System build - recommend me the details.

[X-Warrior]

Yes, I always unplug for the night. After shutting down of course :roll:


Anyway, I've done memtest today.

Started this morning on the first DIMM, went to work, and when I returned it had done 23 passes without error.

I figured it would be faster than I thought, so I put in DIMM #2 already. And it failed on the very first run. And the next. Out of 5 passes, it had one without fail.

Then the turn goes to DIMM #3, which did 5 passes without error.


Brings me back to DIMM #2, just in case. And guess what? 3 passes without error.

Nice...
Maybe it's had power too recently? If it was going to fail again, it would've done so in the first or second pass, but it didn't.


So I called tech support again. Told me Memtest doesn't tell me enough, and if I want to be sure, I'd have to send back at least the memory and the MoBo so they can check out which is at fault. And if I have doubts about anything else, send it along. I guess I could do that if I can't definately find the culprit. He mentioned 5 things: MoBO, RAM, CPU, GPU, PSU. Each of which could be sent back (all in one giant package) for them to check out.

Alternatively, he suggested updating the BIOS. I told him boot failed when I disabled hyperthreading, upon which he suggested that the MoBo is defective for that reason alone if it does that, and in need of replacement.
Anyway, I'll be trying the BIOS update once more. Maybe it was a one-time event. This time I'll be loading the failsafe defaults before disabling HT. And if it fails, well I suppose I might consider inserting a floppy drive to use other means to update.

[X-Warrior]

The second BIOS update of my life (yes, I've done one before some years back):

I loaded failsafe defaults, then disabled hyperthreading and rebooted. Then the weirdness started again. For a while. I pressed the power switch: no result. I pressed the reset button: presto! Or maybe a coincidence I don't know. Maybe it just needed a few reboots to disable HT (1 per CPU?). In any case, HT was succesfully disabled and Windows booted normally. DXDiag confirmed 4 CPUs instead of 8.
Read the manual very very carefully. I closed all programs I could which I deemed unnecessary and could close. Read some more. Then I installed and ran @BIOS. Internet update: nothing found. Typical. I'll do it manually then. Downloaded the latest BIOS (I had read somewhere that I shouldn't skip too many, but I suppose from FB to FG wouldn't do too much harm). Read some more manuals to make sure I wasn't doing anything wrong. I checked the 'load CMOS default after BIOS update' and 'Clear DMI pool data' (that one might be interesting if I ever want to give the X-Fi another shot, though I still think it's knackered).
And I pressed the button... some of the scariest moments of my life, but I knew I had to push through.

And then... well, I'm posting this from my own rig. It was a great relief to see it just boot as it should. Nothing went wrong. The BIOS is now the latest. All that remains is to see what happens next Monday. Problem solved due to the 'improved memory support' which is supposed to have been added in the latest BIOS update? Problem not solved and system parts finding their way back to the retailer?

Time will tell... for now I guess I have the rest of the week to enjoy this thing again ;)

[whismerhill]

at least the memtest indicates that something is faulty which is better than knowing nothing...

memory problems can be pretty random which is why there is several passes otherwise there would be no point in checking several times for faults ...
so when you say "If it was going to fail again, it would've done so in the first or second pass" seems to me like a bit too quick of an assumption...
I understand that the first few times DIMM 2 failed a lot but still ...

the fact that DIMM 1 did no less than 23 passes without problems also tends to make me think that maybe the problem lies within DIMM 2
In my humble opinion, you would need further testing of DIMM 3 and DIMM 2.

however what the tech guy said is true, it's not normal that changing HT settings makes the computer fail, however this might be unrelated to your main problem... That said, if a couple of reset/reboots make the thing work again, I think it should be ok enough (but then that's only my opinion)

good luck and have fun :)

[Gecko]

X-Warrior - I'll throw in my vote as well...

Whenever a PC does weird, funky stuff like yours is, I've found it is almost always:

1) bad memory stick
2) faulty PS

Any power supply can go at anytime (I just had a brand new Antec 750 blow on me for no apparent reason). Sometimes they go out gloriously with a gunshot bang and leave behind a user with a laundry emergency, other times they die slowly. This is when they cause all kinds of crap and it's difficult to diagnose. Bizarre, intermittent failures are the usual suspect. The voltage dips just enough for a split second and the mobo locks things up, or you get memory dumps, etc.

I've found with faulty memory, the failures are usually a bit more consistent and isolated, like under a heavy load.

Memtest only really scratches the surface. Many OEM shops will offer full memory tests for cheap, have you tried this?

I'm also guessing you have a high-quality surge bar/UPS? I've lost two PCs to surges, one was behind a surge bar (not strong enough I guess, which is why I shelled out $$ for current Belkin bars). A surge could also cause damage and similar symptoms.

The mobo and/or BIOS wouldn't be my top suspect.

You may already know all this, just thought I'd throw it in anyway.

Good luck!

[MobsterOO7]

If you don't already have one, invest in something to test your PSU. I bought a little gizmo with lights on it. More hardcore people use a multimeter.

I also recommend getting a UPS, make sure to do plenty of research beforehand though, there are some UPSs that can actually damage a computer. Depending on what you get, they can get pretty expensive. I got mine half price on ebay, IIRC.

[X-Warrior]

I do have a surge protector between the wall and the pc. It's one that was cheap, but did do well in some tests at the time. And hasn't caused any problems in previous builds. The indicator light is still burning, which means that it hasn't had a surge yet. Aside from that - the Dutch power network is among the better ones worldwide I believe. It's pretty stable. Not perfect of course, but still reason enough to not make the standard current the prime suspect.

If the bios update doesn't solve the issue (which I'll know next Monday evening at the earliest), I'll try installing my old 600W Zalman the week after to see if it happens with that thing in as well. At least that PSU has been tried and tested, so I'm quite sure it's a good one. Switching should be a breeze, since they're both modular and the cables seem interchangable.


Another thing the tech support person reminded me of is that I have 2 years of warranty. I.e. plenty of time to do the testing without rushing.


Another interesting thing I noticed today is that now my new BIOS tells me in the startup dialogue that the processor runs at a 24x multiplier, at 3.2GHz. The BIOS menu is still set at 23x (3.06GHz) though, and Windows (DXDiag/CPUZ) tells me no different. Maybe it was already there before and I just didn't notice. Maybe it's stince the BIOS update. Anyway, doesn't seem to be doing any harm, so I'll take it as it is.


As for electric testing devices, I have a cousin who might just have the stuff (and skills) I need... but first the things I can do myself :)


edit: strangely, my keyboard seems to be having some small issues now. It's a Logitech G15. Sometimes the memory won't load on Windows boot, and I need to unplug and reinsert it to get it working. Just the 6 programmable buttons, but still. I wonder whether it's related, or merely a software thing. It happened only once before, and now twice today. Fyi, all 8 USB ports at the back are in use. 5 devices are actually on all the time.

[Gecko]

This one is a bit of a reach but - you might want to make sure you're not starving the PSU by overloading the circuit. My Antec 750 draws a full 10A current at 115V, its power cord is rated for 13A! Not sure about Europe with 230V, but I'd assume your average circuit capacity is less current since you have higher voltage. If you add in room lights, monitor, audio equipment, or another computer, etc. it's easy to get close to the max (15A here).

[X-Warrior]

Underpowered system, hmmm...
Well it does all have to come from one wall socket. Attached are:

- The pc.
- 1 monitor
- 1 speaker set (5.1)
- 1 external harddisk (off)
- 1 printer (off)

And one free slot in the plug bar. Hmmm... maybe the bar doesn't supply enough power through one plug for the build? That would be nearly funny if it were actualy the cause.

The only thing really drawing much power is the pc itself. Oh, and there's also a light bulb above me, but it's one of those saving lamps, shouldn't draw too much.

Yes, I think I'll just try replacing that thing soon as well, if problems remain.


By the way, doesn't the formula go like Vol * Amps = Watts? Therefore, to get the maximum PSU output of 750W, I'd need 750/230 = 3.3A. Theoretically. By the same theory, you Antec would only draw 6.6A max... but I'll just try to see if I can find something along these lines.

[Gecko]

That is the formula, but I'm not sure it's that easy? The power supply converts from AC to DC and then supplies the various rails with the differing voltages. I did take physics at university but I haven't slept in a Holiday Inn Express in a while...so my memory for equations is foggy.

This is what Antec lists (from their site) as input current draw for my TP-750 : 10A @ 115VAC, 5A @ 230VAC.

So, if you are running your power supply from a bar with other stuff on it, that bar should be rated for 15A minimum or higher. i.e. it should have a nice thick 10- or 8-gauge power wire. If not, it may be hampering draw and heating up - is it warm to the touch?

This is all a pretty long shot though. :wink:

I'm still betting on faulty PS or memory.

Good luck, I can appreciate how frustrating it is to build a new system only to have issues with it. Like I stated earlier, I just built my system a few months ago and the PS blew up on me (literally) a couple of weeks ago. Quite disappointing, usually Antec's are bullet-proof.

[X-Warrior]

The power bar is rated at 2300W @ 250V and 10A max. Which is pretty standard. Remember I'm running everything on 230V here in the Netherlands.

Doesn't feel warm to the touch, but I just booted, so give it some time :P

According to this article: http://www.hardwaresecrets.com/article/775/7, the 750HX should draw no more than 881W at full load even.

All in all, the numbers seem to indicate the power bar should be plenty. I doubt any of the other devices will draw >100W or so all by theirselves. Or do they?
Nevertheless, it won't hurt to have a hideously expensive rig like this on a quality power bar, even if it were working perfectly already. So I will be getting me one most likely, with 15A/3500W, or something along those lines.

If it doesn't work, there's always the option of trying the old PSU to check on PSU failure.
Ahhh, so many things to try... and only one attempt per week so I can see whether it actually worked. But at least I can enjoy the system in the meantime. Once Windows is running, everything's as stable as it'll ever be. And boy, is this thing fast.


Whoop, just activated Windows again. I had better not need another reinstall :roll:

[Gilly]

there is no way the power bar or overloaded circuitry would cause this problem, the load he is running would not cause it to excede the ratings

the 10A limit is for safety and doesn't mean that the world will self destruct if it is exceded.

A good extension cord would be 1.5mm^2 (x3) and therefore would allow 15A/16A easily, a ring main (2.5mm TE wired in a ring both to from the consumer unit) in England is rated with a 32A breaker.

There COULD be a PSU problem, but it is more likely to be the RAM, as we had already established that DIMM 2 was dodgy. However.... a couple of related issues, how is the BIOS battery for voltage? lots of the BIOS settings seem to be completely different to settings that software is reporting, which is very odd IMO. And it could be a faulty motherboard with regards to memory slots/ memory power phase.

RAM shouldn't hold charge after power has been removed, so that shouldn't really explain why it works sometimes and not others.

[X-Warrior]

how is the BIOS battery for voltage?

I have no idea actually. Never really thought this far... where do I check? BIOS?

lots of the BIOS settings seem to be completely different to settings that software is reporting

Please clarify? All I can come up with currently is the 3.2GHz(x24) mention at first boot. Which is strange, but likely appeared after the BIOS was flashed. Hence it is probably not related to the initial issues.

And it could be a faulty motherboard with regards to memory slots/ memory power phase.

RAM shouldn't hold charge after power has been removed, so that shouldn't really explain why it works sometimes and not others.

If all else fails, I'll be RMA-ing the MoBo along with the RAM. Before that I do want to ensure that the PSU is working correctly. If it does, there is little else that could cause it.

[X-Warrior]

On a side note:

My pc case is currently standing on the floor. Which comes with its share of static and dust. Anyone know of a small type of 'table' on which I could put it something like 10cm above the floor? Wheels would be a nice addition.

[Gilly]

if your outlet socket is grounded properly static should not play a role at all.

and for the BIOS problems I was meaning how randomly it showed 4GB RAM, not 6GB, the flashing problems and then the odd multipler.

and yeah, best way to check the battery, although the chances of it being the cause is very slim is to take it out and use a multimeter on it. Or just buy a brand new one and slot it in.

I had a machine (K8 MSI shit board) that now has BIOS problems every time it loses mains power (switched off at night) becuase the BIOS battery is shit. It's back in England, so it's not causing me problems anyway!

[whismerhill]

there is no way the power bar or overloaded circuitry would cause this problem, the load he is running would not cause it to excede the ratings

I second that & the rest of his post.

ho and by the way
when estimating max input current take max rated Wattage & divide by efficiency
so 750W with an efficiency of 80% (for example) means a draw of 937W

[X-Warrior]

It's Monday again! And that means glitch time.
Well, seems the BIOS update didn't cut it. It was displaying 4GB again at first. This time I managed to hurry into the BIOS setup though, so Windows did not boot and crash, resulting in system restore, etc.

Anyway, once in the BIOS, I looked up which memory channel was missing: Channel C. (I'm glad I could actually see that.) Now the middle DIMM of the three was the one that failed memtest. Bummer, I thought at first, but then I checked the MoBo manual: Left = channel 0, Right = channel 1, Middle (2) = channel 2. Bingo, so it is that DIMM which failed. Unless I'm making a mistake assuming 0 = A, 1 = B and 2 = C.

For next week I plan to switch the DIMMs at channel 1 and 2. Just to see which one fails this time.
If it's channel B, then the RAM goes RMA and I should be done.
If it's channel C again, then I'll be checking the PSU the week after that. If all is well with the old PSU, then the new PSU will go RMA. If issues remain with the old PSU, then the MoBo is most likely at fault and I'll be RMA-ing both the MoBo and the RAM (just in case, and to let the guys at Alternate figure some things for themselves).

You'll understand that I'm hoping for Channel B to fail next week. All in all, I should have this whole thing sorted within a month from now.

And if none fail next week, then erm... I think I will crash then.

[Paradigm Shifter]

I'd seriously think about RMAing the RAM. But with a channel going missing on an i7 system it could be RAM, CPU or mobo. :( And hit it with all the RAM in with memtest for... I dunno... 10 passes? Usually if RAM is bad it tells you pretty fast, but I've not had memtest give me errors with an i7 system... given the integrated memory controller on the CPU it might manifest differently?

[X-Warrior]

I've done memtest per chip, and the middle DIMM has failed it at first, right at the first pass already. When I retried later, it didn't fail anymore. Maybe because it's had power recently...
It's not utterly bad, it just needs too much time to get started it seems. Very strange, yet it is what I'm experiencing.

That said, it should fail in channel B next week if it is indeed faulty as I suspect. That should be all the proof I need to RMA the memory alone, without any adjoining components (which require a lot more work to get prepared for RMA). If channel C fails again, then the MoBo is more likely to be at fault. Or the CPU, as you mention, but I'm still keeping that as an unlikely.


Toughest bit is that it takes a week for anything to manifest itself again, unless I voluntarily leave everything off for an extra day. And I don't really want to do that. Besides, formally speaking I have 2 years to figure it out.

[X-Warrior]

What joy, it did the trick again today. And it's Tuesday!

Either it's a coincidence, or I had it off for just long enough, or the problem's slowly getting worse.

Of course, I had not switched the DIMMs yet, since I hadn't expected this (but I was ready to stop the boot with the DEL button, so no harm done once again). First thing I did now was turn the thing off, unplug, and switch the DIMMs. The supposedly faulty one is now on channel 1 (B). With luck (or lack of it, whichever you prefer), it'll fail again tomorrow and I'll know more sooner than anticipated.
And no, after the switching it went straight to 6GB. There was a sparkle of hope that the relatively quick unplugging would cause the failure already, but alas, it did not.



And the keyboard programmable buttons didn't work again right at startup. However, simply restarting the profiler software brought it back, so I'm still going with software issue on this one.

[whismerhill]

hum I wonder something....
maybe you could try the "total power off reset technique"

on some systems (mainly old ones but you never know) it could happen that some resident power could be stuck inside some component (don't know how to say it better in english)
and doing the following could resolve it : 0-shut the computer down 1-unplug the power 2 - press the power ON button continuously as if you wanted to power on your computer for 10seconds 3 - release the button, replug the power & continue as usual (for laptops you need to add to step one the removal of the battery)

In your case, maybe, just maybe it could simulate like the computer was off for a longer time than it really was

and to be extra sure it would probably be better to unplug power rather than just turn off the computer power supply switch