More CPU Woes

[Jason Dunn]

Last month I picked up that AMD 64 X2 4400+ and wrote about my initial problems with it (http://www.digitalmediathoughts.com/forums/viewtopic.php?t=10648), and I thought things were fine. For a while, they were. But in the past week, the Windows Media Center Edition 2005 machine that the CPU is in has given me three blue screens of death, referencing the ultra-vague MACHINE-CHECK-EXCEPTION error. According to the Microsoft KB article (http://support.microsoft.com/?kbid=329284) on it, my problem might be any one of the following:

• System bus errors
• Memory errors that may include parity or Error Correction Code (ECC) problems
• Cache errors in the processor or hardware
• Translation Lookaside Buffers (TLB) errors in the processor
• Other CPU-vendor specific detected hardware problems
• Vendor-specific detected hardware problems

So basically CPU, memory, or motherboard. Memory is easy enough to check - I ran memtest86 (http://www.memtest86.com/) for an hour and my 2 GB of Kingston RAM came up clean. That leaves the CPU and the motherboard, both of which are much more difficult to test. I've scoured a few forums (including AMD's (http://forums.amd.com/lofiversion/index.php/t48192.html)) and the issue is frustratingly hard to nail down. Some people blame the power supply, some blame the CPU itself, some claim the NVIDIA IDE drivers are at fault (I don't have them installed, though I did at one point). At this point, I'm close to simply putting the old 3500+ CPU back in and using the 4400+ CPU down the road. As much as I love assembling customs PCs, there's something to be said for the highly-controlled closed box approach of a big box vendor where all the parts are verified to work together. Any suggestions for me?

[Damion Chaplin]

Without swapping out the mobo or chip, I would say try swapping in a different set of RAM sticks. Memtest86 might say they're OK, but they may still be creating slight incompatibilities with MCE.

If the RAM proves fine, try throttling down your CPU to the 3500+ mark and see if it has any affect (it probably won't).
I think you're using a SFF case, so you may be having heat problems, which will of course produce any of a billion error codes. Try running it for a while with the case open.

Then I'd say try a new motherboard as it's the next-cheapest thing to try.

This exact same setup worked with the 3500+ CPU? I assume you wiped and rebuilt the system after putting in the new chip...

[sub_tex]

While I don't have any advice on your current issue, I will say that I am now way beyond the time when I had patience to deal with that type of stuff.

I order my boards and CPU from Monarch Computer, and have them do the burn in, stress tests, etc for me.

When it arrives in the mail, The CPU and heatsink are already assembled and put in the board, ready to go. I need only add RAM, stick it in the case, and install the OS.

Much much nicer.

I would go that route for your next machine if you can. Having a vendor do all of that stuff before you get it makes your life easy!

And you can still be assembling your own box as you want. You just let them handle the two issues I can't stand the most: CPU and motherboard woes.

[bcre8v2]

Hi Jason,

Sorry to hear about the bad experience.

If I could offer one suggestion, it would be to swap out the primary boot drive and replace it with a spare HD.
You could try re-installing XP (or Media Center) on the new disk and see if the problem migrates with the new OS. If not, then it's an OS / driver issue - possibly caused by an update???

This may actually take less time than some of the other troubleshooting techniques suggested @ MS or AMD.

Regards,

-Steve

[Jason Dunn]

Without swapping out the mobo or chip, I would say try swapping in a different set of RAM sticks. Memtest86 might say they're OK, but they may still be creating slight incompatibilities with MCE.

Thing is, this is the same RAM that's been in the system for over a year. So unless it's some strange interaction between the RAM and this new CPU, I don't think it's the RAM.

I think you're using a SFF case, so you may be having heat problems

System temp is the same with the new CPU as the old CPU, so I don't think that's it either.

Then I'd say try a new motherboard as it's the next-cheapest thing to try.

Since it's a Shuttle, the motherboard comes with the machine, and it's not something I can purchase by itself.

This exact same setup worked with the 3500+ CPU? I assume you wiped and rebuilt the system after putting in the new chip...

Nope, I didn't wipe and re-build, I just swapped out the CPU. CPUs don't require drivers to function within Windows, it's like swapping in a new hard drive, Windows just recognizes it. Although in this instance the "Cool n' Quiet" driver that I needed to fix that MCE problem does make me wonder a bit...

[Jason Eaton]

There is one thing you might need, not sure if you installed the amd x2 optimizer. It essentially adjusts the cpu timing in different modes. Some programs skip the windows api timming. This causes memory errors and such in certain things. I have seen this occur in some games so perhaps, if you haven't installed it, this is also causing you troubles.

http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_871_13118,00.html

It is the first link on the page.

update: sorry just read the problems link that showed you did that.unless you were talking about the cool & quiet patch back then... :)

[Jason Dunn]

Thanks JohnSmith, I'll give it a try...

[dirtboy]

I'm a shuttle owner as well. One thing I have noticed is that the box is so small that heat issues aren't as noticeable as in a big case with a lot of air. My shuttle gets really hot when I am doing something worthwhile, but as soon as I stop it cools down very quickly. In fact, I never would have noticed if I hadn't used a program (SpeedFan works nicely) that logged the heat readings to file. As soon as I exited a program and launched a monitoring program the case had totally recycled its volume of air several times and returned back to normal temps.

Another problem I had with my shuttle was that the motherboard headers for the heat pipe fan went out. I moved the fan to another header and that one went out within a month. I finally gave up on the smart fan control and used a molex adapter and plugged it into a power supply rail. Its loud, but it finally works.

My Mediaportal machine is running a passively cooled FX5200 video card, and that thing totally sucks. Until I get the cash for another card, I have the case off and Walmart desk fan blowing inside.

I know that Cool N' Quiet driver requires you to uninstall previous versions before upgrading and it might require a new version to be installed to work properly. Who knows what kind of crud it leaves hanging around. During the install it might check the processor model and configure itself for that specific processor.

If all else fails, FFR.

[/offtopicsomewhat]

My next major rig will be a full tower sitting next to me at my desk on the floor. The only thing I really need access to is the USB/Firewire ports and the power and reset switches, with an optical drive in the top bay. Cooler Master has one with the ports on the top of the case I might go for.

I really, really want to get a "real" HTPC case, but I think I am going to wait it out until the chip guys bring down their heat. Vista has me scared that the GPU fan is going to running full blast constantly with that "Aero" interface. Just think of the stock cooling those cards have now compared to CPU coolers. If both are running full blast, I don't think the A/C in my house will be able to take it down here in southern MS.

[xycury]

Jason:

thought i'd give it a whirl.

1st Heat- something not to ignore, my personal HTPC had heat issues that would make everything act really weird, BSOD, program crashes, video drops, DVD playback, ect ect. Just check with SpeedFan or other similar programs that can record temps for you.

I finally moved to a ful tower with ample fans.

2nd, just a bad mobo. I have a motherboard that is simply a beast that is extremely picky and ended up RMA'ing it, got a different version and everything worked flawless.

3rd Bios revision, any new versions? support maybe Cool n Quiet from shuttle? i know those are important and if your board is one of the older 939 setups then it may need an upgrade.

4th Reinstall, swap an old HD and see how it goes, i've often reinstalled a few times because of some sound/video change that i couldn't recover from.

My bet, something completely not right with the mainboard. (bios, general hardware) I think your memory is fine, and the cpu shouldn't be at fault, but just with that mainboard.

[Damion Chaplin]

Nope, I didn't wipe and re-build, I just swapped out the CPU. CPUs don't require drivers to function within Windows, it's like swapping in a new hard drive, Windows just recognizes it. Although in this instance the "Cool n' Quiet" driver that I needed to fix that MCE problem does make me wonder a bit...

Yeesh. I know a CPU swap doesn't require a reinstall of the OS, but I always do anyway, just to give me a clean slate to start from.

My rule of thumb is the more fundamental a component is, the more important it becomes to give Windows a clean base. That basically means changing CPU, mobo or C: drive. Everything else, including the video card, is just the walls and roof on top of that foundation.

And, more to the point, you just can't rule it out until you've done it. :?