the small LSI image too.
The only way to get that board functional with the small LSI image again
was to physically remove the CMOS (Serial EEPROM) chip.

So, it goes into my "dud parts" pile.

You can try the 32kB image and see if that works.
There will not be any configuration utility but it will boot.

And if you don't need it to boot you may be able to set it inoperable from
the other card's bios utility so it doesn't try to activate itself next time.
May or may not work but worth a try.

 

[Lady Margaret Thatcher]

Regardless of the operating frequency, PCI-X Mode 1 uses a slightly different
bus-level protocol than PCI (all signals get a full bus tick to make
transfers, no need for "sneak clocks" in the io ring to make timing back to
the bus from hot-off-the-bus inputs).

The comparitively rare PCI-X Mode 2 is quite different electrically; PCI-X
Mode 1 and PCI use single-ended signaling while PCI-X Mode 2 uses differential
signaling, includes ECC at the bus level and is more packet oriented...

/daytripper

Thanks. I guess my next question is:

If I have a motherboard with 64-bit wide PCI slots, but without any
mention of PCI-X in its specs, can I use a PCI-X board in this
motherboard? The motherboard is an ASUS A7M-266D.

Thanks.

 

[J. Clarke]

Thanks. I guess my next question is:

If I have a motherboard with 64-bit wide PCI slots, but without any
mention of PCI-X in its specs, can I use a PCI-X board in this
motherboard? The motherboard is an ASUS A7M-266D.

Your board uses the AMD-768 chip, which supports PCI 2.2. PCI-X is
_supposed_ to be backward compatible with PCI 2.2 so PCI-X boards _should_
work fine. What _should_ happen and what _does_ happen are not always the
same, but in general PCI-X boards should work

 

[Lady Margaret Thatcher]

That is normal for the LSI bios. They all do that.

Yes, it seems that whenever you first insert an LSI board into a new
system, the board reports "invalid configuration."

That is one of my recommendations for getting a bad-flash board through post.

Well, I figured it out on my own, and it did work. But imagine the
poor slob who doesn't have a second board. What is he to do?

I'll take that as 'after you flashed it with the other boards image'.

Yes, I should have been more explicit. All the above behavior was
AFTER I flashed the 'bad' board with the 4.16 BIOS, as copied directly
from the 'good' board.

That may be it's problem. A bad CMOS chip or CMOS image.
I have had that happen with Compaq boards.
That particular Compaq boards had a smaller CMOS chip and the
full blown LSI bios couldn't run with that (configuration error).
It would run with the small bios from LSI though.

My two LSI boards are the identical model. I believe that this board
uses a 64 KB BIOS. Which BIOS would I pick up for a 32 KB image?

You can try the 32kB image and see if that works.
There will not be any configuration utility but it will boot.

OK. That would be inconvenient, but I could always move the good
board into the bad board system when I needed to reconfigure the bad
board BIOS.

And if you don't need it to boot you may be able to set it inoperable from
the other card's bios utility so it doesn't try to activate itself next time.
May or may not work but worth a try.

That seems like a good idea. That way I could have 4 SCSI channels in
my PC.

Thanks.

--maggie--

 

[Folkert Rienstra]

Yes, it seems that whenever you first insert an LSI board into a new
system, the board reports "invalid configuration."


Well, I figured it out on my own, and it did work. But imagine the
poor slob who doesn't have a second board. What is he to do?

Ok, my fault, apparently I wasn't clear.
That is *one* of my recommendations. Better so?

Yes, I should have been more explicit. All the above behavior was
AFTER I flashed the 'bad' board with the 4.16 BIOS, as copied directly
from the 'good' board.


My two LSI boards are the identical model. I believe that this board
uses a 64 KB BIOS. Which BIOS would I pick up for a 32 KB image?

The one at LSI that says 32kB or less, maybe?

OK. That would be inconvenient, but I could always move the good
board into the bad board system when I needed to reconfigure the bad
board BIOS.

There is no reconfiguring the 32kB bios.
It doesn't use the CMOS, (well, not for configuration, afaict).
As to whether the controller will work in the OS may still be questionable
as the driver will examine the CMOS and may get stuck on that.

To be clear, this is with the current bios. The 32kB bios can't be disabled.

That seems like a good idea. That way I could have 4 SCSI channels in
my PC.

Or use it with an IDE drive for boot in another system.

 

[Joe Average]

Have you tried it in a different PCI slot? Or a different motherboard? With only
one CPU installed?

No, I haven't tried any of those options. But I certainly will. It's
a bit of a pain inserting the heatsink/fan for the CPUs because of the
spring-load holddown design, but that's a minor detail.