K8WE SCSI not working!

Hi guys,

I have finally kicked my K8WE into action today after several issues like memory and PSU choice.

BIOS: 1.01
RAM: KVR400S4R3A/512 x 2
PSU: Silverstone ST65ZF
CPU: AMD 248 x 2

It is working but …the SCSI controller doesn’t seem to find any of the connected drives.

I have tried the same drives driven by an old Adaptec 2940 Controller card placed in slot 2 (33 MHz) and they work fine.

Tyan doesn’t tell you this before you buy it but the onboard SCSI controller runs off the 100 MHZ PCI-X bus.

I wonder if this has anything to do with my problem.

The manual doesn’t incorporate any information about the SCSI controller (a bit poor for a 640 Euro mobo) so I am hoping that someone smarter than me may have some suggestions.

This is my first experience on a workstation mobo so please bear with me if I am not too familiar with some of the BIOS / mobo settings.

Thank you all

RMA the board.
It's most likely a bad controller. Let's just say Tyan and testing are fundamentally incompatible, and there's a high chance you would be better served by a Supermicro H8DC8.

Also, while it's off the 8131's 100MHz bus, it runs at 33Mhz or 66Mhz, mood depending. It also makes the bus highly unstable with additional devices. So if you have to, for some insane reason, have the blatantly inferior (8131 vs 8132 is NOT a contest) K8WE, bear that in mind.

I always though Tyan was really reliable?

Key word, was.
Since then, they have really, really gone downhill. I have not run into any of their Opteron boards that has not either had various problems, be them minor or major, or has had failed promises galore. (The K8QS Pro was originally promised to be dual core with a BIOS update. Surprise!)

Furthermore,
With SATA storage connected I get the following error message:

> Out of space for option ROMs
> Option ROM for the following device-PCI Mass Storage Controller on motherboard
> Bus 0A, Device 06, Functions 00 Was unable to run due to memory constraint.

…..so SCSI does not work.

If SATA is disabled then SCSI is available, Cntr.+C enters the menu, but upon scan for devices only No 7 (controller) is found. No DRIVES!
The drives work if on Adaptec Pci controller so they are fine.

The manual shipped with the M.Board is Rev 1.00 and does not even include SCSI contr. Instructions. Is there a later one?

BIOS: 1.01
CPU: AMD 248 x 2
RAM: KVR400S4R3A/512 x 2 (on PCU1)
PSU: Silverstone ST65ZF

As I said; RMA the board.

It will not work. The end. The K8WE has had the OROM bug since day one, and it is still not fixed. This is purely due to incompetence on Tyan's part. If it was a real design issue, the Supermicro H8DC8 and H8DCi would exhibit similar symptoms. They do not.

In other words: this board is not going to work for your usage, period. Return it and purchase a Supermicro H8DC8.

These are issues that have been long present in the K8WE and are still uncorrected. Despite their claims, OROM is still completely broken and unusable. PCI devices requiring two-stage OROM/Firmware absolutely do not work period. This board will not work, and will never work. The K8WE was a rushed product. Note the date of the last BIOS update; 10/13/2005. Tyan has already discontinued active support for this board - they've given up on getting it working.

First of all thanks for the time taken in answering me.

You are most probably right and I thank you for your concern.

But you will pardon me if I find it hard to believe that a group of engineers spent a few weeks locked up in a lab eating sandwiches drinking diet Coke to design and construct a board that does not do what they wanted it to do.
Even if so, who allowed it to leave the lab and reach the production line?
And more surprisingly the distribution department!

The reason I went for this board is that, due to my usage of Video Switching/Editing, I could do with what they promised to supply.

2 independent buses 100 & 133, one 33 bus, “ independent Sata controllers, 1394 x2 & Scsi” are all valuable in and out gates for video streams.

The H8DC8 does not specify independent buses, only one Sata controller & no 1394.
You see where I’m coming from?
Still, as you said, it’s no good to me if it doesn’t work.
Having said so, it does everything I expected minus the SCSI so I could simply exchange it for the non Scsi model.

I’ll still like to give it at least a chance (update bios and so on) and that’s why I am asking around to see if anybody out there has had it working.
Or have they all gone back?

Ps: from Italy, where I am, it takes forever to RMA anything, which is probably why I am reluctant.

I'd chime in on this, but I don't have any experience with the SCSI model. Mine doesn't have the SCSI controller. What troubleshooting steps have you taken concerning the SCSI drive problem?

I would find it hard to believe the SCSI controller is dead, but it is a possibility. I service a lot of systems, and I rarely find onboard motherboard components that just go bad. Especially out of the box. I'd double check your configuration, make sure you have the latest BIOS if you haven't already and go from there.

logic28 said:

But you will pardon me if I find it hard to believe that a group of engineers spent a few weeks locked up in a lab eating sandwiches drinking diet Coke to design and construct a board that does not do what they wanted it to do.
Even if so, who allowed it to leave the lab and reach the production line?
And more surprisingly the distribution department!

Click to expand...

I've been asking that of Tyan since release. They refuse to admit to the design defects, problems, or provide anything resembling adequate support. They're doing fire-and-dump; throw a 'hot' product out there, give it three months of support, then stop all work on it.

The H8DC8 does not specify independent buses, only one Sata controller & no 1394.
You see where I’m coming from?

Click to expand...

Yes, and I design high end workstations including NLE for a living.
The H8DC8 uses the AMD8132 PCI-X, which has two 133MHz buses. The H8DC8 is 2x 133MHz, 1x 100MHz arrangement, with PCI32/33 on the nF2200. The H8DC8 uses the AMD8132, which is VASTLY improved over the AMD8131.

Having said so, it does everything I expected minus the SCSI so I could simply exchange it for the non Scsi model.

Click to expand...

Again. Wrong. You need to listen. The board will not do what you need. Period. End of story, end of argument.

I’ll still like to give it at least a chance (update bios and so on) and that’s why I am asking around to see if anybody out there has had it working.
Or have they all gone back?

Click to expand...

Again; you're wasting your time. I've dealt more in depth with these boards than anyone else. These are not going to work for NLE. Especially with hardware encode/decode accelerators.

Ps: from Italy, where I am, it takes forever to RMA anything, which is probably why I am reluctant.

Click to expand...

*sigh* Try RMAing something with Tyan in the US. Their average turnaround on a support request by email last I checked was three months. And it's not gotten any better by all reports.

AreEss said:

Yes, and I design high end workstations including NLE for a living.
The H8DC8 uses the AMD8132 PCI-X, which has two 133MHz buses. The H8DC8 is 2x 133MHz, 1x 100MHz arrangement, with PCI32/33 on the nF2200. The H8DC8 uses the AMD8132, which is VASTLY improved over the AMD8131.

Click to expand...

I am not trying to antagonize you here, but just trying to clarify things.
The H8DC8 manual shows the 8132 driving 1x 100 Mhz bus and 1x2slots 133 Mhz bus not 2 133Mhz buses!

Since I have to work with multiple drives playing and capturing uncompressed video simultaneously (min 70 Mb/s each 100 preferred) it is essential to have multiple drives controllers fully independent from each other.
Again the H8DC8 only provides 1 controller x 4 ports Sata 3.0 Gb/s as opposed to 2 controllers x 2 Sata 3.0Gb of the Tyan.

My understanding is that each controller can truly carry 3Gb/s giving me a total of 6Gb/s simultaneous read or write or any combination of the two.
The H8DC8 only provides one controller for a total of 3Gb/s amongst the 4 drives. Correct?

It is also true that I could add a Pci Sata controller to the H8DC8 but I would be wasting a valuable 100 or 133 Mhz bus.
In addition I would have to add also a Pci 1394 controller, since the H8DC8 does not have one, using another valuable Pci-X slot.

So, once again, where would I be better served by the Supermicro? I do not understand.

Exactly what, beside my Scsi problem, could give me trouble on the K8WE whilst working away on the Newtek Vt4 platform?

Okay, let me just spell this out for you nice and simple.

The K8WE is not capable of 70MB/s on the Adaptec. It is not capable of 70MB/s on the SATA. It is just not capable, the end, quit wasting your time.

The only board that will work for you is the H8DC8 or H8DCi.

BusA, PCI-X 133MHz, Slot A1, A0 (bottom, second from bottom)
BusB, PCI-X 100MHz, Slot B0 (third from bottom)
BusC, PCIe nF2050, Bottom PCIe16x
BusD, PCI32/33, Slot D0, PCI slot.
BusE, PCIe NF2200, Top PCIe 16x, PCIe8x (4x signals)

And the only person you're fooling is yourself if you think you're going to get 3GB/s out of SATA, much less 6GB/s. That's just flat out wrong period. For one, even if it is PCIe, it's single or two lanes, which means you tap out at 500Mbit/sec anyways. That's before you factor in the drives, which are not going to give you anywhere near this mythical 3GB/s number anyways. It's called marketing bullshit.
Onboard SATA RAID? For NLE? Nope. Not going to work. Seen it tried. Hardware encoder or not, it's slow. >70MB/s consistent? Dream on. You need a real SCSI RAID controller, meaning an LSI 320-2X in SlotA0 on the H8DCi. The VT4 resides in SlotB0, and your IEEE1394 goes in the PCI32/33 slot or SlotA1.

Ok...Ok... keep your hair on..

I feel as if I've insulted you. It was not my intention.
Anyway

I was wrong!

With the latest Bios all the problems disappeared.

The board is now functionig in all its glory but I'm still testing it very carefully after the way you've punished me.

I promise I will never buy another Tyan, so if we ever meet at the local bar at least you'll buy me a drink.

I am seriously pondering over your kind advice but still testing the board.

By the way, I never said that I was expecting 3Gbit/s ( NOT G.BITES as you quoted) from Sata but meerly that whatever each controller can provide should double if you have 2 controllers. I assumed that the Supermicro having just one Sata contr. would have to share the data rate amongst 4 ports.
Was I wrong again?

I'll keep you posted

AreEss said:

It will not work. The end. The K8WE has had the OROM bug since day one, and it is still not fixed. This is purely due to incompetence on Tyan's part. If it was a real design issue, the Supermicro H8DC8 and H8DCi would exhibit similar symptoms. They do not.

Click to expand...

Does the Supermicro boards support booting from USB devices? Are the network boot proms enabled by default (both of them)?

Do you realise that the optional rom memory space comprises no more than 384KB that has to be shared between add-in card BIOSes, video memory (both graphics A000 through A7FF IIRC, and text mode B800 through BFFF) and system BIOS? There's not much space for any wiggle room, and enabling every bootable device is most likely a bad idea... (Tyan could be faulted for that I guess -- their default should be more conservative vis-a-vis memory) This is btw one of the main motivations why Intel and MS (among others) are drafting up a new specification for future BIOSes (or rather the lack of such) - EFI.

But you, being so knowledgable, could perhaps cough up a suggestion for how Tyan could actually fix their mishap?

BikeDude said:

Does the Supermicro boards support booting from USB devices? Are the network boot proms enabled by default (both of them)?

Click to expand...

Totally irrelevant on both counts. OROM can be loaded/unloaded as necessary, if the BIOS is done right. Booting from USB is an internal BIOS function.

BikeDude said:

But you, being so knowledgable, could perhaps cough up a suggestion for how Tyan could actually fix their mishap?

Click to expand...

Okay, you find out why they don't alloc OROM properly. I'm not a BIOS writer. That's ignoring the fact that the BIOS flat out pukes all over itself on certain OROMs, regardless of space. Which is, no surprise, the most frequently encountered bug.

Oh, and while you're at it, since you're apparently so cozy with Tyan, why don't you get us a new BIOS for the K8WE and email support replies in less than 6 weeks?

AreEss said:

Totally irrelevant on both counts. OROM can be loaded/unloaded as necessary, if the BIOS is done right. Booting from USB is an internal BIOS function.

Click to expand...

AFAICT, if you e.g. disable the network boot proms, they won't occupy any of the opt rom memory space. So Tyan do this already.

That's ignoring the fact that the BIOS flat out pukes all over itself on certain OROMs, regardless of space. Which is, no surprise, the most frequently encountered bug.

Click to expand...

I've seen similar issues with other boards. It's not an issue observed solely on Tyan boards.

Oh, and while you're at it, since you're apparently so cozy with Tyan, why don't you get us a new BIOS for the K8WE and email support replies in less than 6 weeks?

Click to expand...

Well, they do not answer my mails either, and what BIOS features are you missing..? (more features = more bloat = less space for opt rom = be careful about what you wish for)

I have an Adaptec 29160 controller, and none of the K8WE's BIOS revisions I've used have kept me from booting from one of my SCSI drives. What device(s) are you booting from AreEss?

BikeDude said:

AFAICT, if you e.g. disable the network boot proms, they won't occupy any of the opt rom memory space. So Tyan do this already.



I've seen similar issues with other boards. It's not an issue observed solely on Tyan boards.



Well, they do not answer my mails either, and what BIOS features are you missing..? (more features = more bloat = less space for opt rom = be careful about what you wish for)

I have an Adaptec 29160 controller, and none of the K8WE's BIOS revisions I've used have kept me from booting from one of my SCSI drives. What device(s) are you booting from AreEss?

Click to expand...

I had to disable the network boot roms in order to use my LSI MegaRAID U320-2 controller. And old controller, but still a solid performer. Definately boots correctly also.

Tyan really did need to include more option rom space though. I shouldn't have to disable anything to make a damned SCSI controller work.

Dan_D said:

Tyan really did need to include more option rom space though. I shouldn't have to disable anything to make a damned SCSI controller work.

Click to expand...

With bios v 1.02 you do not need to disable anything.

I am now using the onboard LSI scsi successfully

maz