My only slowdown right now in setting up the Shuttle for Sage is that the system has only a singe PCI slot - which was occupied by a SCSI card to run the boot drive and CDROM. So, to clear out that slot I had to swap in a new boot drive and optical drive, which means I have to also install Windows onto that new drive.

Unfortunately I my car is awaiting some repairs, otherwise I would have gone out already and picked up another spare mobo and memory which I'd use in the original system while sending the current parts off for repair/replacement.

I ran the memtest on a single dimm for an hour and have just now put back the previously removed module and started another test round. I'll run it the whole day if necessary to see at which point in time it starts to show failures.

Everything always comes at just the right moments in time too. I leave for Vegas at 7am and should be using this time to get together stuff I need to pack as well as practicing some flash and umbrella photo techniques in preparation for some photos I need to take on the weekend. Not to mention putting up an out of office notice on my site, doing some last minute shipments and preparing some auto-responders.

I can't find it now, but I remember reading in the last few weeks about someone having issues with their Asus motherboard, dual channel ram and ACPI mode. They fixed the issue by using an older bios.

If I had just updated to a new BIOS out of the blue I would strongly be leaning toward that type of problem as well. But the system hadn't changed in that regard since I put it together in December (using the original BIOS it came with).

You can't blame a poorly written application on the OS. The difference between the Mac and the Windows (although Windows has admittedly gotten better in recent years) is that when a program crashes with the Mac, it doesn't take the rest of the system with it.

...

Um, clearly the symptom is that accessing the RAM doesn't work properly. But the question is whether that's because a RAM module is faulty, or the motherboard, or maybe even the PSU.

Peter

The biggest issue here is that it's not something that happens instantly nor can be verified right away from a cold state.

The MEMTEST has been going for some hours now (new run) and no errors have been reported yet. The memory modules are quite a bit hotter than when they're running normally in the system - just noting this by touch.

My PVR software is happily running on a different system right now, but I don't have access to all my previously recorded content because that's still siting on SATA drives in the original system. The spare machine has only PATA, so it's going to be handling only the record duties while I'm away and until I can get this memory stuff figured out.

Now I'm in a scramble to finish up all the other crap I need to do before leaving in the morning - which it turns out is more than I was anticipating. Scramble time.

Oh, well there's your problem. The universe knows when one is going away, so breaks as many things as possible to make one's life a living hell.

Tru fax.

Update: Memtest has gone through 16 complete passes and not a single error yet. That's with both modules in place and over 5 hours of run time.

WTF.

I'm planning on leaving it until at least morning, when I have to leave. Should I leave it running all week while I'm away?

Do you have a blow dryer, put it on low heat and aim it at the PSU does it start to fail?

To help reveal any intermittent solder joints on the mobo, give it a few moderate taps with the plastic handle of a screwdriver while running Memtest86 or better still XP.

Sounds like the PSU.

I'll try both the hair dryer on PSU and tapping.

What circumstances are pointing to the PSU right now? I suspect there's more energy required when running Windows, but how about the fact that it was failing memtest this AM and Windows had been failing to boot for at least 6 hours...

PSU failure cause the strangest issues. I had a PC that would run 100% fine but would fail to burn a DVD 1 out of 10 times a few months when buy and it wouldn't burn DVD's anymore. I replaced the burner and it wouldn't burn. I tested both in a different PC and they worked. A friend had a PSU Tester and it showed that my PSU was bad.

The PSU is under very little load just sitting there running memtest.

But with SageTV up and running, recording a program or two, and transcoding or commflagging them, things start to suck a little more power.

I'm betting that eventually something gets hotter than it wants to get (a failing component or whatever), and then it misbehaves until it gets to cool down substantially again.

PSU is the most likely candidate, but it's hard to tell without being able to swap in a known good PSU.

Cheers

Ok, I'm back from Vegas... Unfortunately I came down with a cold the night before leaving and it's still with me now. The trip was still cool (lots of hiking in the "mountains"), but I was only able to actually taste food 3 times during the whole week. The rest of the time I might as well have been eating salted cardboard.

I ran the memtest all afternoon yesterday and hit the PSU with a hairdryer for quite some time on/off. The PSU itself got quite hot and the air blowing out the back of it, even after turning off the dryer, was very warm for some time. No errors reported.

While heating the PSU I also knocked on it and the rest of the case, something that never naturally happens.

The test has continued to run all night and right now it's at over 20 hours without a single error.

If I didn't know any better I would say there don't seem to be any hardware faults.

What's next?

I really hope I'm not going to regret writing this later, but right now everything seems to be okay.

After 30-something hours or running MemTest and multiple sessions with the hair dryer and knocking on the case, PSU and mobo, I decided to reinstall Windows.

I formatted the boot partition, wiping all traces of the past WIndows install. So far I've installed only SageTV, TweakUI and Service Pack 3 along with the drivers for the mobo and capture cards. The system has been up since last night without any signs of issues, recording and playing back for pretty much the entire time.

Fingers crossed.

I knew I'd get hammered for saying things were ok.

Not more than 30 minutes after posting, SageTV quit out. I went to the system and it was still up and running. Trying to start Event Viewer produced an error. Starting a few other apps produced similar errors. I restarted the system via Windows shut down menu and immediately booted into Memtest.

Continuous errors reported. It was climbing into the thousands and the bottom of the screen was solid red and scrolling non-stop.

Turned off the system at the PSU for 10 seconds and restarted. Now it ran for a couple of minutes with only two errors reported. Shut it down the same way for 10 minutes and now there are no errors reported.


WTF. It seems that these memory errors only happen after running Windows (and some software) for a day or so.

What can I do here other than just buying a completely new system (and putting a fresh copy of Windows on it)? I don't even know how I'm going to file warranty claims on any of the hardware I have now because I can't prove definitively that any single piece fails at any specific time.

PSU.

Need to swap the existing PSU with another known-good one and see if that cures it or not. Got another PC in the house somewhere?

I have a crappy spare PSU here that could be used as a freebie loaner, and I have better ones that unfortunately cost money.

But I do plan to be in Toronto next weekend..

Cheers

I agree with Mark and think you should try a differnt PSU. If it isn't the PSU it has to be something overheating. Do you have large fan you could aim into the case when your geting the continuous errors in memtest?

With the capture card running (perhaps even just with its driver loaded), not to mention the other peripherals and their drivers, the system will be using more power than it does with a bare boot into Memtest. I agree with the others, it sounds like your PSU is flaking out under load.

Peter

I can see the load issue, but it's not a heat issue. The system was absolutely cool when I saw the issue and saw he errors in memtest. Much cooler than when I had run the memtest for 30 hours. And certainly much cooler than when I blasted the PSU and memory with the blow dryer.

I have another PSU, I just need to check to see if it will work with this mobo/setup at all.

TUrns out I don't in fact have an extra PSU as I had thought. I'll have to pick up a new one this week sometime.

I've also filled out an RMA a Thermaltake's site - without the advance replacement option.

I'm thinking of buying an OCZ StealthXStream 500w which can be had for $72 minus a $30 mail-in rebate.

For the TT RMA I'll have to ship the PSU to California. I'm wondering if it's even worth it, given the high price of shipping things. The RMA has a 90 day window anyway, so I have plenty of time to decide. And plenty of time to test the new PSU.

Now I'm also wondering if I should buy new RAM and/or mobo while I'm at the store for the PSU - just to save another possible drive out there later.

If you still don't have it sorted by the weekend, then I could pack a spare system in the Scoobie before we head down that way. I'm free on Sunday..

Thanks again for the offer Mark.

The replacement PSU was put into action last night. So far the system is still running. If it doesn't fail, I'll restart it myself in a couple of days and immediately run memtest. Which a few days ago would have shown at least a few errors.

I just did the restart now as I noticed that the SageTV service had quit at around 6pm.

Sure enough, by the time memory test#3 had completed there were already over 60 errors reported.

Seems like it's not the PSU.

Now I'm going to look into returning the mobo and memory at the very least, but I think I should try swapping out the processor as well. Unfortunately I just can't get any conclusive test that proves the faulty component (so I might forward it along to the manufacturer)

I wonder if any of these components can be cross-shipped. At least they're a lot lighter and easier to ship than a PSU.

I just can't wrap my head around the memory failures even if the contents of memory were still intact/corrupt upon restart. The test should be inserting new data, so the previous contents should have no bearing on the testing. This is puzzling because the longer I leave the system off, the less errors will show up (off for even close to 1 minute removes all traces of memory errors and they never come back regardless of how long I run memtest or blast with a hair dryer).

It seems like only running Windows with the other apps I run causes the memory to test bad.

If you have multiple DIMMs, run them one at a time and see if it's a particular one that's having problems.

Have you ruled out the TV Tuner card? I don't recall how it's figured into your other diagnostic tests.

Thanks for the reminder. I just realized that I only did the DIMM isolation testing when using MemTest and not while running Windows for an extended period.

I've now popped out one of the 1GB modules and we'll see how it goes in the next couple of days.

I also already come up with a plan to run Sage along with my drives on my other system if I should need to. I did this while I was away in a more cut-down fashion.

Tony, I haven't ruled anything out, it just didn't cross my mind that the TV tuner card would be the cause of the memory errors though. Especially after rebooting the machine. I ran my primary capture card in my alternate system for the whole week I was away without any issues.

I can conceivably do a card shuffle using that other system to isolate the cards. I'll run through this memory isolation first and then move onto that, likely setting up my other system for primary recording and continuing the testing on my primary system to see if I can get it to a stable state and identify any suspect parts.