added an AMD X2 to my server, now it regularly reboots itself. help!

[phelme]

anyone out there running an X2 processor in their server?

I had a super stable Sage server for a year, but then I wanted to handle more processes simultaneously like transcoding, comskip and the like so decided to put in a 3600+ X2 in my ECS board. bad move, since then the darn thing reboots 1 - 2 times a day. no panic event log entries, no blue screen (that I've managed to see, doesn't mean it's not happening), no dumps of any kind (yes, memory dumps are enabled). it just reboots. there is no pattern that I can find, it happens any time of day, with heavy load or no load.

so I've tried the following:

-upgrade to a super heavy duty CPU fan, core temp never gets over 36c, even while cranking on comskip, transcoding and HDHR recording simultaneously. usually sits around 16c at idle.
-upgraded ATI drivers for video
-upgraded HDHR firmware and software
-upgraded AMD software
-no new BIOS available
-no motherboard nForce chipset upgrades are out there

I'm frankly stumped. I thought either the ATI drivers or the HDHR upgrade would fix it, but no. It's probably the ECS nForce4 A939 board itself as it is a couple years old and the last firmware update is where they added X2 support. maybe it's just flaky. no crash events or memory dumps must mean it's something sudden & immediate, which is probably hardware. or something isn't happy with dual cores.

I'll probably run the nVidia stress test software for a long period of time and see if anything crops up. I've done 20 minute tests which showed nothing.

anyone have any ideas? if not, I guess I'll just remove the X2 or get a new board.

thanks in advance.

[mattdcknsn]

I would recommend installing Speedfan and have logging enabled. This should tell you if its a temperature problem. For stress testing Prime95 really puts a load on the computer.

Are the reboots happening at specific times? any predictability?

[MeInMaui]

I would consider doing a Repair Installation of Windows. There may be some driver problems with just changing the CPU. If you boot into your Windows CD and go through the installer, there should be an option to do a 'Repair Installation'. This will essentially reinstall a fresh copy of all of the windows files over your existing ones and set up the proper drivers for your motherboard, CPU, etc.

You will have to re-install of of the updates and security fixes after that, but then you should be good. Make sure you backup anything import on your system drive first (like wiz.bin or in fact your whole SageTV folder). HTH

FYI, here is a link to instructions for doing this when replacing your motherboard.

Aloha,
Mike

[PGPfan]

The repair install is a good idea, but be VERY careful about how you do it. Follow the link Mike posted above. I know you get 2 prompts for the 'repair install' and you DON'T want the first one - it will mess up the system in a big way (I believe it is called the 'Repair Console').

-PGPfan

[phelme]

Originally Posted by mattdcknsn

I would recommend installing Speedfan and have logging enabled. This should tell you if its a temperature problem. For stress testing Prime95 really puts a load on the computer.

Are the reboots happening at specific times? any predictability?

I already have SpeedFan installed, which is how I know the temperature isn't an issue. The new Arctic Cooling Freeze fan I have installed is the first thing I changed when the problems started. It never gets beyond 40c and usually runs around 28c when a process like Comskip is going. 16c at idle.

The reboots are completely random. Any time of day or night. Usually I get one a day. Sometimes two.

I would consider doing a Repair Installation of Windows.

I don't think it's Windows related, as I never had a problem before and if it was Windows there would probably be memory dumps occurring or at the least, panic events in the log, which there haven't been. Now having looked a bit around on the web, I think it's not a driver issue at all. It's probably the hardware itself.

I found a thread on AnandTech where some guy had a similar issue. He had an nForce board which worked great for years, then he upgraded to an X2. Suddenly he started getting random reboots with no BSOD or memory dumps either. Hmmm, sounds familiar. He did stress testing, memory tests, fan upgrades for possible heat issues etc. and found nothing. He even RMA'd the new X2 CPU for another and the reboots continued. Finally, he started pulling off USB devices. Bingo. When he removed some XBox wireless device the reboots stopped. He did some further digging and it appears some nForce 939 boards have issues with X2's and USB 2.0. He put in a PCI USB 2.0 card, hung his miscellaneous USB stuff off that, switched the USB board support from 2.0 to 1.1 in the BIOS and hasn't had a problem since.

I'm currently snow boarding in Park City, UT (18" of fresh powder last night! ), but will be home late tomorrow so I'm going to do the same thing he did. I already have a USB 2.0 card (it's a Firewire hybrid) so I'll just plug my stuff into that and turn off USB 2.0 on the motherboard as well. I do have some oddball USB stuff connected as the server shares devices through a KVM switch. The first thing I'll probably do is completely isolate the server and unplug all USB stuff from it. If it stops rebooting then I'll know it was something on the USB bus giving the board with an X2 fits. The server should probably be off by itself anyway, but I've never had problems with this configuration in a year.

I'll post back in a couple days with the results. Thanks for the ideas!

[Motofreak75]

if its rebooting after this upgrade with no warning, I would suspect a power supply, if you have a spare, I would throw it in and see, make sure its above 380watts. and it you have a more harddrives then larger PS is needed.


If it was an issues with blue screen or the logs were telling you something then if could be a main board/processor/ram but completely random reboots with no logging is a power issues and dual cores take a bit more watts to run.

I'm not telling you its a power supply 100% but many times I have seen this and nine times out of ten its the power supply.

[phelme]

Originally Posted by Motofreak75

if its rebooting after this upgrade with no warning, I would suspect a power supply, if you have a spare, I would throw it in and see, make sure its above 380watts. and it you have a more harddrives then larger PS is needed.


If it was an issues with blue screen or the logs were telling you something then if could be a main board/processor/ram but completely random reboots with no logging is a power issues and dual cores take a bit more watts to run.

I'm not telling you its a power supply 100% but many times I have seen this and nine times out of ten its the power supply.

Since I was going from a 45 watt CPU to a 65 watt and that's not a huge bump, it was not at the top of my list to check, but I will! Thanks!

[edbmdave]

I just rebuilt my PC going from an AMD 3000 to a AMD 64 X2 4800 (939). I have been having regular lockups and reboots. I have swapped power supply, memory, motherboard (I had two identical) CPU Cooler, reload the OS, reinstall Sage from scratch. Removed all my encoder cards, and slowly checked each one. After changing XPs recovery options to stop and capture dumps, I am getting STOP 7Es. What I have noticed is that SageTranscoder is always running when it hits. I wonder if the transcoder trys to make use of some instruction set that doesn't exist in the AMD X2 (939)s??

I am going to try and swap CPU today I hvae another X2 64 (3800) that I am going to swap out and see what happens. If anyone has any ideas...

[IVB]

Same memory brand each time?

Run a memtest on it, see what that shows. I had an issue with incompatible memory with one of my mobo's, but it didn't reveal itself until I had a lot of stuff loaded, after which things went hinky.

[krasev]

I would check HAL for this processor on your computer....
most probably you had "ACPI Uniprocessor" (you check it in Hardware\Devices\Computer) and looks like you have wrong HAL... most likely non-ACPI some "MPS Multiprocessor"
Boot from WinXP-CD or CMDcon (if installed on a computer) and when you see message something like "F6 for third party drivers..." hit F5... you will be prompt for choosing HAL... hit up or down key and I think for your processor and mobo choose "ACPI Mulriprocessor" then if booting from CD go to Repair with console and log on your OS then exit and rebot... if booting from Hard Drive CMDcons make shure your logon to your OS... then exit and rebot.
Hopefully Windows starts check your HAL (you check it in Hardware\Devices\Computer).
Good luck!

[edbmdave]

HAL is already ACPI Multiprocessor, and I ran MEMTEST86 for 12 hours on the memory, no errors.

Both good thoughts? Others? After superbowl I am going to swap out the processor.

[phelme]

Originally Posted by edbmdave

I just rebuilt my PC going from an AMD 3000 to a AMD 64 X2 4800 (939). I have been having regular lockups and reboots. ... If anyone has any ideas...

wow. you're having it worse than me. at least mine runs.

just an update from me and my issue with occasional reboots, I've been running now for about 30 hours with no reboot. Yesterday I unplugged all my USB devices from the motherboard USB ports and am now using only the PCI USB card. I also turned off USB 2.0 in the BIOS for the onboard ports. If it runs OK for another 48 hours, I'll mark this problem solved.

[blade]

I had a dual core Opteron that ran great for a few months then I started having random crashes and reboots. The cpu would run prime95 on both cores at once for hours on end without a hitch. I finally RMA'd the cpu and haven't had a single issue since installing the replacement cpu.

[MrD]

Dual Core optimizer installed? Disable cool n quiet? That MB may have power plane problems too. Spontaneous reboots signal power or hardware faults. Also check to see if you have any swollen capacitors on that board. That is a sure sign of trouble.

[JonTom]

I've never had much success w/ ECS mobos, but it's been a while.

I had the same problem with my X2-4000 when I first installed it. But it went away after the first couple of days and hasn't come back (knock on wood). I don't know what I did that fixed it, unfortunately, but there was lots of updating on all systems at the time as it was a virgin install.

[edbmdave]

This is almost funny. I bought a MSI K8N Neo2 and when I plugged in the fan on the CPU, I looked in the normal spot for a connector. (right near the power supply on top edge of the board.) Found it and plugged it in.

For some reason this connector was for the system fan, not the CPU. Which is located in a spot slightly different that any other motherboard I have used. I moved the connector and how have 6 days of uninterrupted service. YEAH!

[phelme]

I haven't had any issues with my ECS and the X2 CPU since I moved off the motherboard's USB, so I guess the AnandTechies were right and that's where the problem was.

hopefully it'll last another year or so and then I can look into upgrading the entire server into a quieter case. I'll probably go for some tiny Shuttle-like thing and put all the drives in an outboard enclosure. though two of my client/network encoders are in nMediaPC cases and they are very quiet.