Hardware Analysis - I have now tested a Tyan Thunder K7 board...

CPU posting at lower speeds

chris mason — 2002-02-10T17:11:25-05:00

This is NOT a MOBO or Power supply problem.

Jumers ju-52 and ju-53 are backwards.

Just switch the jumper settings on the mobo and your problem will go away.

I had two xp1800's posting at 1350 mhz.. did the jumer switch and they now post at 1533.

Fustrated with Tyan Thunder K7 (times three)

James Taylor — 2001-11-01T17:25:33-05:00

Receintly I ordered parts for 3 servers (so, everything listed x 3):

Generic 19" rackmount case
NMB SD025A460WSW 460W PS (tyan recommended)
Tyan 2462NG (no scsi) bios v2.06
2x 1ghz Athlon MP w CoolerMaster heatsink/fans (amd approved)
512 Kingston pc2100 ddr ram (tyan approved)
Adaptec 1200A PCI raid card
2x 40G ATA100 7200 rpm WD400, running raid 1
OnBoard Video
OnBoard Nic
Creative Labs 52x CDRom
Mitsumi 3.5" fdd

----------------------

To make a short story long. This same story goes for -three- systems. Each one has the -exact- same problems.

I get everything installed hardware wise and I go to power up the system for the first time. I plug the power in (which is off of a APC 1000UPS) and the only thing that happens is the nic leds flicker for a second. No fans turn, nothing. I unplugged everything and still the same result. I finally found a FAQ on a site and found that if I did the "Clear CMOS" jumper then about 2 seconds after I plugged the system in it would come on.

Then I noticed that none of the LED's on the chassis were working, nor the reset or power switch. I removed power, studied the jumper leads for the leds, etc, flipped them around for the heck of it, and tried again. Still the same problem. None of the LED's came on, nor would the power/reset switch work. I know the HDD LED's work, because they work fine when I plug them into the RAID card. Maybe since I only have the cdrom as on the ide channel that's why I don't get a HDD light? *shrug*

If I remove power for too long, then I have to reset the CMOS to try to get it to power back on. So far we haven't had a blackout here so I'm lucky. I hope we don't lose power here for more than 5 minutes after the UPS would die....

So, to combat these problems, I figured I'd try to upgrade the BIOS to the 2.08 thats out. Bad idea. I used a Win98 boot disk (removed everything out of the autoexec and config so nothing would be loaded) and ran the flash utility (phlash16.exe) off of Tyans web site. I used the 2462v208.rom file with the util. About 30 seconds into the actual flashing, I got an error:

"Undefined error -8956 Please reboot with crisis diskette"

So, I followed some other directions for a recovery. I put the 2462v208.rom on a blank disk, and held ctrl+home as I applied power. Unfortunately it doesn't even -try- to read the fdd. Now I know the fdd works, because how could I have gotten to the flash part if it didn't? Now the board is dead. No activity other than the fans coming on, but it still has the same power problem as listed before.

I've emailed Tyan, about the LED and switch problems, but they haven't gotten back to me (and it's been a week). I also mentioned the BIOS update problem. I'm thinking I'll just have to RMA the dead board, but what about the other two?

On one of the other servers I went ahead and loaded win2k/exch2k and it seems to be -very- stable. I just pray the power doesn't go off for too long.

Anyway, I guess I'm just wondering if anyone has any other input as to what to do. I'm scarred to flash the other two boxes to 2.08 because of what happened to the first one.

Thanks
James

Thunder K7 issues (or not)

James Howard — 2001-07-29T20:18:58-05:00

I have been using the K7 Thunder since early June as part of the pre-release program with AMD and had very few problems with this setup. Granted, all of the systems I am building are being used as servers so problems with various video cards etc. are not something I would encounter. Currently I have either built or performed final QA on 15 different systems and had no problems with the exception of one board having a DOA Adpatec controller (would not POST with the controller active)

Allof these boards were from the original runs that shipped with BIOS 1.1. Upgrading to 2.06 resolved most problems that I encountered.

As far as power supplies go I am using the NMB's and again have had no issues with these either. I will state that this board is extremley picky about how the DIMM slots are populated due to 'loading issues' but this ws resolved with one call to Tyan's west coast office and a conversation with the field engineer.

The majority of problems I encountered were either improper jumper settings for the bus speed resulting in either no POST or improprer CPU speeds, instability due to overloading the memory bus (corrected by changing memory vendors while waiting on a updated module design fro Dataram), and ipmproper BIOS settings by production staff.

Here is a list of components being used in these servers.

K7 Thunder board Rev A. and Rev B
BIOS 2.06
Dataram 256 and 512M registered ECC modules (1G in testing now)
1G and 1.2H Athlon MP CPU
JMC 400020 H/S (listed up to 1.4G on AMD recommended list)
Netgear G620 Gigabit ethernet card
AMI Elite 1600 and Enterprise 1600 RAID cards
various IBM, Seagate, and Quantum U160 SCSI hard drives
Seagate and Sony DAT drives on second Adaptec channel
Sony AIT drives on second Adaptec channel

Operating systems:
Windows 2000 server and advanced server SP2 with pre-SP3 hotfixes
Red Hat linux 6.2 (w/modified 2.2.19 kernel)
Red hat linux 71.(w/modified 2.4.6 kernel)
FreeBSD 4.3

The kernel modifications are specific to our companies needs, the systems were also tested with stock kernels with no issues.

I find this board to be as good as can be expected for a first run on a new design. I have a few friends who are building systems around the K7 Thunder for 3D animation and CAD work. I will post more info about how they fare as soon as I hear from them

Update: My final look at Thunder K7 Rev. A

Brendan Harnett — 2001-07-24T19:18:10-05:00

Finally, after no less than two nights of tweaking and prodding, I was able to get the Thunder K7 board to operate properly. This did require a BIOS flash, several CMOS clears, and some toying with Onboard Ati vs. The Gloria III card. Never the less, the board does actually work now, and is as stable as I have ever seen a motherboard.

I'm honestly not sure if the boards are flawed. This one, apperantly, is not. Instead Tyan has just made a board with lots of 'qurirks' rather than 'bugs' that is a configuration nightmare. Every time everything seems to work fine, some new 'error' pops up that refers to nothing you can manually change. So it's time to clear the CMOS again...

I've got two more, now. This time, revision B boards. I'll keep you all posted as to how those turn out. Hopefully, they will go a bit faster, but I will be attempting to set up RAID sripe sets on both of them using the very same Adaptec card everyone seems to be having problems with. Wish me luck.

If these go well, I'll stop posting my warnings about this board all over the web. If they do not... well, that's what test systems are supposed to determine. I'm just a little ticked off that Tyan would bundle such a finicky board with such a poor manual. Honestly, the manuals they used to send with super 7 systems were far superior. I have set up hundreds of machines, and this thing had me baffled for about 12 hours! In the end, all my problems were what Tyan would call 'user error.'

What about people with less experience? If I'm generating this much 'error' what about the people who decide that THIS will be their first home built system? Of course Tyan's going to hear about it.

I know a WS/Server board like this isn't supposed to go to home users, but what did AMD and Tyan thin would happen? For many, this is the most anticipated Mobo release in years!

Tyan Thunder K7, Something Not Right?

dieu dieu — 2001-07-24T05:10:20-05:00

I have been investigating and testing the K7 Thunder two weeks prior to its official release with anecdotal results. I must say that this debate is quickly becoming more heated than politics, or religion itself. So heated, that I have received flame mail by users who claim that we unlucky users are nothing short of contemptuous witch hunters. Unfortunately, these few users are the minority. Regardless of the limited number of satisfied K7 owners, there lies an equally numbered group of unhappy customers. So great, that I have personally offered my telephone number to those who seem to be at a loss with the K7 Thunder. The areas of expertise range from system integrators, business owners, website owners, to end-users. This list is hardly to be considered by Tyan as simply end user-error. Are we all senseless fools who lack the basic intelligence to the proper design and configuration of an AMD based rig? Is it simply a coincidence that many websites are reporting the exact same phenomenon with this product? Is it odd the K7 Thunder is already at a revision �B� so early in its life cycle? Is it unusual that Tyan has changed some of the physical features of the motherboard? Is it odd that the K7 Thunder has had numerous BIOS upgrades the first week of its launch? Is it odd that Tyan is placing the weight of these issues on a power supply manufacturer; while in the same breath, denying that any of these issues exist? And secondly, where is AMD during all this unrest?

I have personally lived the nightmare that many of you are going through. The difference is, I reported these issues much before any of this became common knowledge. When I mentioned these problems to Tyan, they simply regarded all my concerns as isolated end-user error. When my first K7 Thunder failed to operate, Tyan refused to work with or replace said motherboard. Their comments were that they never experienced any problems with the K7 Thunder; therefore it must be user error. It is important to note that my problems where spontaneous looping reboots, black screens, and error messages that Tyan couldn�t figure out. To date, I am still baffled by the sudden death of my new Prophet III video card that failed to operate shortly after being inserted in the first K7 Thunder. Later, it was determined that the Prophet III was damaged and needed to be replaced. So, that was another $400.00 down the drain. Subsequently, the second K7 Thunder has worked with no abnormalities whatsoever. It is also interesting to note that no configuration changes were made between the first and second K7 Thunder. That is, I used the exact same equipment with a much different result. The result was that everything worked and continues to work according to any reasonable expectation. What does this tell us? It tells us that there are certain K7 Thunder�s circulating that provide a much undeserving end user experience.

Lets all hope you pull the long straw�

DieU
http://www.AMDmb.com

[No Subject]

Sander Sassen — 2001-07-23T21:56:22-05:00

Hi Brandon,

Thanks for taking the time to respond in this thread, I know you're a busy guy so it's very much appreciated. But allow me to comment on some of your suggestions.

First off, let me just make sure I make it very clear that I've FIRST tried two different, new, 460W NMB powersupplies, both manufactured at a different date, so not from the same 'bad' batch, if any. BEFORE trying out other WTX powersupplies, one 350watts, from a dual Xeon workstation, and two 300watts, from two Compaq workstations. Only to find out that they had the same issues as the NMB powersupplies. So if something damaged the motherboard it was the NMBs, not the others. I just tried the others to see whether the NMB powersupplies could be the cause of our problems.

Furthermore I've rebuilt the entire system three times and you can be sure that I know how. I've been building computers for the past ten years and I'm confident I'm skilled enough to tackle any issues, whether hardware or software. All the suggestions you've made I've already checked into and made sure that those aren't a problem. The ONLY thing that we're using that is not on the compatibility list is the Crucial Registered DDR memory, but my contact at Crucial tells me that they've done extensive testing on their memory and Tyan's Thunder K7 and have seen no problems or compatibility issues. Furthermore 100s of people use Crucial memory in their Thunder K7s with no problems whatsoever, and yes, I've also replaced the memory already to see whether that could be the cause.

To me it all looks like I've got a defective motherboard that somehow passed all QA tests but still has quirks. Friends of mine have a similar setup like ours, even with the same Crucial memory, and have no issues at all. Like I mentioned I've replaced all parts one by one, to see what could be the cause of our problems, BUT the motherboard, so to my opinion all of our problems are just due to the motherboard and should be fixed once we get a replacement.

Looking forward to your reply.

Thanks and best regards,

Sander Sassen

CEO, Hardware Analysis
Email: ssassen@hardwareanalysis.com
Visit us at: http://www.hardwareanalysis.com

Tyan's reponse to the problems reported.

Brandon Hammersley — 2001-07-23T19:01:10-05:00

The majority of the issues reported have been one of the following:

1) BIOS mis-reporting CPU speed.
This is caused by not setting the CPU speed jumpers correctly. J52 & J53 were somewhat confusing in the manual. An addendum has been published. ftp://ftp.tyan.com/manuals/thunder/a_s2462_100.pdf
2) System not getting power when power button is pressed.
This can be caused by not having proper loading for the NMB power supply. NMB has an app note on their website & Tyan has published an FAQ. http://www.nmbtech.com/cgi-bin/htmlos.cgi/002031.1.1835820677
http://www.tyan.com/support/html/f_th_k7.html
3) System not functioning correctly, ie lockups, reboots or other instability.
In most cases this can be attributed to not using approved components;
for example CPU's, Heatsinks, Memory, Power Supplies etc.; or not using the latest BIOS version. I can't tell you how many times I have had trouble calls of instability and it ended up being the heatsink was not installed correctly on the CPU, using too much thermal grease or no thermal grease.
http://www.tyan.com/support/html/b_th_k7.html
Always remember to clear the cmos after flashing the BIOS by unplugging system power & changing J21.

As for your board Sander, perhaps one of the variety of power supplies you tested damaged something on the board. As you stated "I tried a whole variety of different powersupplies, as the Tyan board simply uses a WTX power connector, so just about any WTX powersupply goes. Even a 300-watt WTX powersupply wasn't able to power the motherboard with just the two CPUs and the memory, it wouldn't even boot properly."
I will contact you by email to resolve this issue

Thunder K7 Results

Jim Nesta — 2001-07-21T03:22:23-05:00

I built the system myself about a week ago and everything installed fine and is running with no problems at all. The mobo and processors are from Monarch Computer. Here are the system's specs:

Tyan Thunder K7 (S2462UNG) - bios ver 2.06, FSB 266 MHz
2 X 1.2GHz Athlon MP processors with Taisol CGK760.. HS
2 X 8cm HSF (36 CFM) with 6 -> 8cm adapters
3 X 256 MB PC 2100 Crucial Registered DIMMS
Sony FDD
Plextor Ultraplex Wide
DPT Millennium SCSI RAID card w/ 4 Cheetahs on three channels
ASUS V8200 pure GF3 card
Adaptec 2930U narrow SCSI card
100 MB internal SCSI Zip
NMB 460W PS
Lian Li PC 70 case
W2K Pro(sp2).

The onboard Ethernet is disabled, while the SCSI is enabled. I loaded Nvidia ref drivers 12.41 to run the video card. I've never had a problem with this system. It is very fast and stable. This is only a test setup right now; I'm going to rebuild it in a couple of weeks when I'm ready to optimuze drivers and settings. I'll let you know if there are any problems.

[No Subject]

Sander Sassen — 2001-07-20T22:49:23-05:00

Oh, one other site I noticed has something up on the Thunder K7 issues, Chris can be trusted for telling things like it is and he's reporting the same set of problems some of us have.

http://www.gamepc.com/reviews/hardware_review.asp?review=thund...d=&tp=

Have a good weekend,

Sander Sassen

CEO, Hardware Analysis
Email: ssassen@hardwareanalysis.com
Visit us at: http://www.hardwareanalysis.com

[No Subject]

AfterBurn — 2001-07-20T05:57:24-05:00

Following the discussion on AMDmb.com forums, it seems that most, if not all, problems are directly related to the NMB powersupply, and NOT the Tyan boards, so I wouldnt be so sure that swapping the mobo will magically cure your problems, but a new PSU might. Of course there are always exceptions to the rule.

[No Subject]

Sander Sassen — 2001-07-19T22:22:30-05:00

Okay,

We got a reply from Brandon at Tyan, I sent him all the info requested and also asked whether he could provide us with a new motherboard as we think ours is indeed defective. I'm still waiting for a reply and am hoping he'll be able to discuss some of the problems on the phone.

One thing that puzzled me though is the fact that he mentioned he has read this thread and the news item yet failed to respond on this own. I had to send an email to tech support once more to get a response from them. That, to me, is rather odd, I'd have jumped on the occasion and would've tried my best to help out in every possible way. But thats me, Tyan might have a different perspective.

Best regards,

CEO, Hardware Analysis
Email: ssassen@hardwareanalysis.com
Visit us at: http://www.hardwareanalysis.com

Thunder K7 working stable

Joost Buijs — 2001-07-19T09:51:02-05:00

The 10th of june I bought a Tyan Thunder K7 at a local dealer here in The Netherlands. I don't know what revision it is. It came with BIOS v 2.01 and still has the metal levers on the ZIF-sockets. I took me a couple of weeks to get it running because the PSU was not available. The Athlon-MP is also unavailable overhere so I am using 2 Ahtlon 1200 AXIA. The system is running now for about 2.5 weeks and is completely stable. No issues at all. The only problem I had was that it posted at 1066 MHz. (120 Mhz. FSB) instead of 1200 Mhz. This was my own mistake I interchanged the settings for J52 and J53.

Tyan Thunder K7 BIOS v 2.06
PSU NMB SD025A-460WSW
Memory 2x 256 Mb. Crucial registered DDR PC2100
CPU 2x Athlon 1200 AXIA
Cooler 2x Spire copper 5P53B3 (http://www.spire.nl)
Case Addtronics 8600WTX.
Quantum Atlas V 18.3 Gb.
Plextor Ultraplex 40
Plextor Plexwriter 12/4/32
Asus V7700 Geforce 2 GTS
Soundblaster live player 5.1
Diamond HCF 56K PCI modem

Tyan k7

Mike Jacobs — 2001-07-18T18:51:55-05:00

I just ordered the NG version of the board, w/o onboard scsi, and I'll let everybody know how it goes. I figure/hope it's one of the latest revisions of the board, given that the boards w/o scsi didn't ship until much later.

Bummer about the bad boards...looks like I got lucky.

Dennis Brletich — 2001-07-16T15:23:35-05:00

I have been using my K7 Thunder for about a week non stop and have not had any problems.

I must have gotten one of the new revision boards that I mail ordered from http://www.unitedmicro.com

I bought one of their combo's I saw on Pricewatch.

Came with the Mobo, 1.2gig Palomino's, Volcano II heatsink/fan. They matched a deal for the NMB power supply too, since one of their customer service guy's mistakenly told me they didn't have it for sale there.

On my purchase order it even says "Power Supply 460WT NMB ATX tested OK." Don't know why it said ATX, because it didn't have ATX connectors.

My board didn't have a metal or plastic lever, but rather plastic clip holders on each side that made clipping the Volcano II's very easy dispite the closeness of the capacitors by one socket. You will have to remove the thermal tape and replace with gease, though since you can't overclock this board, you may just want to leave the tape on.

The boards BIOS was v1.07 not v.2.06, so you will want to flash it on the first boot as this will save you a lot of reboots and clearing the CMOS.

The Software did not come with the new BIOS but it did have an AMD 762 northbridge drivers and AGP patch as well as a butt load of LAN drivers and diagnostics for the 3com LAN jacks as well as a utility for booting you system over a network.

I've only been building systems since 1995 and am by no means an expert. This is my first dual MP solution and the only problem I had was flashing the new bios, but that was do to my own ignorance as I have never flashed a W2K system bios before.

The only other gripe I've got is that the SCSI runs slower then the ATA100 but I still have to update the new drivers from Tyans website which may fix this.

This system has been extreemly impressive with multitasking and is a joy to watch the MP's splitting the chores in the Task Manager when running multiple Apps. I can even Burn CD's while running UD client in the background, surfing the web and watching the stock market on my tv tuner card, with out missing a beat. The Ram hardly gets used. I have it networked to two other computers as well running Win98SE, Win98, and everything works great.

It looks as though there are some legit issues with the board in one of it's releases.

I set up each component one by one and installed drivers and optimized before moving to next component. Graphics card was last. I used the onboard ATI first.

Make sure your DDR is Registered.

Good Luck All,

Dennis

OS=W2K Pro
NMB 460W PSU
Tyan K7 Thunder S2462UNG
ATX Supercase Full Tower (Not recomended)
AMD Athlon 1.2mhz MP Palomino's X 2
Volcano II heatsink/fans
1024MB Registered Crucial PC2100 DDR ECC(4X256)
Onboard Adaptec AIC-7899W SCSI controller
9.1G IBM Ultrastar Ultra160 LVD SCSI-3 10,000rpm/4MB buffer (Not recommended)
40G IBM Deskstar ATA100/7200rpm/4MB buffer
Chaintech AGP GeForce 2MX 200 32MB SDRAM (Not recommended)
WinGo TV Tuner (Not recommended)
Iomeg 100MB Zip Drive
Acer 52X CDROM
TDK 12x10x32x32 CDRW
SB Live! 5.1 Sound
AltecLansing SurroundSound Speaker System
Teac 3.5" Floppy
Netgear NIC (Just wanted to see if it would work...it does.)
3Com USR 56K Data/Voice/Fax
A butt-load of case fans
Microsoft Natural Keyboard
Logetech Optical Wheel Mouse
Samsung SyncMaster 955df 19" monitor

[No Subject]

Robert Kropiewnicki — 2001-07-16T14:52:05-05:00

Interesting post from DieU on the amdmb.com forums. For those who are not aware, DieU has been one of the main people involved in troubleshooting the Tyan Thunder issues.....

"What power supply are you using?

Without waiting for NMB's statement regarding the power supply issue, I will say that they are aware of certain compatibility issues with their power supply.

Essentially, the NMB power supply requires a minimum current for proper operation. In other words, it requires that enough devices be connected to the power supply to establish a load. If one were to setup the machine with little or no devices to draw current, the power supply may not turn on. It was indicated that the power supply was designed to shut down. Though this phenomenon is atypical, its effect is temporary.

NMB and Tyan are also working together to determine some of the voltage issues that may be responsible for system stability. It seems as though the +12V fan connectors may be suspect to system instability. NMB and Tyan are investigating the voltage relationship between the four +12V fan connectors above the processor sockets and between the 8-pin and 24-pin connectors of the power supply. Currently, NMB believes that this problem can be isolated if they determine which of the four +12V fan connectors draws current from either the 8-pin or 24-pin connectors of the power supply. This may be an issue of load balancing if you will, between the 8 and 24-pin connectors of the power supply.

Though much of this may make little sense to the non-engineering/electrical type, NMB is going to release an Application Note on their website on Tuesday. This should help clear up any uncertainty or question you may have.

As more information develops, I will update this information accordingly. "

The whole post was in italics so it may be a response he had gotten from Tyan to whom he had sent out for information.