I think I killed it

[Vexorcist]

Time to unlurk now and see if anyone knows what went wrong I guess. So I built a new system after my laptop died, and amongst the various bits and pieces I've ended up with:

MSI Z77 MPower (I just now realized the pun in mpower...)
Intel i7 3770K
Corsair... *ahem*... 'CMX8GX3M2A2000C9' 2x4GB DDR3 RAM

The CPU is incidental in this AFAIK but it seemed potentially relevant.

I read when installing it that the RAM would generally like to start out furthest from the CPU and skip rows. The MSI instruction book shows what appear to be DIMM 4 and 2 populated in the case that there's only two sticks of RAM. I say 'appear' because it's listed 1-2-3-4 on the MB itself and in the book, with 1 being nearer the CPU, but that may not be true (I'll get to this).

When I started everything up, the BIOS showed the RAM at being only slightly above 1k MHz (I forget the exact number now but less than the 1.333 MHz 'SPD' thing on Corsair's page). I naturally wanted to get it to 2k MHz but I figured I could leave it alone for the time being and get things stable first, which actually took about 2 days simply to dig up drivers for every conceivable part and the smaller pieces thereof.

Last night I figured I'd finally got the last driver, windows seemed to have quit annoying me with BSODs for a bit, and I left the PC on copying some backed up files from the network. A few hours later, it was still running, but the screen was black (still on) and though I suspected it happened because I forgot to kill that 'put hard drives to sleep' option, it would not wake up or otherwise respond so I turned it off and went to bed.

When I got up today and tried to start it up, windows had a new BSOD (3B, 7E, 7F, something about CI.dll) every time I attempted to start it, or repair it, even from the CD. Even my random ubuntu live-usb faceplanted and complained about corrupted LZMA data. Looking at google results suggested memory issues were the common factor. Some people had said they ran memtest86 for hours and come up with nothing, but removing some/most of their RAM still fixed it.

As memory had also been a common theme in previous BSODs (albeit broken up more and with driver issues also being possible factors some of the time) I finally just took out the RAM in DIMM 2.

I think.

Windows managed to boot from the CD and run an uneventful check for errors, so I rebooted (and had to select the HDD manually to stop from going to the BIOS over and over and over... not sure what's up with that, but okay) and windows managed to start up. Things seem to be working fine, except for bewing down to 4GB of RAM - what MSI ControlCenter calls 'DDR3-1066 (533MHz)' - but I figured the other stick was just demented.

However, MSI CC also says that slot is 'Dimm 2'. According to the little printed text on the MB, that slot is 4, not 2. And in either case I'm not sure how the slot at the end of the row could possibly be 2, but it's their own control centre. Is it just wrong? Could the slots actually be really weirdly ordered, and that other stick of RAM was just screwing things up because it was out of place?

I'm sort of hesitant to just start moving things around to see what happens, in case I actually cause damage to something instead of finding out what the problem is. If I started it up with both sticks of RAM in the places they've always been, but swapped around, would anything happen to the (currently) decent one if the slots are really weirdly arranged?

 

[alxcsb]

Full specs first, please.

 

[airdeano]

welcome to the OC3D forums.

starting from the CPU moving outward, the first slot is DIMM_1, then DIMM_2,
then DIMM_3 and lastly, DIMM_4

in the above photo, the RAM is placed in the 2 & 4 slots.

Serial Presence Detect (SPD) RAM speed in BIOS is usually set to 1333 by
default. usually determined in the POST. which is normal.

to achive the tested speeds of the RAM, youll have to either use the XMP
file (embedded in RAM) or a manual setup within BIOS.
Speed - 2000
Voltage: 1.65v
Timings: 9-10-9-27

as for the BSOD errors:
3B is usually a voltage supply issue to the CPU usually too little
7E is usually a USB protocol interupt error (timeout)
7F is usually a graphics error on a timeout communication between a GPU sys file

for grins, id reset the BIOS, verify it is the latest BIOS available, when in the
BIOS press F6 Load optimized defaults, reboot and then boot with one
module in slot 1, then slot 2, etc.. then do the same for the second stick...
if no errors, verify the modules are firmly installed and the end clasps are
engauged.

 

[Vexorcist]

The unidentified hard drives are from older stuff:

MSI Z77 MPOWER Motherboard

Intel Core i7 3770K w/ Corsair Hydro Series H100i CPU Cooler

Gigabyte GeForce GTX 670 OC 4GB (as much memory as everything else, now )

Corsair CMX8GX3M2A2000C9 8GB (2x4GB) DDR3 (one is removed currently as per previous post)

Corsair 'Force' Series GS 128GB SSD (win7 'ultimate' 64 bit is here)

2x 750GB hard drive* & 1x 320GB hard drive* (storage & miscellaneous programs I don't want on the SSD)

Pioneer DVR-220LBKS Black 24x DVDRW OEM

Corsair AX860 Platinum Power Supply (via CyberPower Value GP 1500VA Line Interactive UPS, if it matters)

Coolermaster HAF XB Case

*ST9750420AS ATA Device x2 and WDC WD3200BPVT-22JJ5T0 ATA Device respectively, according to device manager. I recall seeing Western Digital on the little one, hence that part of its name, but not sure of the others except they're out of my dead alienware mx17r3.



OS is updated except for language files, and drivers are (I think) all up to date. It's not had another BSOD since removing that bit of RAM so far, which is probably almost a record by now.

Just watch it explode now that I've said that though...

 

[airdeano]

yeah, the BIOS is not updatable through windows/microsoft update..

thats a lotta drives..
sure you got them plugged into the corresponding SATA port?
the 128 SSD should be plugged into SATA 1 or 2 ports with a SATA 6gb
cable (usually black with a white inlay)
the 2x 750 and 1x 320 and DVD should be plugged into SATA 3-6 ports
with any black SATA cable

 

[Vexorcist]

welcome to the OC3D forums.

starting from the CPU moving outward, the first slot is DIMM_1, then DIMM_2,
then DIMM_3 and lastly, DIMM_4

in the above photo, the RAM is placed in the 2 & 4 slots.

Serial Presence Detect (SPD) RAM speed in BIOS is usually set to 1333 by
default. usually determined in the POST. which is normal.

OH! The photo just reminded me, I initially had it set up with the integrated graphics connected as well as the nvidia card, but something somewhere would short and prevent it from starting (I believe related to the case's audio cable having a spare 'AC 97' plug which is now covered in insulating tape so it can't touch the case wall, but I didn't know that at the time) so I took it all out and started with CPU/MB, CPU/MB/RAM, CPU/MB/RAM/I-GPU and so on, solving the problem accidentally along the way. But as nothing seemed immediately worse for lacking the inbuilt graphics (as long as the nvidia card was plugged in) I left that disconnected from the PSU after testing it that way. Which I see it's not plugged in in the picture either.

I'd thought integrated graphics aren't necessary if you avoid using them (on past machines I've even had to disable it to make my AMD GPU's updates work properly, which never made any discernible difference) but would anything else draw power from its plug? The manual just says 'this connector is used to provide power to the graphics card', and leaves it at that. Hopefully I've not starved the RAM of power by making it rely on the 8-pin and 24 pin plugs.

for grins, id reset the BIOS, verify it is the latest BIOS available, when in the
BIOS press F6 Load optimized defaults, reboot and then boot with one
module in slot 1, then slot 2, etc.. then do the same for the second stick...
if no errors, verify the modules are firmly installed and the end clasps are
engauged.

Assuming MSI's LiveUpdate can be trusted to know what the BIOS is, it matches what's shown on this page. I was a bit confused at first because LU calls it 'H.70' but I see now this is in the file name of their latest version.

The picture shows the same way I had my RAM as well. The one closest to the POV in the photo is where my current stick of RAM is, which that control centre refers to as Dimm 2, so I guess the control centre might just be having an off day or something, idk.

I'll try those steps and see what happens.

thats a lotta drives..
sure you got them plugged into the corresponding SATA port?
the 128 SSD should be plugged into SATA 1 or 2 ports with a SATA 6gb
cable (usually black with a white inlay)
the 2x 750 and 1x 320 and DVD should be plugged into SATA 3-6 ports
with any black SATA cable

the SSD is in SATA 1, 2 is empty, and the 3 other hard drives + DVD drive are in the 3-6. The cables I have are mostly white though, which I think came with the board, and one yellow one for the DVD because I had it lying about.

Update:
Okay so: I cleared CMOS with their fancy button, which I think just reset my boot order, and placed the RAM in DIMM 1 & 2. The former 'good' one from DIMM 4 is now in 1, and the other is in 2 (as in closest and next-closest to the CPU, respectively).

MSI ControlCenter says the same thing in 'speed' (DDR3-1066 533MHz) but tells me they are in Dimm (sic) 3 & 4. :mellow: I'm vaguely surprised it didn't turn out to be using 'Dimm 2' to refer to 'whatever RAM I notice first'.

On the plus side, it did boot without any of the errors, and I killed one of those twist-ties that cables come wrapped in, which might've been causing a short I think. Also a noisy fan rattled itself quiet again.

UPDATE: another '3B' BSOD, saying something about 'system service exception'. I got into bluescreenview in safemode(networking) and found that this and one of the previous ones it still had (some were missing, even though one was from last night) were being blamed on 'ntoskrnl.exe'. Which doesn't look encouraging. Oh, and that fan is back to it's normal rattling self, too.

031013-7332-01.dmp 3/10/2013 4:49:17 PM SYSTEM_SERVICE_EXCEPTION 0x0000003b 00000000`c0000005 fffff800`03144e7f fffff880`09910b50 00000000`00000000 ntoskrnl.exe ntoskrnl.exe+75c40 NT Kernel & System Microsoft® Windows® Operating System Microsoft Corporation 6.1.7601.18044 (win7sp1_gdr.130104-1431) x64 ntoskrnl.exe+75c40 C:\Windows\Minidump\031013-7332-01.dmp 8 15 7601 292,528

031013-8642-01.dmp 3/10/2013 11:00:46 AM SYSTEM_THREAD_EXCEPTION_NOT_HANDLED 0x1000007e ffffffff`c0000005 fffff800`030c62bb fffff880`035327c8 fffff880`03532020 ntoskrnl.exe ntoskrnl.exe+6a2bb NT Kernel & System Microsoft® Windows® Operating System Microsoft Corporation 6.1.7601.18044 (win7sp1_gdr.130104-1431) x64 ntoskrnl.exe+6a2bb C:\Windows\Minidump\031013-8642-01.dmp 8 15 7601 292,432

Now google tends to suggest a hard disk error, but if I run chkdsk again that'd be the third time and it found nothing previously, so you may be right about the CPU power. There's no chance of windows actually doing that to itself, right? Like trying to 'save power'? That would be so annoying and yet unsurprising.

 

[airdeano]

move DIMM_2 to DIMM_3 to take advantage of DDR.
right now id worry about stability than what the memory reports as for now.
when installing the OS forthe first time, you only had the SSD (assumming this is
the OS drive) connected?

 

[Vexorcist]

The OS is on the SSD, I unplugged the others to install it as I'd heard of it causing issues, then connected them afterwards. I do keep some programs on other drives though if I don't need them much, like a media player.

I removed the second RAM module a while ago and moved the more stable one to DIMM 2 (which the command centre still insists is 4) then left the computer running memtest86+ via usb. Unfortunately I went to sleep as it was approaching 2 & 1/2 hours or so. It had no errors by then and was coming up on a 4th pass, but later I was barely awake and the system rebooted, so I guess something failed eventually even with this one. IDK if it was necessarily the RAM's fault or what though.

Something else I noticed by chance, the 'Click BIOS 2' when I look at DRAM 'strongly suggests' not setting it over 1.65v, and it's 'auto' setting is awkwardly sitting in the midst of a list of selectable settings, just below 1.5055 and above 1.4980, while my RAM is 'tested' to 1.65 and the corsair site says 'SPD Voltage: 1.5' so I hope that doesn't imply the mb is intentionally starving it or anything. Can it reduce that like the CPU or is RAM constant? I've never seen that auto thing move, but it may just be really weirdly positioned, I guess.

Also, XMP is disabled by default (even when they were in 2 and 4, or 4 and 2 depending whether I ask the written instructions or the software), which is >apparently< a thing this RAM does/can do. :huh:

Are default settings generally trustworthy for stability?

 

[airdeano]

for now, yes.. then once stable then move to efficiency. this will weed out any
other problems that might creep into the overclock or general operations.

 

[mrapoc]

Take out anything unnecessary for now and run it with the bare essentials. This way we can eliminate or at least start to work out what is causing it.

So only the SSD with windows installed, like 1 stick of ram (if it still crashes swap ram stick/slot) no other unneeded stuff and see how it goes.

Run Seatools (from seagate) to determine a faulty drive, chkdsk will only do so much to repair errors, it won't necessarily inform you of a dodgy drive

 

[Vexorcist]

Well a few things since last night:

-I set up memtest86+ again as I went back to bed, with only the 'good' RAM, and it rebooted much sooner than last time but I still didn't see any errors first.

-I swapped it for the 'bad' RAM and have confirmed that it is, in fact, bad. Failed to load windows, went into memtest, 23 errors in mere seconds. Didn't make it reboot though. :mellow:

-The good/better RAM has now run from DIMMs 1-2-3-4 without noticeable difference. Of course, the MSI software still insists they're really 3-4-1-2. I don't know what to make of that.

So that's something I guess. Right now I'm running it with only the bootable RAM stick again, as I want to figure out whether there are still any problems. The system itself has never crashed for me with just this one, but memtest eventually reboots if I leave it going, even if I've not seen errors.

Could there be an error in RAM severe enough to screw up memtest if ever it finds it? Or can I just assume those were unrelated?


In other news, sometimes after doing stuff with the case open (e.g. switching out RAM) I still get a short that prevents the system from starting. Either that or I'm misinterpreting something else (press pwr, lights & fans twitch, everything stops in a fraction of a second). What's weird is that it seems to be related to putting the top panel on the case at the moment (HAF XB is basically a box) and I can't see anything that looks like it should be bothered by that.

I've also put it on running in the past which never caused things to fail, this only ever happens when first starting up. I guess I'll just try to move/insulate anything that even goes near the roof... though I'm starting to suspect it's some weird thing about electricity that I don't fully understand.


The hard drive I suspect is okay, but I'll look for seatools and see what it says, should give me something to do while I wait to see how this RAM fares by itself for a longer time.

 

[mrapoc]

Memtest shouldnt reboot its blue screening or something

Rma your ram

 

[Mark3Max]

Welcome to the forums Vexorcist,
Sounds like you've really had your problems with this build.
Hopefully you don't have any problems getting an RMA on the ram that's bad.
I would suggest you get 2 new sticks of ram just to make sure that all ram is good.
On the case , check and double check all your wires for exposed or cut sections and make sure your not pinching any when you put the top panel on.
Check all your connections and make sure it's all cabled right , it sounds like you have a short somewhere. Good luck , I hope you get it sorted out.

 

[Vexorcist]

Well seatools' basic tests (there was a warning about wiping things out if I tried to access the advanced ones...) passed the SSD and then the other drives as well.

Memtest shouldnt reboot its blue screening or something

Rma your ram

I'm really confused by that result TBH. With windows running I've yet to get a single crash with only that stick of RAM installed, even though it's been on for about 4 hours yesterday and 7 today, and I had Prime95 running for a bit earlier as well. Yet memtest eventually reboots without displaying errors on the RAM.

On the other hand, the other stick won't even let windows boot if it's the only one there, causes BSODs randomly if both are in, and immediately shows errors on memtest, but they're not enough to make memtest reboot the system.

I mean I could send back both, but A) I'd be without RAM on this system, and B) ...

A large percentage of warranty returns are tested to be without fault. If the product is found to be not faulty, a $25 service fee per item may be charged and the non-faulty items will be shipped back to you at your expense. As such we recommend troubleshooting and testing with another system to make sure an item is faulty before returning it.

Depending on whether or not they consider both sticks one 'product' as a result of selling together, and how thoroughly they'd test it (assuming it did actually cause the reboot), I might end up still having wasted money, and getting that stick back anyway. For reference I spent $69 on both of them.

Then there's the chance it would be damaged by the time it got back here. :mellow:

Welcome to the forums Vexorcist,
Sounds like you've really had your problems with this build.
Hopefully you don't have any problems getting an RMA on the ram that's bad.
I would suggest you get 2 new sticks of ram just to make sure that all ram is good.
On the case , check and double check all your wires for exposed or cut sections and make sure your not pinching any when you put the top panel on.
Check all your connections and make sure it's all cabled right , it sounds like you have a short somewhere. Good luck , I hope you get it sorted out.

Hi, yeah it's been a bit... special... compared to last time. I don't really mind too much, I mean I'd be annoyed if I was forever fixing it and never using it, but for now sorting it out has at least been interesting. Sort of like cars in a sense, I never learn anything about them until it breaks.

I've had a fairly extensive look at the wires when I took it all out and reassembled it before, but nothing seems damaged from what I can see. First I found a spare plug on the audio cable that was touching the case, but I moved it away and taped it and it didn't happen any more for a while. Later one of those little twist-ties looked like an end was a bit close to a plug on the cable it was bundling up, so I took that out and put one of my cable ties where it had been and that stopped it that time as well.

For the top panel though, IDK. Very few cables touch or go near it, mostly just the ones to the video card as those power cables are pretty stiff and I didn't want to bend them too much. They don't look cut or frayed though and the spare plugs seem to stick off where they aren't near other things. I've even put it back on now and it's working, but I'm sure it's just waiting for the next time I take the lid off and add/remove something so it can fail again...

 

[Vexorcist]

Okay! So... as it happened, my brother also built a new pc recently, and I got 8GB of older 1333MHz RAM from his previous machine and sent back my irritating ones (I emailed them and they said it does count as one 'item').

My explanation to them:

RAM tested with other parts from invoice* on win7 64 bit, default BIOS settings (includes XMP set disabled). One might be okay, other causes random BSOD when both are installed, or consistent BSOD when attempting to load windows if by itself. All DIMM slots tested and don´t seem to make any difference. System has not crashed with only the better half installed, but eventually reboots without errors being displayed when tested on memtest86+; 4 passes initially. Other module (in serial #) immediately displayed over 20 errors when tested. Running now with 4 modules of older RAM instead and the system is still stable.

*this being the system from this thread.

Their response:

I have tested your memory for a full day in various configurations without any luck confirming your fault. So far the memory has been tested with and without XMP enabled in dual channel and single channel (each module). I am using Memtest86+ 4.2 which you indicated faulted for you immediately with one of the modules, but this does not appear to be happening for me! I have run the modules for at least 2 passes in each configuration, and ran it in dual channel for about 4 passes. I also booted into Windows and ran Prime95 stability test for around an hour without any crashes.

Please let me know if you have any suggestions in terms of what I can do to fault this.

That's very interesting because for me, with both installed it simply would not go a day without crashing - multiple times - and with only the one, the system would BSOD the moment it touched windows. The same windows installation as I have right now, which hasn't had a single BSOD without that particular stick of RAM plugged in.

If it was compatibility of any sort, I'd expect both sticks to have awful results... if it was an issue with the board, I'd expect to notice differences when moving them around, and of course, I'd expect to see problems with all 4 slots filled - and all 4 have been fine with the current RAM ever since.

Is it possible that this system would always try to use the faulty area first, hence that stick not being able to load windows, and showing immediate errors in memtest? And whatever he's using doesn't, so it would take longer to become apparent? :huh: