AMD users - system temps?

[mdmint]

Originally posted by Crowdx42
Heat is more likely to give a hard lock if the heat is system related (not hard drive). Hard drive overheating is something that could cause a reboot like what you are talking about BUT also causes BSOD's a lot of the time (cause the system can't write to the disk, the OS is still in memory). Have you your OS set to reboot on system failure or to show BSOD? You need it set to BSOD so that you can see what error it is giving you.
As far as 7200 drives being "not that fast" , some of these drives do get VERY hot, read the reviews on NewEgg or other online retailers, overheating is a common problem due to bad case ventilation, the bigger drives are MUCH more prone to heating than the smaller ones.
I STILL recommend running Prime95 to check the board,memory and cpu first.
Patrick

I said 7200 drives aren't that fast by today's standards, not that if not properly cooled they wouldn't or don't get hot. As you said the problem isn't the drives, it's poor case ventillation usually the root cause of HD overheating problem.

I totally agree when getting unknown errors, like these sponstaneous reboots being spoken of with no error reported, diagnostics are logical tool to attempt diagnose cause. While I agree overheating heating HDs could be the problem, it seems my comment on power supply being common cause of sponstaneous reboots was totally dismissed. It is a definite possibility of spontanesou unreported reboots. Fine, I've only been working on PCs, Servers, SANS etc for 17yrs or so in field service.

[Crowdx42]

My apologies about your power supply point, did not dismiss it,,, you are VERY correct that it could be a power supply issue.
So far the original poster has done no real trouble shooting and both of our comments have been very valid. I did not mean to sound like I was dismissing any of your advice, I was just adding and restating previous advice to the original poster. Maybe the brand of power supply could also shed some light on the subject? Power supplies are one of those nasty parts of the computer which can have intermittent issues as is described here, only real way to prove or disprove that theory is for the power supplies to be switched (or if he had test equipment which I doubt and also in intermittent failure does not always show up).
Again apologies mdmint, there was no dismissal of your advice intended. I too have near 20 years of experience with trouble shooting both normal desktop/server computers and also CNC computer modules.
Again sorry
Patrick

[edgley]

I have startet having random reboots in the last few weeks.
I am Intel based.

I am begining to think PSU. I think I might swap it with my other system that has a good PSU and see if the problem follows.

[rotaryracer]

Sorry, guys...I've had clients in at work all week and am going blind trying to keep up. I'm hoping this weekend to do some serious troubleshooting on the box.

My power supply is a Enermax Whisper 300W adjustable powering 2 160GB drives and 2 optical drives, plus the mainboard/CPU - no intake/exhaust fans. The PSU is about 8-10 months old (new when I first built the HTPC). Currently I have the fan control turned all the way down for low volume.

I will DL Prime95 on Friday and start stress-testing the crap out of it. Hopefully I should be able to post meaningful results by Saturday.

[Crashless]

Try cranking all your fans up. If you have them down, and you're worried about heat, wouldn't turning them up be the prudent thing to do?

[rotaryracer]

Actually, I'm not sure if it's heat related or not and the problem is intermittent enough to be difficult to narrow down. I'm going to try resolving this issue with number of fans and RPMs staying where they're at - if it proves out cooling is the issue, then I will definitely increase airflow (and unfortunately, noise).

Hopefully running Prime95 will help me determine if it's hardware related or if I need to flush and completely rebuild the software.

[BergoniaC]

I have a 2500+ and it idles at 48C with the system fan and CPU fan at 100%. Under normal load it hits 52C. I never really max out the load, so I can't give that temp. Never had a problem heat or reboot. Also, I have my Ideq inside a wood entainment center with a glass front closure.

It could be a bad driver somewhere. I had a system crash/reboot problem when I was using the built in Video card. I replaced it with a Nvidia 5200 and it hasn't crashed/reboot since installment.

[rotaryracer]

FINALLY had a chance to breathe and went and downloaded Prime95. Installed it, selected 'Stress Testing', and started the default Torture Test. Veeerrrryyyy interesting results:

Test #1 - 1280K blocks - FATAL ERROR. Test wouldn't even run...something about receiving an answer of 0.4999999888 when expecting less than 0.4.

Test #2 - 1024K blocks - FATAL ERROR. Test failed after 3 minutes with one error; reason identical to test #1.

Test #3 - 8K blocks - Passed (so far). Test ran for 1 hr 40 min and was still chugging when I turned it off (had to record some shows). I'm going to restart the test this morning at 20K blocks to pick up where I left off and see if/when it breaks.

Soooooo.....my guess, based on what Prime95's test descriptions are, is that I've got a stick or two of bad memory (or the memory is overheating). The test seems to run fine when I'm putting more stress on the CPU and L1/L2 cache (Test #3), but fails when I'm testing the RAM (Tests #1 and #2). Does that seem to make sense?

The RAM I'm using was a Circuit City cheapie special - (2) Centon PC2700 512MB DDR. I've always used Kingston memory on my PCs and never had any problems like this, so who knows?!?!?

I don't know if I will have a chance today (I'd like to stress-test the HTPC a bit more on Test #3 to see if/where it fails), but I may try to swap out my 1GB of Kingston PC3200 into the HTPC, button it back up, and start Prime95 again. If it passes #1-#3 with flying colors, then CrowdX42, I owe you a beer or three!

Mdmint, regarding power supply, I'm also going to install Motherboard Monitor and see what the voltage ratings do when I run the machine under load with Prime95. Although I wouldn't be too happy about replacing what's supposed to be a decent PSU, I'd do it in a heartbeat to get stability.

THANK YOU to everyone whose offered ideas to this point - it's very much appreciated and hopefully will help me get my Sage box running like it should!

Jason

[rotaryracer]

Update....I let the Torture Test #3 (hard load on the CPU, not much load on the RAM) run for most of the morning. At 4 hours 50 minutes, it croaked as well with a similar error message as outlined above. I'm starting to wonder if it might not necessarily be "bad" memory, but too MUCH memory. Remember I said I was running 1GB with (2) Centon 512MB PC2700 DIMMs? Here's what Kingston recommends for my motherboard (ABIT NF7):

Using 266mHz Modules: Virtually any configuration up to 3GB can be reached using any combination of Kingston's 128, 256, 512MB, and 1GB modules.

Using 333mHz Modules: Supports only two banks at 333MHz using two 256MB (single bank) modules, one 512MB (double bank) module, or one 1GB (double bank) module.

Using 400mHz Modules: Supports only two banks at 400MHz using two 256MB (single bank) modules or one 512MB (double bank) module.

I need to pop my case tomorrow morning and see if these are single or dual-bank modules (the Kingston 512MB PC3200 DIMMs I have in another computer are dual-bank). Since my ABIT manual doesn't go into as much detail as Kingston, it's hard to say if this is actually the problem. If the Kingston info is correct and everything plays nice when I yank out one module, then I've suffered through 8-10 months of relative instability for no reason....



I did also run MBM5 and although the core voltage did fluctuate a bit under load (1.65 at idle down to 1.60-1.62 at full load), nothing dropped dramatically.

I'll keep you all posted!

Jason

[dagar]

Here's what Kingston recommends for my motherboard (ABIT NF7)

That banking issue is due to the NFORCE2 capability of not supporting the 3 DIMM slots with anything but the 200/266Mhz DDR DRAM. 333 and up limits to two. The banks confuses the issue as it is not in reference to the DIMM (double or sinlge sided) rather the slots on the mobo.

IOW, you should be fine.

Are you overclocking? Or did the mobo BIOS incorrectly apply timing parameters? Such as CL2 when the memory (could be) CL2.5 or CL3? Try relaxing the memory timing.
------

On your original posting:

It shouldn't be a problem. Though it sounds possible that your system is suffereing from a screwy bit of SW. It could be that your system is heat-soaked -- multiple components are too hot. Try opening your case and laying it down (so the heat can rise). Stick a 'household' fan over it and move air. It sounds like memory problems though ... esp from your prior post.

[dagar]

Originally posted by insomniac
I only use my athalon system if i need a room heater. Its much more efficient than my wallboard units

You must run Via C3 or Intel Pentium M CPUs. Otherwise it's an asinine thread-hijacking comment

[rotaryracer]

Well, I'm not sure if I lucked out and yanked the "bad" stick of memory or if the Kingston info pertaining to dual-bank/double-sided memory is accurate, but I've just cranked Prime95 for the last 9 hours straight without a problem running with one 512MB DIMM. I'm also running the 'blended' test, which means my CPU has been running at 100% utilization and my memory is about 99% utilized - this is the test I couldn't even get my system to start when I was running 1GB without puking. I'm thinking if it can survive this, hopefully it will survive Sage!

I'm going to button it back up for now and put it back upstairs running with just 512MB. Keep your fingers crossed - if this works, I'm going to be one happy dude!

Jason

[rotaryracer]

Spoke too soon....it actually had run 10 hours, 25 minutes and just died. Still, that's WAY more stable than it was previously. CPU temps were about 59-60C and case temps were about 44-45C (my basement must be a bit cooler than the living room!).

I'm still going to take it upstairs and run it hard....we shall see.....

[dagar]

Oh man. I hate memory errors ... what was I saying?

They can be insidious.

-----------
Edit.

That sux.

Have you relaxed the memory timing? Sometimes the 'autodetect' in BIOS for the timing isn't great. I've noticied it esp on my AN7 boards; the offspring of the NF7. I run 1GB and have noticed that problem too.

[rotaryracer]

So far, so good.....(fingers crossed, knocking on wood)...

I had two shows taping last night while I watched a movie off DVD disc (not streaming from the server). No glitches, hiccups, errors, or problems of any kind! I'm going to try and beat on it all this week, but this was a postive sign.

On another note, I noticed my memory usage is actually down from where it was when I was running 1GB. Typically RAM used would be around 320-370MB....with 512MB total, it's down to 240-250MB used. Am I hitting pagefile more?

Anyway, I'll keep you all posted as the week progresses, but we're definitely looking better!

Jason

[dagar]

Great to hear!

Yes, it's hitting the pagefile more. XP is a bigger RAM hog than other versions. And that is what I see with my systems running 1gb vice 512mb. I haven't bothered to research it but, my 512mb systems use around 80mb of pagefile where my 1gb systems use ~5mb. Though YMMV you are probably ok. Though I have noticed on some 512mb systems that the amount of 'thrashing' increases when you dip below 150mb of 'available' physical memory. Sage isn't a memory hog, from what I've seen. BF1942 is an depending on the mod will require over 400mb of ram to fully load!