Reboot on HDD activity (need help)

[robal]

Hi TPU users,
This time, I need help from you

My PC (see spec.) is 100% stable during games, CPU + GPU + RAM stress tests, etc. I can run them for 12 hours without a hiccup.

Still, a simple HDD stress test (or just copying couple of large files) will always cause a reset (no BSOD, just instant reboot) for 100% within two hours.

So far I've tried:
- clean Win7 install on a different HDD, using different cable
- switching HDD from internal (SB600) sata to additional SATA controller on my board (separate from south bridge)
- using different RAM (DDR2 800)
- disabling any "auto tweaks" in BIOS

Nothing helps.
At this point I'm convinced that my mobo fails. (but why both SB600 and G-SATA ?)

My NB and CPU temps are very low.

One could suspect PSU, but since it's super stable at maximum loads, why would it fail at almost idle ?

Did anyone had or heard about such instability ?


Cheers,

[digibucc]

that's weird.

mobo, could be the ram slots. did you try single sticks in different slots?
that could still be it, and ram would (i think) be the most likely culprit for
failing stress tests with different hdds, cables & ports. ( so something to
do with the ram on the mobo, rather than the sticks themselves)

you did more than most would have before asking ... as i said, this is weird.

[mrhuggles]

i had that happened, it was when i was heavily overclocked and it turned out to be that the southbridge was at its limit, my motherboard couldnt drop the southbirdge clock

[Steevo]

I believe my board does the same, increasing my southbridge volts helped my overclock.

[robal]

Originally Posted by digibucc

that's weird.
did you try single sticks in different slots?
that could still be it, and ram would (i think) be the most likely culprit for
failing stress tests with different hdds, cables & ports

Thanks for tip. I've actually set up a testbed with different PSU, different GPU and, as you suggested, one stick of ram in slot that was always unoccupied.
It's running now. We'll see

Originally Posted by mrhuggles

i had that happened, it was when i was heavily overclocked and it turned out to be that the southbridge was at its limit, my motherboard couldnt drop the southbirdge clock

I don't overclock. But SB seems quite hot to touch.
That's why I added some temporary air flow for it for this test run.


Thank for replies so far. I'll report this test run result.

Cheers,

[mrhuggles]

does your silence mean its working fine now? i sure hope so.

[micropage7]

try to set from advance system setting - advanced - startup and recovery then turn off automatically restart
so although there any err the system wont suddenly restart, it just show the warning but not restart

[robal]

Originally Posted by mrhuggles

does your silence mean its working fine now? i sure hope so.

No
It means that I'm taking testing seriously (has to run for 12 hours before I accept as 'stable' and make any hardware change). So it takes days, unfortunately :/

The test run with different PSU, different GPU, relocated RAM and added SB airflow was proven stable. Yay !

Since then, I've done another test run, with PSU and GPU reverted to original. That was stable too !

Now only two things remain: Relocated RAM and SB cooling.
I'll revert one of these and run tests tonight.

Thanks for your suggestions and support !

Cheers,

[de.das.dude]

just in case,

have you tried running HD tune pro? does it show any errors?

[Beertintedgoggles]

As others have already said and it seems like you've figured out too, it sounds to me like its the southbridge either not having enough volts or overheating. In your case it sounds like overheating.

[robal]

OK, test update:

SB cooling is not the problem. The system is 100% stable without additional SB cooling, and allowing it to hit more than 45 deg.C

That leaves RAM.
I've tested twice, for both sticks (using one at the time) and it's stable.
Digibucc, you were right.
How is it possible that RAM configuration is responsible for HDD activity crash will probably remain a mystery.

I've got a few more configurations to test:
- both stick in dual-channel, but using other 2 slots
- both stick in single channel

Cheers,

[robal]

Another update.
My God... that is so painful...

After DAYS of testing I've concluded that fiddling with RAM sticks and settings can merely reduce the probability of crash, not prevent it.
I've made test runs longer, and made step back to reconfirm findings.

The system is unstable even if it's running on one stick of lower-rated memory.

My current suspicion is in CPU / Mobo now.
I'm running tests on different CPU now.

[Phrosh]

wow, a fresh thread on this topic.

same issue here. bought myself a new cpu. updated bios from f1 to f7. everything works fine: games smile, memtest is happy, too and prime95 is full of glee . nothing is dangerously hot, in fact everything at norm temps.
but when i come to copy files from one to another HDD the system reboots within, let's say one hour, or freezes with just a single repeating sound from my speakers (not from the case, those are disconnected). last thing only happened to me twice (but after another), others several times more...

everything worked fine for about 3 years with my old cpu until then. the only component i changed was the cpu. and for this i updated the BIOS and reinstalled win7 x64.

so it is highly possible that it has got something to do with eiter cpu or MB (more precise: BIOS / settings). the fact that prime95 did not see any errors AND that since the invention DMA, file transfer operations between hardware components are mainly non-cpu controlled, that only leaves the MB / BIOS.

my mainbord is a Gigabyte MA790FX-DS5 (AM2+)
my new cpu is a AMD Phenom II X4 940 and with the newly flashed BIOS the strongest compatible CPU for this MB (you can apply BIOS F8beta instead and use AM3 cpu's... but... meh).

the next things i will do is try to flash the BIOS to the F8beta instead of my current F7final. if that does not work, i'm gonna install windows on a different drive (although you did try this, too) and maybe play with some BIOS settings (i already did this, but obviously with no success).

i also found some people with the same problem but threads dated 2 years back and more... one found out, that these problems don't occur in ubuntu live. i will also check that out. this one might lead me to think that this may be a sofware problem rather than a hardware problem, but after you said you re-installed windows, i'm not so sure anymore...

anyway, if i get any update, i will post it here!

[Phrosh]

ok, i just copied 150 GB from one HDD to another! system remained stable all along.
i object the problem as "solved", after about a week of "trial-and-erroring", anger and dispair...

i updated the BIOS to F8beta but apart from some more OC'ing features available nothing changed.
i just found the system specs on the left hand side, noticing you have the same mainboard as i have and the CPU i first wanted to buy, but finally decided to take the 940, because the 955 is not supported by the latest stable BIOS release (F7). nonetheless, i ended up using the F8 beta BIOS, yay <.<...
if you haven't yet, i strongly recommend you to update to F8beta, becuase otherwise the 955 is not officially supported.

ok, enough of my discoveries of hidden forum features . how i (apparently) fixed it: turned the SATA mode to "Legacy IDE" in the BIOS. yes, you heard right, "legacy", not "native"... it's fuckin' win7 x64, so i have no clue, why i had to do this - this was just a result of trial and error.
maybe someting with IRQs, since the controllers IRQs are handled differently in this mode, dunno.

anyway... this works for me, hope it does for you as well.

PS: hope you forgive me my first 2 posts being double posts

greetings from germany,
Phrosh

edit:
ok, forget this. system crashed again.
but the period of time the system ran stable seems to be greatly improved. this might be a coincidence, though.
still no 100% stable system, still same problem. a new windows set-up is following on my side...

edit2:
stability really improved. copied about 250GB with one crash during the procedure. still one crash too much.
before that, the system shot itself after about 10 minutes...

[robal]

Hi Phrosh,

It's so fortunate indeed that you've found the thread.
We've got similar configuration and seem to be suffering with the same glitch.

You have no idea how many times I've announced (to myself) that the problem is fixed, until longer testing proved me wrong.

Right now I'm sure of only one thing.
My system was 100% stable when I replaced CPU with old Athlon X2 5000+
So you're right. It's CPU related.
I'm suspecting that the mobo/NB f#$cks this up...

Note that 5000+ has 1600 Mhz HT. Phenom2 has 2000 Mhz (on this mobo).
I've forced my NB and HT to run on 1600 on my Phenom2 955.
So far it's stable, but I'm not announcing anything until I'm 100% sure.

Cheers,

[Phrosh]

hey!
okay, sitting here in front of a brand new windows installation and got the first crash. was not hard to tell this would happen, but worth a try, especially with an installation with changed BIOS settings.
did not help, so at the moment i'm really out of ideas...

my previous CPU was an Athlon 64 X2 6000+. this lets me hope you might be damn right with the NB / HT attempt you mentioned above. my old CPU's HT was at 1GHz o.O...

i'm not quite sure, but i think my NB and HT run at 1.8 GHz, as far as i remember.
i don't know what to do anyway, so now i'm gonna underclock these two friends, like you did and see what happens.


edit:

about 5h stable, copying data from different HDDs to another with no crash at all.
seems like you got the right solution! high five!

i always liked this mainboard... but underclocking the north bridge to get a system running is not quite what i understand under a "high end board"... i'm pissed because i paid 160€ for this mainboard and damn happy because we finally found a solution at the same time ^^. i went 200 Mhz down, but you even 400. if we're lucky, it still runs when HT/NB are back to normal again - would not be the first time changing something and resetting it back to its previous state solved a problem .

maybe an overheating problem? too stupid gigabyte did not attach a sensor to the chipset, so temperatures could be monitored <.<.
speedfan shows a mysterious "temp3" on the initial tab, but this just seems to be a bug. "temp3" is 92 - 93°C here, but due to the sensor not being attached this value rather shows the socks of my grandma than it shows any chipset temp.
i will be busy installing the dozens of software i need for my "new" system - you could try to replace the heat transfer paste (looked this up ) of the NB. as i read gigabyte used everything but the best there . i got a really bulky cpu cooler, so i would defer to you .

anyway, thanks for sharing the prob and your attempts

edit 2:
no more crashs while copying files. played ut3 with max details and 3dvision - so the north bridge can't really be any more busy like then. system crashed after a while. burnt myself when touching the chipset coolers. so this really looks like a cooling problem to me. gnaargh... i don't want to rip all the stuff out again just to replace some stupid paste... but i think there is no escape ^^'.

[robal]

small update:

I've been running my system for weeks now with 1.6GHz HT and it seems stable.

[TheMailMan78]

Have any of you reset your CMOS?

[robal]

More than once. Tried different BIOS versions too.
It's a very non-trivial issue going on here.

Cheers,

[Phrosh]

for me there is only the choice between two different bios revisions - but i tried both.

bought myself a chipset cooler. i accidentally dropped the good gigabyte MB - this solved the problem for me, because it does not work anymore now.

because it does not make sense to buy an am2+ board today, i'm upgrading to a sandy bridge - after years of amd... i bought the new cpu BECAUSE of this board and now i have to sell it again, because it does not fit into a newer one... how ironic... and expensive... and stupid...

with a broken MB i can't reproduce the problems anymore... anyway, you could get yourself a northbridge cooler and go to 2ghz again, i think. i read the manual of the MB and found out, that the initial bios wasn't even able to have a higher north bridge clock than 1ghz, so i think the MB itself wasn't built to run a higher NB clock than 1ghz. so even 1,6ghz is like "overclocking" for the NB.

[robal]

Small update.
My system has been rock solid stable for this entire time (more than a month).
In the meantime I've upped NB and HT clocks to 1.8 Ghz and that is stable too.

And yes... I think it's a 'legacy' of this board being AM2 board, pushed to AM2+ by BIOS update.

Cheers,

[Phrosh]

glad to hear that you fixed the problem.
but due to the fact that it actually is an overheating problem, the system might still crash under certain conditions - and you're still not running on your 2 GHz NB clock, so you can rather speak of the problem being "avoided" than "solved", imho... but that's gigabytes fault - and maybe a bit ours, because we bought the mainboard

anyway...
pretty sure people with the same issue will appreciate this as well, when they find this thread .

I'm in love with my new sandy now - a graphical EFI BIOS with mouse support definitely kicks ass .