System build - recommend me the details.

[packerfan]

Interestingly what gimp chugs on photoshop often doesnt. (6000 pixel canvas)

[whismerhill]

My issue turned out to be Avast failing to find a required file then dying, making Windows stick in a 'load Avast, program crashes, load Avast' cycle. Uninstalled Avast... it's all fixed. Ran chkdsk to see if there was any problems with the HDD - I can only guess that Avast updated badly and killed itself, as everything else is fine. Now running ClamWin instead.

wow didn't know that could happen ...
was it win7 x64 or 32 ?
is clamwin any good ? I wonder how an open source antivirus might fare...
too bad there's no "av-comparatives" review of it

[Paradigm Shifter]

Windows 7 x64.

I've had Avast on my laptop do a similar trick a few times before when there was no internet connection, where plugging it into a LAN fixed the 'Windows hanging on black screen with cursor' thing. That wouldn't fix the Windows 7 problem though.

ClamWin is pretty good - it's not permanently 'memory resident' (ie: scanning everything you do all the time) so you need to be more cautious online (although let's face it, running Firefox with Adblock/Flashblock/Noscript should save you from website nasties) and make sure to scan any downloaded files before running/opening. It does have an Outlook plugin for e-mail.

I can't say how good ClamWin is at detecting viruses, as I've not had it or Avast complain about any files yet. It updates regularly, uses very few resources (my Win 7 memory footprint has dropped from 1412MB 'clean boot' to 1118MB) by switching from Avast to ClamWin...

[X-Warrior]

Latest update:


Well, apparently neither my sound card or PaintShop Pro were at fault.

Started again today and bam - memory dump at windows boot. Restart, system restore, bang, BSOD mentioning 'Memory_management'. Retry, bang, BSOD talking about Page_Fault_In_Nonpaged_Area.

Turn off, eat diner. Take out plug from wall, reinsert. Reboot. System restore. Success. Problems solved.

...the temporary problems that is I'm afraid.

Just ran a harddisk check with the Intel toolbox and it found nothing. So the disk should be ok.

Could still be the RAM. But...

Everything put together points me more and more towards something with the power distribution. As if parts don't get enough power all the time. It started shortly after I installed the BluRay player (a few days I suppose), which is attached to the same cable as the harddisks. I think the next thing I'll try is to put the BDplayer on a dedicated power cable, see if that helps...
But surely one cable should be able to support 2 HDDs and a BD player? And the soundcard was attached to something else... Then again, that one was on the same cable as the case fans I think. All in all, perhaps I should be using more cables to spread the load from each PSU port...

[X-Warrior]

Had a look inside my case and apparently the BD player was already on its own private SATA power cable. So I don't suppose it was sucking too much from other components...

I have however replaced the cable that powers the hard drives. And relocated some of the plugs in the PSU. See what happens now...

I shortly considered putting back the sound card already, but quickly figured I'd leave it until a bit later. When things are stable again. At least I can now do so with less fear of breaking things again.

[Gilly]

Your memory voltage is too low, and/or memory timings too tight, and/or maybe you have a defective piece of memory.

Download CPU-Z, take a screenshot of the memory tabs showing your memory settings and SPD pages in CPU-Z. Go look in the BIOS what voltage you have the memory running at.

Also, wouldn't be a bad idea to check it with memtest86.

[Paradigm Shifter]

Gilly's right - sounds like defective memory. Either the RAM is bad (always a possibility) or it's not getting enough volts for the given speed.

However, don't give RAM on an i7 system more than 1.65v... you'll risk frying the memory controller in the CPU. It's not quite as simple as that, but the general rule of 1.65v saves you from any potential pain.

[X-Warrior]

Now just before I read the above two posts, I had downloaded CPU-Z and already checked out my memory timings in the BIOS.



I bought them as being 9-9-9-24, so there's something too tight o'er yonder.

I changed the 'Performance Enhance' in the BIOS from Turbo (default) to Standard, but it doesn't seem to have changed the timings. Guess I need to set them manually.

I think it may have something to do with the memory running at 1066 MHz (something along those lines) instead of the 1333 it is specified at, because the CPU won't do more. In any case, I suppose I should set these timings to 9-9-9-24?

The vendor's site says the memory should run at 1.5 - 1.65V, with 1.65 being default. The downside is that the BIOS doesn't seem to support 1.65, but only goes in steps of 0.1 at a time. Will this cause problems, or is the voltage fine as it is and should I just change the timers?


Edit: now it looks like this:




Forgot to put the performance enhance to turbo again, but am I really going to notice this? Perhaps for stability I should keep it at standard for the time being. Can always put it back to turbo later.

[whismerhill]

I always run memory at least at the higher memory voltage they are certified for
this has no particular side effect since they're built for that purpose except a tiny bit more power consumption...

so unless another component don't like your RAM being too high voltage I would look into getting more juice into them
it can only increase stability which is what you need right now
however don't go above specs either... overvolting does not always brings benefits ...
also relaxing timings slightly too much can't do any harm either ....
but at the same time, at this frequency, your memory should in theory at least support the third JEDEC setting of 8 8 8 22 (which are read from the memory sticks themselves)

Should the problem continue, you might also want to run a bunch of mem tests for several hours to be extra sure http://www.memtest.org/

[Gilly]

the cpu controls the voltage of the RAM on the i7 systems, and therefore bacause of this you can't run too much juice.

Like Whis said, try memtest86 and if that passes, I would format and try again with W7. If their was a sligh problem during installation because of the RAM, just swapping the settings now is too late, the damage may be done.

[X-Warrior]

Reformat? Isn't that rather rigorous when my system is running just fine at the moment?
Of course, it's a matter of a few hours to get everything up and running again, but still, I have activated W7 now, and don't really want to waste it.

What damage are you referring to though? Could there be corrupt files on my system which just happen to go unnoticed most of the time? The Windows directory does have a list of 0 byte files, but they are mostly (error) logs, and a single dll with a huge filename. I'll just list them in case something bad does happen to be among them...
C:WindowsSystem32wfpwfpdiag.etl
C:WindowsdebugPASSWD.log
C:Windowsativpsrm.bin
C:WindowsLogsDPXsetuperr.log
C:WindowsTempDMI2FE5.tmp
C:WindowsSystem32sysprepPantherIEsetuperr.log
C:WindowsTempFXSAPIDebugLogFile
C:WindowsTempFXSTIFFDebugLogFile
C:WindowsTempDMIDF65.tmp
C:WindowsTempDMI4EAB.tmp
C:WindowsPantherUnattendGCsetuperr.log
C:WindowsPanthersetuperr.log
C:WindowsServiceProfilesLocalServiceAppDataLocalMicrosoftWindowsWindowsUpdate.log
C:WindowsSysWOW64configsystemprofileAppDataLocalLowMicrosoftCryptnetUrlCacheContent94308059B57B3142E455B38A6EB92015
C:Windowssetuperr.log

Anything I should worry about in those? Surely empty error logs are a good thing?


Something else now:
I've attempted to reinstall my sound card. And failed horribly. Windows will not detect it no matter what I do. Whenever I remove/insert it the BIOS registers a DMI Pool Data update, so apparently something is seen by the MoBo. Windows may have seen some new hardware for about 3 seconds, and that's the end of it. I've pretty much declared it dead. It's still strange though, as it has worked before, even though it started failing along with Windows on the previous install.
Is it really dead, or is something else possibly going on? Conflict with onboard sound drivers perhaps? Something got corrupted after all that causes the failure? Maybe a Windows reinstall will fix the card. Maybe not though.

On the bright side, these attempts did make me do a lot of reboots, plug pulling, etc. and not a single time did I have a boot failure. Just the previously fine X-Fi won't work anymore...


Hmmm... come to think of it, I suppose I should try the format+reinstall. But I'll do it when the time suits me, there's no hurry at the time being I think. I'll also take a look at memtest, just to be sure.

[X-Warrior]

I've run Memtest tonight. Is it normal that every next run takes twice as long?

Anyway, it's had 3 passes without error. I suppose that'll do. Although I read somewhere that 7 passes is more ideal, that's going to take more than a day if time per pass keeps doubling (starting at about half an hour for the first).


Oh well - Windows reinstall will be planned for either tomorrow evening or Saturday. And I will attempt to take the X-Fi along with it, as something tells me it may well be Windows being nasty about it instead of actual hardware failure.

[whismerhill]

anything in temp folders(like c:temp c:windowstemp and .documents%username%temp ) can be safely deleted if you're not currently in the middle of some software install
some files will happen to be locked by programs since they are currently in use, and that's fine too...

edit: if you can't make that Xfi to work, the best would be to test it out in another system if that's at all possible

[X-Warrior]

Okay, that's it, I'm done fiddling. I am just not going to do anything anymore unless forced to. I will now enjoy my system, which should be stable.


What happened tonight? Well, I reinstalled Windows again, to ensure nothing bad remains from the previous install - back when the memory timers weren't set correctly yet.

At the same time, I gave the X-Fi one more shot. Just in case it was the Windows install that had screwed things up.


The result? Well, the X-Fi is still dead.

So I considered a BIOS update. Who knows, maybe it was the BIOS that did something wrong. The @BIOS guidelines told me to disable Hyperthreading before proceeding. So I did. And then my pc would not boot anymore. It would turn on for a few seconds, and then turn off. Turn on again and turn off. Etc. Didn't reach the BIOS even so there was no fix. Except of course the failsafe CLR_CMOS pins. Which I found under my video card. And boy, it actually worked, the BIOS has been reset. No, I did not forget to put my timers back to 9-9-9-24 :roll:. Interestingly, the BIOS warned me about previous boot errors due to overclocking, hmmm... I suppose that's the standard message when the CMOS has been cleared. On the bright side, at least now I can be sure that no previous meddling in the BIOS has left any remains to bug me.

Anyway, as it stands, I have not updated the BIOS, and I will not update the BIOS. The X-Fi has been put in the box and won't leave it anymore anytime soon.

The only thing that remains is possibly updating the SSD's firmware. It may need a flash before I can trim. But I'm putting that on the long term plans. I just want to enjoy my system now.


@whismerhill: Yeah, I suppose I could put it back in its previous build and see if it still works there. All I'd need to do is install WinXP on there again, on one of the remaining harddisks. Aside from the one harddisk I transferred, it's basically still the complete system. But this one I'll also put on the long term, if I ever feel like it. I do think it's actually dead though. It has worked before in this system. The first install had it up and running. After the crashes it started to fail more and more. I think those were the dying signs. Maybe I hit it in the wrong place. Maybe it got a shortcircuit somewhere. Maybe something else. I do get the impression when pressing it in the PCI slot that I can't go the whole way, leaving it slightly tilted. I actually tried bending the backplate a bit to get more room, but it still didn't seem enough. According to my brother though, it was pushed in well enough, and it's fine just as long as it makes contact.
But like I said, I'm done fiddling now. Any further tweaks will be postponed until a later undefined time.

[X-Warrior]

Interesting... just interesting.


I had the memory dump screen again today when booting Windows.

This time however I saw something unusual before it happened. BIOS memory check at boot: 4GB. That's right, 4. I bought 6, and put 6 in there. Next boot was still 4, but the one after that gave me the full 6GB. I hadn't seen this before because of the big full screen logo which had been on a lot of the time.
It always seems to happen on Monday - everything's off on Sunday, so the power is cut for 40 hours or so. It's almost like one of the memory sticks needs some time to get started. Could be overcome, but it just doesn't sound right, does it?

Apparently it's not the memory timings which were at fault after all...

Perhaps next time (that'll be next week) I see the 4GB memory check pop up I can check the BIOS to see if it's missing one of the sticks in there. If I can't find which is faulty there, I'm afraid I'll have to take them out one by one. One per week, what joy. Hmmm... maybe that could be done faster if I leave a single DIMM out for a longer period...


So... suggestions? Faulty MoBo or memory DIMM? Normal behaviour (I don't think so, but who knows)? RMA?

[packerfan]

Hmm

Try taking out sticks one at a time.

If you have Sticks A B and C, two combos should give you 2GB one should give your 4GB if your mobo is fine and it is the RAM.

Test them in pairs in the same two slots, if you cant manage to get 2GB at all, the slot you aren't using is probably defective, if you do get 2GB on all of them at some point then one of the two slots you are using is defective, move one of the sticks to another slot and try to get 2GB to show up again, if it does it is the stick you didn't move, if it doesn't it is the stick you moved.

Now that you know what the problem is ;) RMA whatever is borked.

Could you leave your PC off for the night and work the next day if you basically wake up and go to work, that could get close to 18 hours off potentially.

[X-Warrior]

Could you leave your PC off for the night and work the next day if you basically wake up and go to work, that could get close to 18 hours off potentially.

That's the normal daily procedure. And it's stable if it's only been off for about 18 hours. It's always stable, except on Mondays, when it's been unplugged for much longer than a day. One day won't cut it. That's the hard part about this thing. Even if I take out the sticks one at a time, it's going to take ages to find the culprit. And if I wait for too long before turning on the pc after I've plugged it in the socked, I may not even have any problems at all. So basically it's remove DIMM, leave unplugged for 40+ hours, boot, check memory. Once per DIMM, and then another few times. Meanwhile, I'd also like to use the thing - now there's two things that don't like eachother. But like I said, maybe I'll find the DIMM missing in the BIOS the first time I notice it's only 4GB again. But I'm thinking too easy aren't I?

Anyway, BIOS is back to default 8-8-8-20 memory settings, as apparently going 9-9-9-24 wasn't the solution after all.

Tomorrow I think I'll call the shop where I bought it. Maybe, just maybe they've seen it before. Never hurts to give it a shot, and to get them knowing that there is probably a hardware issue.

It's still at 1.5V though - any chance 1.6V might solve this recurring very temporary issue?


There's still this strange thing about the sound card (?) as well that's been bugging me for a while... in the PCI devices listing during the boot procedure, there's an 'ACPI controller' which has no assigned numbers. I do believe everything was filled when I first installed the pc. It turned empty when I first disabled the onboard sound controller, but never returned when it was enabled again. I don't know, something fishy may be going on in the BIOS, even after the CMOS reset. It shouldn't have failed after disabling hyperthreading either (though maybe that was the combination with memory timers?).

[whismerhill]

memtest shouldn't miss a faulty module... well I'm not saying it's perfect but still ... it's pretty thorough given enough time...
so somehow I wonder if the problem is either located on the motherboard side OR if there's an external issue

I mean ... unless it's because you use your computer a lot during the week-end that the computer would fail on monday
and in that case you might want to "simulate" heavy usage with 3DMark in loop when you are away working ...

so unless it's that, such a regular periodic event (every monday failure) makes me think of an external perturbation.....

[X-Warrior]

It has been suggested to me on another forum that the memory may need some time to reinitialise after being cut off power for a longer period of time.
Yesterday I shut down at about 23:10. Today I started at about 17:30. No problems, 6GB right from the start. If it is indeed something like this then apparently the memory needs quite some time to 'get lost', so to say. Since not many people would leave a system like this unplugged for prolonged periods of time (I assume), most wouldn't even notice if the same issue was present.

Anyway, calling tech support now. You never know what they come up with...


edit: and they came up with running memtest for every DIMM seperately, at least 5 passes each. I suppose I'll be doing that next then. Of course I knew this was going to be the next step :P. Let's hope 2GB takes a less ridiculous time as 6GB does...
He also suggested possibly the memory controller on the CPU, though he considers it very unlikely. And of course the MoBo slots might be an issue.
I'll just run Memtest with a single DIMM in the slot they're occupying now. That way I can at least narrow it down to a single slot/DIMM combo. If I find an error, I switch the DIMM and check again. Fail = DIMM, Pass = slot. 4 runs to figure it out then I guess...
However, the manual recommends installing any single DIMM in slot 1. Will there be issues if I ignore this and do the above?

Edit 2: never mind the previous question. If a fail occurs when I put everything in slot 1, then the number of tests will be down to 3. That's even more efficient. If the memory is at fault. I guess I'll run it during daytime when I'm at work. Assuming 10 minutes for the first run, and a doubling of the time for every next, it could do 6 passes in 10.5 hours. I should be able to have it do that if I'm at work for 8.5 hours + travel time + other stuff.
And if it's faster, well then it's going to have a lot of passes.
At any rate I should be able to use the system in the evenings this way, and do the tests while it would've been inactive anyway :)

Edit 3: And if the memory test doesn't show anything, there's always the possibility of the PSU not supplying enough power, causing:
Memory slowness.
Sound card failure.
Maybe some other things I haven't actually seen yet.

I may have to install the old PSU some time, just to check what happens. It should be able to handle the system properly, and it's modular as well, so I can probably just switch most cables over. There's something that makes me just not fully trust the thing...

[whismerhill]

It has been suggested to me on another forum that the memory may need some time to reinitialise after being cut off power for a longer period of time.
Yesterday I shut down at about 23:10. Today I started at about 17:30. No problems, 6GB right from the start. If it is indeed something like this then apparently the memory needs quite some time to 'get lost', so to say. Since not many people would leave a system like this unplugged for prolonged periods of time (I assume), most wouldn't even notice if the same issue was present.

do you always unplug the power when the computer is already shut down ? or did I misunderstand something ?