Hi,

I have built a Tiger MP system with dual athlons (1.4GHz ), a Gig of registered corsair DRR RAM, twin 60 Gig Maxtor GL740s running on a Promise mirrored ATA Raid array, a Radeon 8500 and an HP DVD 100i burner.

My aim with this system is to capture video, edit and burn to DVD using Adobe Premier and a Raptor video capture card. I avoided a striped Raid array as my other work demands data security and I hoped that the performance even with a mirrored array would be enough for some serious video work.

I started to build the box on Xmas eve at 3am and by breakfast had XP whirrring away ready to present to the family and to get the bits off the kitchen table. It all worked great except for two minor issues - I couldn't format DVD media using the HP burner, and the video showed a strange glitch on start up in that the XP logon screen was mildly scrambled and seemingly VGA size, but fixed to 1024x768 by powering THE MONITOR on and off. I installed the latest drivers for the Radeon, all the HP fixes and was looking pretty, but the two problems ate away at me....

Yesteday I spent some excellent time with HP support and they fixed the DVD problem by adjusting my MB's BIOS for the DVD IDE settings to include PIO4 / and disabling UDMA - format problem fixed.

Next I read about the need to install MB specific AGP drivers for XP, and wondered if this would fix the video isue. So I downloaded TYAN's AGP driver for the s2460 ( w2K) and a registry update...

Bad mistake..

All seemed to be fine for about 1/2 hour and I was clearing up. I had captured video, burned a kids-opening-presents DVD and was able to watch it on my home DVD player - all as planned. Even the video glitch seemed fixed.

Then I noticed an occasional "tick" from one of the hard drives and the next thing was that on reboot the Promise setup suggested a critical problem with the array - one disk not registering. Patting myself on the back for mirroring the drives, and not believing there could be a problem with the drive, I rebuilt the array with the original drives. All fixed in a few minutes. But the problem recurred, with more frequent drive "ticks", so I took it off line and booted with only one drive from the array. To my horror, this one started making the same noises with finally the array not recognising any disks.

My solution was to rebuild XP from scratch thinking I'd screwed up with the AGP driver and being unsure how to reverse the registry update I'd downloaded from TYAN ( I rolled back the AGP drivers to my originals). All well with reinstalling XP until the final reboot when the array fails again, both drives sounding sick.

Thinking simple and hoping to use a Maxtor drive analysis tool, I took all the add-in cards and DVD drive out and booted to DOS with just a drive on the primary IDE and the Radeon connected. No joy - the drive was duff. Same with the other drive. Now I even had problems with just a floppy connected with spontaneous re-boots, so I took out the Radeon, installed my old and trusty Matrox card and reflashed the BIOS of the MP, resetting the CMOS.

This works. Even so, I had the occasional spontaneous reboot and the system seemed to take for ever to come back after resetting the CMOS ( is this normal?). But irrationally, it seems to have settled down. I have an ultimately powerful machine running DOS 5.0.

So my beloved system in pieces. ... Unsure that I haven't damaged the motherboard and certain I've done something horrible to the Maxtor drives. Can anyone help diagnose and solve this litany of woes? Fortunately this hasn't come out of my own pocket. I'm not a newbie - this is about the 5th machine I've built, but I've hit the wall.

Any help gladly received,

Thanks,

Iain Sanderson.

Iain,

I feel your pain, and I'll do my best to see if we can diagnose and solve your problem. But please see if you can answer the following questions and follow my suggestions:

1 - What kind of powersupply are you using, the Tiger MP needs a 300-watts powersupply at least. AMD-approved that is.

2 - What kind of memory are you using? Registered DDR or Unregistered? Registered will work, Unregistered is 50-50% change at best. Make sure you lay the board flat on the table with a towel or something underneath and then firmly press down the DIMMs, make sure the center piece is fully inserted.

3 - Tyan's BIOS reset isn't really working too well, better to remove the battery for about 2..5 mins with the power cable pulled then re-insert it. That'll make sure you start at factory-default settings.

4 - Memory should be inserted in bank 1 and 3 if you're using two 256MB DIMMs, not bank 1 and 2. If there's just one DIMM use bank 1, 2, 3, 4 and see which one works. The DIMM sockets seem to have caused more problems with other users.

5 - Do a clean install of Windows XP Pro, Home edition won't work as there's no support for SMP. Don't install any 3rd party drivers or registry updates from Tyan or AMD, XP has all the lastest already. Do download the K7 Update from Microsoft's XP Download site, it is among the non-crtitical updates.

6 - Format the HDs using Ontrack Disk Manager, the Promise BIOS writes an alternate FAT table to the drives that DOS Fdisk and Format don't always know how to interpret. The Disk Manager can be found on Maxtor's website under Utils, or just do a search for it on Google. If you can't find it, email me and I'll send it right over.

7 - What BIOS version and board revision Tiger MP do you have? Latest non-Beta BIOS is V1.03, I'd suggest updating if you have and older revision. Board revision can be found by looking at the PLL, the clock generator, if it ends with **-64 or **-54 its not a Rev 3. if it ends with **-155 it is.

8 - Only use WHQL certified drivers for your components and stay away from any 3rd party utils that could affect the system registry or substitute *.dll or *.sys files with alternate versions.

9 - I'm sure you'd have tried this already but try starting from scratch, just the motherboard, memory and a video card and then start adding devices to see which one might be defective or show odd behavior.

10 - Keep us updated, we'd be happy to help out, do list a full system spec. incl. BIOS revs. if you know them.

Good luck!

Sander Sassen

Email: ssassen@hardwareanalysis.com
Visit us at: http://www.hardwareanalysis.com

Sander,

Thank you for your excellent suggestions. Here's some data-

The Bios on the board is v1.3, which is what I reflashed with again from TYAN's website. I couldn't find the board revision you mentioned.

The Memory is a single Corsair 1GB chip of Registered ECC PC2100 DDR( quite expensive too) in DIMM Bank 1.

One of my original Maxtor Drives was damaged - I ran Maxtor's Powermax utility for a couple of days under DOS with no problems, but one drive showed up with an error, the other was OK. Maxtor (and CDW) agreed to take them back and sent me 2 Seagate ATA IV 60 GB drives as a replacement, which I now have.

I connected up the drives to the Promise FastTrack and created a mirrored array using its ROM-based utility. No problems. Then I connected the HP DVD drive and after booting up, the system spontaneously rebooted, without any video output, and with one of the drives making a repetitive "searching " Zing . I had to disconnect it all again and Reset the CMOS to recover.

I'm going to try again, without the Promise card and see what happens, using the low level utility you suggested to format the drives in a Master/Slave set up. Here Goes.

Regards,

Iain.

Sander,

No joy - I'm still having problems despite stripping the system down to its basics (again).

Could it be the power supply? It's a FSP 400watt unit which supposedly has more than enough for the Tiger Mobo. In the current iteration of my woes, I'm noticing that everything seems to be fine until there is a definite fluctuation in the processor heatsink fan speeds (a change of note) associated with disk access noise and then the system reboots. It seems as though the disk is drawing power and causes a power fluctuation in the Mobo, which causes a reboot. After the second spontaneous reboot I have to disconnect all the components and reset ther CMOS by removing the battery as you described.

Regards,

Iain.

Iain,

I will not claim to be knowledgeable at all about FSP power supplies so it's possible that the PSU is not the problem. But as I've told people time and time again in reference to system building, all PSU's at a given wattage rating are not the same.

Perhaps Sander has a recommendation on a PSU to try?

Iain,

I looked up the FSP for you and it is not the beefiest powersupply money can buy. I'm actually starting to wonder whether that might be the culprit here. I'd really advice you to go out and buy a 400-watts Antec or Sparkle and/or another manufacturer from AMD's approved list, it may be the best $60 investment you'll ever make.

Also, make 100% sure your heatsinks are rated for the CPUs, don't use exessive amounts of thermal paste, and NEVER use a shim, or spacer. Furthermore, be sure to check whether the heatsink's fans are spinning and that the sinks are at least warm to the touch.

Hope this helps!

Sander Sassen

Email: ssassen@hardwareanalysis.com
Visit us at: http://www.hardwareanalysis.com

What follows was sent to Tyan support but as yet I have received no response. Keep in mind that at present in Australia our temperatures are in excess of 30oC and the cooling is working hard. Note since sending this I rebuild using the 1.4 specification setting and this again caused considerable effort to load but the Blue Screen of Death problems have ceased. I also enabled the MP interrupt table. The system while stable still seems slow.

The system I have built is SCSI based with registered memory and has an IDE CD-RW. I did install the AGP driver without any problems. If any body has other information I would be interested. The suggestion is that some MP aware software may need some tweating and this may be a registry hack. I have yet to confirm this. Initially I did have sick stick of RAM.

one benchmark suggested latency issues on the PCI but did not help disbled mastering on Network and left SCSI host as a master made no difference

When loading Win2K use ONE stick of memory and turn of BIOS USB support use a PS/2 mouse

What went to Tyan

Your earlier prompt response has enabled me to get the system based on the S2460 board functioning. It has BIOS V 3. I have just completed a system using the board raises some minor issues some I understand relate to compatabilty at the outer edges of the specification in particular memory. The system will be using multi processor aware software.

The questions

1. In the BIOS Advanced settings there are two setting that I could not find reference to in the manual supplied since most refer to the single processor version of the board and these settings apply to multi processor settings located between USB and Chipset setup entries.

The first setting refers to the multi processor interface specification with a choice of 1.1 or 1.4. Who's specification Tyan or AMD and how does this affect the installation of Windows 2000 with SP2? Is there a list of Specification to OS compatability?

The second setting refers to using the PCI interrupt table in the MP I am uncertain as to the purpose and how it is likely to affect Windows 200 functionality. I would like some additional information on this point to ensure I do not have a full rebuild underway if it were to corrupt the installed Hardware Abstraction Layer in Windows 2000.

2. I would appreciate it if there is any addditional information on memory configuration?

The comments

I am using 2xAthlon 1600+MP processors. After a number of attempts to load Windows 2000 with many Blue Screens of Death, I succeeded after removing all memory except one stick. This was also useful not only in the setup stage but also in installing applications.

After finally loading Win 2k I reinstalled the additional memory and found that again the machine had persistent errors resulting in a range of Blue Screen of Death messages. The messages were not consistent. Due to the variability I had though that perhaps the processor temperature was causing some glitching. There was no pattern except maybe USB activity. Page faults seemed to be more evident than others and since this suggested memory I again pulled the memory and the result was a stable machine. I believe that I have eliminated processor instability as a contributor to the errors observed. This could be a problem to be resolved between AMD, Tyan and Microsoft since the Hardware Abstraction layer is probably deeply involved in these processes.

Due to indication of higher temperatures on processor 2 I swapped the processors and fans with no effect on the temperature situation on the processor in socket 2. I presume CPU 1 would carry most of the load at idle. This problem seems to be related to the socket since the differential is irrespective of airflow such as whether the box is open or closed up and hence location is probably not a problem. I had two scenarios until I thought about how the monitoring was achieved, from within the processor itself and the only conclusion is the voltage supply range on the second processor socket is wider and the higher core voltages registered could account for the increased dissipation indicated by the higher operating temperature. Unfortunately I am unable to continuously monitor the processor core voltage and temperature for both processors. Know any good utilities to monitor temperature and possible alert the users to any over temperature? I tried CoolCPU but this caused me more problems and was difficult to remove from Win 2000. Typically, processor 2 consistently runs 3-4C degrees hotter than processor 1. The core voltage range indicated over 5-10 mins was consistently 1.733-1.749V for processor 1 and processor 2 exhibited a similar range but with consistent excursions outside this range with lows below 1.65V and a high of 1.911V being observed using the BIOS monitor function. I did this on several occasions with similar results. This suggests perhaps the regulator has greater overshoot and a longer response time.

I would appreciate any suitable links or advice you may offer

Regards
Brendan Falvey

Brendan,

I wouldn't worry too much about MPS 1.1 or 1.4 as they are both used on various OSs and both work with Windows 2000. The Multi-Processor Specification (MPS) is a specification by which PC manufacturers design and build systems with two or more processors.

MPS version 1.4 added extended configuration tables to improve support for multiple PCI bus configurations and improve future expandability. Most newer versions of server operating systems support MPS 1.4 and, as such, you should change the BIOS setup from the default of 1.1 to 1.4 if your operating system supports the 1.4 version that is.

For instance, both Novell IntranetWare 4.11 and Microsoft Windows NT
Server 4.0 already supported the MPS 1.4 specification. So anything newer most certainly does. However if you use a different OS check with the vendor of your operating system to find out which version to use if you are unsure. Some operating systems may require version 1.1 for compatibility reasons.

As for the MP table settings don't worry, that'll just have the PCI entries show up in the MP table for easy troubleshooting. You'll be able to not only see the devices show up in the MP table but also the IRQs they use. For an ACPI aware OS such as Windows 2000 this is irrelavent as the OS controls IRQ routing, and it has nothing to do with actual IRQs anymore, the cards just think they got an IRQ assigned, in reality they're not. So the IRQs that were assigned during bootup are not the IRQs Windows 2000 will use.

Oh, CPU temperatures, don't worry about those either, CPU0 or 1 (depending on whether you have 0 and 1 or 1 and 2) is doing more work than the other simply because not all apps are SMP aware and even if they are don't use each CPU to the fullest. For example, my dual AMD has the first CPU running hotter than the 2nd also, it is not a problem. Also Tyan's Tiger is unlike most motherboards, use a Tyan approved monitoring util, Tyan has one available here:

http://www.tyan.com/support/html/sm_software.html

That about sums it up, if you want me to be more specific and to the point please post a list of your system specs. incl. the brand of your case and power supply. That will help us diagnose any problems right away, can't be much more specific than what you could've read in this thread .

Kind regards,

Sander Sassen

Email: ssassen@hardwareanalysis.com
Visit us at: http://www.hardwareanalysis.com

Sander

Thank you for the link to Tyan SM. The Tyan System Monitor utility while specified for the S2466 MB seems quite happy with the S2460 MB and confirmed some of my suspicions.

It confirmed the BIOS monitor indications that CPU 2 was running hotter. The greater voltage variation in the supply may in part be explained by the fact that the associated regulator (VRM2) is running at a much higher temperature also. I plan to investigate this with component cooler and see if I can improve the lot of the onboard components. Will keep you advised. I will also be adding two additional case fansd to improve airflow since the box has 3 SCSI HDD 1x7200rpm for system and 2x10k for data and pagefile.
Regards
Brendan falvey

Um, please help, im having a big issue with my system...

im running a tyan tiger mp system with a single mp 1600 processor, with 2 40g maxtor hds, and a raid controller running a RAID_0 config. When the XP professional startup screen loads where there is a bar with green cubes moving.... they show up and then stop, and the computer locks up?

i tried formatting and it does it again!

has anyone seen this before? what can i do to save this?

Kevin,

Do you have ACPI enabled? If so, turn it off and try it again. Let us know if that solves the problem.

k i dunno what ACPI is but i will find out =)

gimmie a couple days, i will get my friend to help me with that.

ill get back you yah soon =P

I just bought a Tiger MP s2460, 256mB (Crucial), one (1)-AMD Athlon 1800+MP, and an ANTEC 400wall ATX psu with 40amps on the +5.5v lead.

I can't get the thing to even POST. Sometimes it powers up when I plug in the power cable; other times, I push the power button. I get no video signal (riva TNT (soundblaster?)), and I don't see the HDD running other than its normal power up sequence. I've changed motherboards at the dealer, but still get the same problem.

do I NEED two procs to get it to work?

The salesmen at the store are no help whatsoever. Has anyone had this problem before?

Mark

tak,

What kind of cooling solution are you using on that CPU? Not an expert on dualies by any stretch but I'm pretty sure you should be able to run it with just one CPU. Have you tried switching which CPU socket the one CPU is in? Have you check to make sure your board isn't touching the case somewhere causing a short?

Thanks for the help...

I use a Thermaltake Dragon ORB III for cooling, but I don't think its the problem. I dont leave the machine long enough for it to get too hot, I think. I heard that ECC takes a long time to register RAM, so I've left it on in limbo-mode for about two minutes before; that's by far the longest.

I thought it might be my case, so I went out and bought an all-aluminum one. I don't see the motherboard touching the case other than thru the standoffs (bent metal clips secured with machine screws through the top of the motherboard. I even bought a new powersupply just to be sure. From what I hear, ANTEC is `good'. But it didn't work for me.

So I thought it was the memory .... Today, I went and got some Corsair RAM because I read on a newsgroup that Tyan boards don't work well with Crucial RAM. ... still SOL.

I've tried switching the CPU to CPU1 from CPU0, but still no luck. MANY others have had similar problems from what i read on the groups, but sooner or later, some seem to get it to work.

Strike one: PSU, so i bought new one
Strike two: case, so i bought new one
Strike three: motherboard, fiddled with ALL jumpers; exchanged for new one
Strike four(?): memory, bought better one

Next batter: bad cpu? two consecutive bad motherboards?

BTW, I've also tried the take everything out except the CPU method, but no luck; same exact outcome.

see that

i have the same Configuration

TYAN 2460 2x1600 MP ATHLONs
1GB DDR PC100

Matrox VC + RR G Seria

my PC Restart and have very big problems with blue screens and many more but for naw with that PC FAN is run normaly

see my solution on following URL http://photos.yahoo.com/bc/ps2x2/vwp?.dir=/myplace&.dnm=PC...p;.hires=t


for instalation WINDOWZE

you need instal MPS Standart PC WINDOZE instalation type no ACPI Multyprocesor SYS

that you can make on

ControlPalnel

System Monitur

Hardwer

Device Manager

Computer

Right Clic Property

Driver

Update Driver

From List

and chose MPS Multiprocesors PC

and i hope after that you need re instal all drivers from factorys for all your hardware all Drivers from TYAN for MB AGP PM etc...

AND DONT FORGATE THE COOLER ))))))))

Have fun after that ))))))

en Joy

MY Ram is DDR PC2100 Sorry for mistake

Wow, I just wrote out a long explanation for you guys, and then it got erased because I didn't confirm my account. So, here's the short version. Take a look at the way you have your fans oriented on you cpu's. I was having the same problem as you guys, and when I turned them fans the right way (lined up the grease with the cpu's) all worked fine. Hope this helps,

Jeff

Hi All,

Interesting to read everybody elses odyssey with the Tiger. just to sum up my experince over the last 2 months

In the setup it is important to disable ACPI functions and persist it may take a number of goes to load Win2k. Usually dies towards the end before it registers every thing. Once it had loaded it is usually fairly stable.

Cooling seems to be the prime source of instability. I replaced the original PSU with an Enermax that has a large fan to extract air from the immediate area of the processors and a wire grill exit that permits a better air flow than the usual pressed slots on most PSU. This resulted in a monitored temperature by the BIOS and System monitor (uncalibrated) utility some 3-5 degrees lower than the original installation. Now looking at other options to lower CPU temperatures.

Memory glitches remain a problem however these may be due to memory leakage since they seem to be time related. Having been in the electronics industry for the past 25 plus years I realise that the DDR clocking on the rise and fall means that the margins for the 133 MHz clock are very tight. Lets look at the pulse as a 5 nanosecond pulse in a 50% duty cycle and assume a rise and fall of 1 nano second. The 1 nano second in physical terms for the speed of light on transmission lines is about 20cm. Here we are talking about a physical event less than the size of the boards. Even with the memory bunched up the pulse margins are very tight and hence needs very careful consideration. I did receive a beta BIOS upgrade from Tyan however I was unable to get it to load either the flash program or data were corrupt or the Real DOS win 98 boot does not recognise 9.3 file names am still pursuing.

The box I built has provided significant iprovement in performance for my client. Fortunately he understands bleeding edge technology and accepts the minor inconvenience. His older PII/350+384 MB+IDE replaced by 2x1600MP+1GB+SCSI reduced a file creation process from 20 hr to 20 minutes. Problems aside the performance increase has been worth the extra grey but fewer hairs.

The SCSI drives are a boon but the 10K version do run hot adding to the cooling load in a closed box.

Just persist its worth it.
Regards
Brendan Falvey

Hello,

Very interesting tread, perhaps some one can help me too. Here’s my Xmas Story:

I put together several motherboards in the past 15 years or so, including the Tyan’s Tiger 133 with dual PIII 800s but my new one, the Tyan’s Tiger MP, is driving me up the wall.

Lets start from the beginning, I use the system for video editing with the Matrox’s RT2500 board and every thing was working fine with the Tiger 133 (after some minor BIOS tinkering and a few slots changes) that give me the confidence to purchase the MP with dual Athlons MP1800+. Here’s my system:

Tiger MP Board (O1MOAA) BIOS 1.03
Dual Athlons MP1800+ (One working on socket 1. Socket 2 is empty) the other CPU failed. Temperature on CPU #1 is 42 C. while on the empty #2 Socket is 32 C.
Thermaltake Volcano 6cu Heatsink
Two sticks of Corsair 512MB 2100 REGISTERED ECC DUAL AMD on slots 1 & 2.
Linksys Etherfast 10/100 LAN Card (For the cable modem).
SoundBlaster Live 5.1 (Driver Only).
Fasttrack 100TX2 Raid Card w/Two IBM 60GXP 40GB ATA 100 Drives. (For video clips only)
Adaptec 19160 Ultra 160 SCSI Card w/Two Seagate Barracudas 18GB Drives. (One for the system and editing software and the other one for ghosting. Two partions on each)
Matrox Millennium G450 AGP Video Card
One Maxtor DiamondMax® Plus 60GB ATA 100 on Primary IDE set as Master. (Mainly for audio clips and other music and data)
One Pioneer DVD Player 105S on Secondary IDE set as Master.
One HP DVD+RW 100i. Idle at present but working fine on the Tiger 133.
One IBM 75 GXP 45GB ATA 100 Drive. Also idle at present but fine on the Tiger 133.
One Startech 450 Watt Professional ATX Power Supply ATXPOW450PRO.
DC Out: +3.3V(30A)
: +5V(45A)
: +12V(15A)
: -5V(1A)
: -12V(1A)
: +5V(2.5A)
OS: Win XP Professional.

The first Tiger MP was received just before Xmas and I had to send it back right after the holidays due to a CMOS ??? problem. I had to reset the CMOS every time I added a card, removed one or after changing a card from one slot to another. It worked fine after the reset with all the hardware described above. It came to a stop when I added the Title Motion’s Dongle to the parallel port. At this time I couldn’t even enter the Bios Set UP.

I received the second Tiger MP at the beginning of February and it wouldn’t POST, lots of drives activities but the cursor kept on blinking on the black screen. That was when I removed the second CPU and BINGO, it came alive. I swapped the CPUS and the cursor blinking on the black screen did confirm that the second one was toasted. I figure that one CPU is better than none, right? Read on.

Reinstalled all the hardware, power up and during the POST two of my drives were not detected properly and the system would not boot up.
Removed the two drives and every thing worked as supposed. Went out, bought new IDE 40/80 cable, reinstalled the two drives and although during the POST all my drives were detected properly, same thing happen, would not boot up. It did not matter what combination I try, Master, Slave, Primary IDE, Secondary IDE, other cables … Nothing. Removed two of the drives from the IDE, boot up, downloaded the necessary drivers and software for my Matrox RT2500 card and shut down to install it and …. Here we go again. Did not matter on what slot I put it in, the POST will stop right after the memory check. With out the card, every thing works. I should mention that the RT2500 works well with the Tiger 133 and so were the two above listed drives, also that other RT2500 users have good success with this card on the Tiger MP.

Another peculiar thing is that it will not go into the BIOS Set Up until the SCSI and the Fasttrack go through their own POST. I never saw this happening with any other motherboards, including the Tiger 133. Is this normal? And if so, why?

I’m I doing something wrong? Did I miss something? Remember, with out the two drives and the RT2500, everything else works just fine and faaaaast. Please, any suggestions?

Thank you,

Ivan

1.I was wondering if anyone has found a good and reasonably cheap out-of -the-box cooling solution for the problem of dual athlons on the tyan tiger?
2. Also, does the tiger mpx board do any better (than the s2460)?
3. Does the bios 1.3 do better with thermal problems than 1.03?
4. Has anyone tried the $25/cpu solid peltier coolers (power=72W each!) or water cooling?
5. Would dual xeons or p4 's be a better solution? (I find the mp1900
only 1.8 times faster than 1GHz pIII for my application, which amounts to a <15% increase in speed for equivalent clockspeeds,
but I have no idea how much slower p4's are compared to PIII's).

Linux (RedHat7.2) seems to at least boot and act normally for dual mp1900s on tiger 2460; however, when pushing the system, especially with burnK7
(the ctcs linux diagnostic for athlons) and my application,
the system hangs after ~ 5 minutes. When
removing the case cover and putting in a cool breeze,
burnK7 runs for > 100 minutes. I stopped it at this point.

I'm using a Volcano 5 fan/heat sink, dual mp1900 tyan tiger s2460 bios 1.03.
Enl mid atx 400w case with extra front fan (has rear fan, power supply fan,
and front fan).

My experience is a bit different than the rest of this thread in that I see the
machine idling at about 45 degrees C (from the bios).
The cpu's heat up to > 79 degrees C (based on rebooting after crashing and looking at the bios set-up) when running burnK7 for ~ 5 minutes.
This seems much higer than the temperatures reported on this thread of 35-45 degrees.
6. Perhaps the mp1900 is significantly hotter than the mp1800, or the heat sink
(dealer-installed) is not properly attached to the cpus?

Finally, Jeff Humphrey had success by "turing the fans the right way ,(
lined up the grease with the cpus)".
7. Could he or anyone else elaborate on this?

Thanks in advance,

Kevin