Various and Constant BSOD's.
 

[B]Various and Constant BSOD's.[/B]

[b]Relevant Files here:[/b]
Minidump and DxDiag Files available from [url=https://www.dropbox.com/sh/5kl04yh3iyubsyg/AAAivHSeCFYdsssIQTPZ_o9aa?dl=0]Dropbox[/url]

All components are new as detailed below:

[B]CPU:[/B] Intel Core i5-4690K 3.5GHz Quad-Core Processor
[B]Motherboard:[/B] MSI Z97-GAMING 5 ATX LGA1150 Motherboard
[b]CPU Cooler:[/b] Cooler Master Nepton 140XL 122.5 CFM Liquid CPU Cooler
[B]Memory:[/B] Crucial 8GB (2 x 4GB) DDR3-1600 Memory
[b]Video Card:[/b] MSI GeForce GTX 960 2GB Video Card
[B]Storage:[/B] Crucial MX100 512GB 2.5" Solid State DriveMSI GeForce GTX 960 2GB Video Card
[b]Case:[/b] NZXT Phantom 530 (Red) ATX Full Tower Case
[B]Power Supply:[/B] Corsair CX 430W 80+ Bronze Certified Semi-Modular ATX Power Supply
[B]Optical Drive:[/B] Samsung SH-224DB/BEBE DVD/CD Writer
[B]Operating System:[/B] Microsoft Windows 7 Home Premium SP1 (OEM) (64-bit)
[B]Case Fan:[/B] Corsair Air Series Purple 2 pack 52.2 CFM 120mm Fan
[B]Case Fan:[/B] Corsair Air Series Purple 2 pack 52.2 CFM 120mm Fan

Since day 1 I've had problems with BSOD's, My initial investigation's lead down the routes of un-installing Avast and then running memtest86/64. When I run memtest64 in windows it produces errors, I believe this is due to using all available memory and getting into a deadlock state. When I run memtest86 off a USB drive no errors are reported. I've left it running for 10 passes, no errors found. I believe this is the better memtest to run, unless advised differently.

However I bought an identical set of memory and have tried them and still I'm getting random BSODs.

I'v tried each of the four sticks in each of the four slots available, I'm still getting BSOD's In Slot Two and Four together and Slot 1 and 3 together. I've tried with each of 4 sticks in all 4 slots.

However the more I've investigated the less I think it's Memory, I as certain as you can be that it's not memory.

I have re-installed my system 4 times, fully updated Win7 with all patch's, only used manufacturer drivers for the Motherboard.

I haven't Overclocked anything, everything is running with default settings. Although the CPU is running at 3.7 GHz instead of the stock 3.5.

Bios for the motherboard hasn't been updated as it's the current version 1.9.

I have run sfc.exe /scannow and no problems were reported.

I've set "verifier" and no driver issues reported

I've Breadboarded and stripped the whole thing down to only crucial parts required to get Win7 to load and I'm still getting BSOD's.

I haven't connected USB 2/3 Cables or Sound cables. Only cables connected are power to motherboard and JFP1 System Panel connectors and the stock CPU fan cable.

I've examined the motherboard back and front and I haven't noticed anything that looks out of the ordinary.

CPU has been taken out cleaned, pins in the socket appear undamaged. I've set it back up with new Thermal Compound.

Temperatures haven't been a problem.

Please advise how I can go about determining what is the root cause of my problem.

Prime Blend Test produced one error within 5 minutes, I stopped the test and rebooted and started it again, no error seen for the last 30 minutes.

[code][Wed Apr 08 18:31:49 2015]
Self-test 12K passed!
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.[/code]

[B]BSOD'S in this Order:[/B]

[code]SPECIAL_POOL_DETECTED_MEMORY_CORRUPTION
PAGE_FAULT_BEYOND_END_OF_ALLOCATION
BAD_POOL_HEADER
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
IRQL_NOT_LESS_OR_EQUAL
SYSTEM_THREAD_EXCEPTION_NOT_HANDLED
NTFS_FILE_SYSTEM
BAD_POOL_HEADER
SYSTEM_SERVICE_EXCEPTION
SYSTEM_SERVICE_EXCEPTION
BAD_POOL_HEADER
MEMORY_MANAGEMENT
SYSTEM_THREAD_EXCEPTION_NOT_HANDLED
IRQL_NOT_LESS_OR_EQUAL
NTFS_FILE_SYSTEM[/code]

Maybe you need a beefier PSU :devil:

PSU is well within limits, even with the GFX card installed, it isn't installed at the moment though.

So unless the PSU is actually faulty then it's not the PSU.

I'm trying to determine where the fault is, hardware or drivers.

PSU at this time is an unlikely candidate, unless you can provide a more specific reason why you say

[QUOTE=paulunderwood;399664]Maybe you need a beefier PSU :devil:[/QUOTE]

That's not related to Wattage Bias.

[B]Prime Failures and a BSOD[/B]

[Wed Apr 08 19:50:15 2015]
FATAL ERROR: Rounding was 4.460210523e+285, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Self-test 896K passed!
Self-test 32K passed!
FATAL ERROR: Rounding was 2.173430923e+189, expected less than 0.4
Hardware failure detected, consult stress.txt file.

[COLOR="Red"][SIZE="4"]WITH WARNING[/SIZE][/COLOR], you might have to bump up you CPU voltage [B]a little[/B] and retest. :smile:

If small FFT torture works and blend fails, then it is likely a memory (or memory controller) problem. Otherwise, CPU voltage is a reasonable guess.

[QUOTE=badbud65;399667]
FATAL ERROR: Rounding was 4.460210523e+285, expected less than 0.4
FATAL ERROR: Rounding was 2.173430923e+189, expected less than 0.4[/QUOTE]

Um..... 4.46 x 10[SUP]285[/SUP] is notably larger than 0.4.

Crank up the voltage some. Run Prime95 within your capacity to do so and stay under 80C as much as possible. That platform overclocks into the 1.2V - 1.3V range so you can go pretty high and then just bring it down to reasonable levels later. If you STILL fail with voltage that high, then you're almost assuredly looking at bad hardware.

In order, my guess is bad CPU, bad PSU, then bad Motherboard.

Memtest from USB is probably running Linux. It's quite possible that the two operating systems are configuring the hardware differently. I would try disabling the various extensions/options/hardware in the BIOS and see if the problem goes away.

[QUOTE=Mark Rose;399693]Memtest from USB[/QUOTE]

Unless you are running a multi-threaded memtest you are not coming remotely close to stressing the memory bandwidth of a Haswell chip. Prime95 blend test will max out your memory bandwidth. Last I looked, two years ago, multi-threaded memtest had bugs that made it unusable.

[QUOTE=Prime95;399695]Unless you are running a multi-threaded memtest you are not coming remotely close to stressing the memory bandwidth of a Haswell chip. Prime95 blend test will max out your memory bandwidth. Last I looked, two years ago, multi-threaded memtest had bugs that made it unusable.[/QUOTE]

I was on that website a while back to grab a copy of it for some testing. I didn't use the multi-threaded one but I recall seeing something about fixes.

Memtest86+ will find physical problems on the DIMMs not related to memory bandwidth by virtue of testing all but a few KB of the stick whereas anything running in Windows has to be assigned to the specific bad sector and then have that fault become relevant (hence the random Windows BSOD's vs. hours of stability in a stress test or a game or whatever).

If the memory controller is just not stable enough, then George is correct: Prime95 is a better tool. There's just the AVX scare.

Memtest working from stick but not from disk smells fishy, and there is no hardware reason to be so. I know a couple of viruses who do that. Ever played with or installed something called "Protect search"? (don't!)
Any non-normal activity when you open a browser? (like more ads than usual, home page changed, default search engine changed?)

[QUOTE=Prime95;399695]Unless you are running a multi-threaded memtest you are not coming remotely close to stressing the memory bandwidth of a Haswell chip. Prime95 blend test will max out your memory bandwidth. Last I looked, two years ago, multi-threaded memtest had bugs that made it unusable.[/QUOTE]

Good to know!

[QUOTE=Mark Rose;399720]Good to know![/QUOTE]

I wish I'd known that. I had a server that consistently crashed once MSSQL was running on it and hit a certain memory usage. Memtest ran successfully over and over, but I suspect now it's because it wasn't really thrashing it enough. (I finally found the bad module and replaced it).

I hope I remember this info for next time.

Further Information: Installed Ubuntu
 

[QUOTE=LaurV;399716]Memtest working from stick but not from disk smells fishy, and there is no hardware reason to be so. I know a couple of viruses who do that. Ever played with or installed something called "Protect search"? (don't!)
Any non-normal activity when you open a browser? (like more ads than usual, home page changed, default search engine changed?)[/QUOTE]

The system is a totally clean system, nothing installed except Win7 and I still get BSOD's so it's unlikely to be a virus or malware.

[QUOTE=TheMawn;399701]I was on that website a while back to grab a copy of it for some testing. I didn't use the multi-threaded one but I recall seeing something about fixes.

Memtest86+ will find physical problems on the DIMMs not related to memory bandwidth by virtue of testing all but a few KB of the stick whereas anything running in Windows has to be assigned to the specific bad sector and then have that fault become relevant (hence the random Windows BSOD's vs. hours of stability in a stress test or a game or whatever).

If the memory controller is just not stable enough, then George is correct: Prime95 is a better tool. There's just the AVX scare.[/QUOTE]

Prime Fails before the machine fails most times, I just need to know what component is most likely to be the cause of the error.

I have a sneaky suspicion that my CPU could be an RMA'ed component as it's clocked at 3.7 GHz not the stock 3.5 GHz, although it's Boost value is 3.9 GHz so it may just be that.

However would you expect to get a new CPU that was clocking higher than it's stated stock value?

[B]Further Information: [/B]
I installed Ubuntu anyway, no loss in doing that, then I ran Prime for a good 10 hours overnight along with 'stress' and finding that the machine is still up and running after being left on overnight.

However I'm not sure of a few things, even though Prime was running and causing some load when looking at 'top' and CPU heat was hovering around 45 C, indicating it's being loaded as the idle temp is around 20 C. I'm not sure even in this situation the machine is being stressed in the same way, so not having any errors doesn't mean there still isn't an issue with hardware. Although it could mean a Driver issue in Windows.

[B]Also:[/B]
The machine when first turned on seems to function normally for up to 2 hours, then at around the 2 hour mark the BSOD's start, after this when rebooting it will BSOD within 5 to 10 mins constantly until left off for a good half hour.

[QUOTE=TheMawn;399684]Um..... 4.46 x 10[SUP]285[/SUP] is notably larger than 0.4.

Crank up the voltage some. Run Prime95 within your capacity to do so and stay under 80C as much as possible. That platform overclocks into the 1.2V - 1.3V range so you can go pretty high and then just bring it down to reasonable levels later. If you STILL fail with voltage that high, then you're almost assuredly looking at bad hardware.

In order, my guess is bad CPU, bad PSU, then bad Motherboard.[/QUOTE]

As mentioned above changing the voltage might not be a bad idea as I have a sneaky suspicion that my CPU could be an RMA'ed component as it's clocked at 3.7 GHz not the stock 3.5 GHz, although it's Boost value is 3.9 GHz so it may just be that. (I don't know much about that though)

However would you expect to get a new CPU that was clocking higher than it's stated stock value?

[QUOTE=badbud65;399732]As mentioned above changing the voltage might not be a bad idea as I have a sneaky suspicion that my CPU could be an RMA'ed component as it's clocked at 3.7 GHz not the stock 3.5 GHz, although it's Boost value is 3.9 GHz so it may just be that. (I don't know much about that though)

However would you expect to get a new CPU that was clocking higher than it's stated stock value?[/QUOTE]

I don't know the answer to your question, but it seems unlikely that a consumer could change the stock characteristics of a chip. I am certainly curious if there is such an answer.

One case I do remember: on the early Athlons, at least the socketed ones, people would bridge certain cut traces on the CPU base around the actual core so as to defeat multiplier locking. That would not seem to apply here.

[b]NOTE:[/b] The system is BSOD'ing with just the bare bones installed, the GTX960 isn't installed whilst I'm troubleshooting.

In fact it's literally being [URL="http://postimg.org/image/7wyhu1d9n/"]breadboarded[/URL].

I reinstalled Win7 last night, no drivers, no updates installed and ran Prime95 Blend Test and it BSOD at just over 4 hours, no errors in Prime seen.

Voltage to CPU set to 1.2v and Memory set to 1.5v, boost turned off.

My next test will be changing the PSU to a Corsair CP-9020077-UK Builder Series 650W CS650M ATX/EPS Semi-Modular 80 Plus Gold Power Supply Unit, later this evening and test again, if after that I still get a BSOD then I'll replace the Motherboard.

I've already order a replacement CPU which I'll get next week and if none of the above resolves the problem then the CPU will be the only item that hasn't been replaced left to test.

I have my MSI board's CPU voltage at 1.215v. This i7 is stable at 4.2 GHz with 2400MHz RAM :whistle: However, I can not run George's FPU libraries without the Arctic Freezer 7 going crazy before an "auto reboot". You have better cooling. So I think you can up the voltage a little more. I think my board is set to "adaptive" too. HTH

I was just watching an on-line video and the box fell over, after being up for about ~40 days. this gave me a chance to check the CPU voltage: 1.26v :smile:

Fingers Crossed.
 

After three weeks of meticulous testing and replacing parts all to no avail with diagnosing the cause of the BSOD's the culprit of the problems has been determined to be the CPU.

I've recently found out that when I run memtest from a clean startup no errors are shown on the CPU, however after leaving the PC on until it BSOD's and then performing a memtest I get thousands of errors on the CPU. Also if I turn the machine off for 30 minutes and then run a memtest again no errors are shown.

I get a new CPU on Wednesday, and if as assumed that is the cause I'll report back with an update.

This sounds like instability due to heat...

[QUOTE=TheMawn;399986]This sounds like instability due to heat...[/QUOTE]

This. Particularly since OP says "since day 1 I've had problems...." Sounds like the cooler isn't installed properly, or no thermal paste (or 100x too much, as many people do the first time building a machine). If your new CPU works fine, try putting the old CPU back in; chances are that whatever you learn by a 2nd CPU install will carry to the 3rd and the original CPU is fine.

OP says it's a liquid-cooled setup; if the radiator has little heat by the time it BSODs, the cooler isn't doing its job.

[QUOTE=badbud65;399662]Temperatures haven't been a problem.
[/QUOTE]

What is your basis for this particular statement? I had some bad BSOD issues myself that were 100% due to temperature problems. I missed one of the plastic wrappings on my cooler :gah:

In my case, the monitoring software showed fluctuations between 30C - 50C whereas the "weaker" cooler I upgraded from would hang around the high 20's and stay there. Then suddenly I would get a bad crash, due to temperatures spiking into the automatic shutdown region before the monitoring software could even display them.

[QUOTE=VBCurtis;399996]This. Particularly since OP says "since day 1 I've had problems...." Sounds like the cooler isn't installed properly, or no thermal paste (or 100x too much, as many people do the first time building a machine). If you're new CPU works fine, try putting the old CPU back in; chances are that whatever you learn by a 2nd CPU install will carry to the 3rd and the original CPU is fine.

OP says it's a liquid-cooled setup; if the radiator has little heat by the time it BSODs, the cooler isn't doing its job.[/QUOTE]

[QUOTE=TheMawn;400036]What is your basis for this particular statement? I had some bad BSOD issues myself that were 100% due to temperature problems. I missed one of the plastic wrappings on my cooler :gah:

In my case, the monitoring software showed fluctuations between 30C - 50C whereas the "weaker" cooler I upgraded from would hang around the high 20's and stay there. Then suddenly I would get a bad crash, due to temperatures spiking into the automatic shutdown region before the monitoring software could even display them.[/QUOTE]

I have been monitoring the CPU temperature at all times and it maxes out at 60C when performing heavy duty burn tests with Prime, in idle mode it hovers around 20/25C, which considering the ambient temperature is 20/22C is fairly good.

I've reseated the CPU 4 times, cleaned it with Isopropyl alcohol 99% and used Arctic Cooling MX-4 Thermal Compound and I used the pea method, not some other random odd method that are out there.

Observing the graphs created by OpenHardwareMonitor shows that Cores 1 and 2 look reasonably stable, with load, heat etc, whereas Core's 3 and 4 are very spiky and jump around erratically. The graphs haven't been drastically different between re-applying thermal compound. This is why I believe I've whittled down the problem to being the CPU.

This isn't the first PC I've ever built, although it might as well be as I was last in this arena some 15 years ago and many things have changed. One main thing I've got out of doing all this troubleshooting is that I've learnt a tonne of new stuff, which has been fun.

[QUOTE=badbud65;400048]
This isn't the first PC I've ever built, although it might as well be as I was last in this arena some 15 years ago and many things have changed. One main thing I've got out of doing all this troubleshooting is that I've learnt a tonne of new stuff, which has been fun.[/QUOTE]

This is the essence of desktop-building! I feel the same way- while I like things to Just Work, troubleshooting involves discovery about all sorts of hardware/BIOS/control items, which I find fun. If you're correct and you got a born-bad CPU, I'll be interested to hear of that.

Round and Round we go
 

Fitted the new PSU and guess what it BSOD'ed at just over 4 hours once again.

Just out of curiosity I decided to run memtest86 again and this time it started failing after about one hour, this is somewhat annoying as the memory was previously tested and passed. I bought a new pair and all passed in each slot even after leaving it on overnight with all four sticks installed and completing 10 passes.

So my memory has become damaged since installed, or they were already near failure when previously tested and managed to pass.

I know it is time consuming, but try running memtest on each stick individually until you find the bad ones. Maybe the memory is not compatible with the board. I would try running it at 1333MHz, and also flash the BIOS with the latest image. :smile:

[QUOTE=paulunderwood;400227]I know it is time consuming, but try running memtest on each stick individually until you find the bad ones. Maybe the memory is not compatible with the board. I would try running it at 1333MHz, and also flash the BIOS with the latest image. :smile:[/QUOTE]

Erm, again I've stated that I've already done that and more besides.

I wouldn't buy components for a build if they weren't incompatible, I'm not a total n00b!

If you'd have bothered to read the OP, you'd know that's been done.

Sorry but I've posted this problem on many forums, most were helpful at the beginning, I learned a lot. I'm now 4 week's in diagnosing this problem so simple basic suggestions, most you could guarantee I've done.

Again sorry, I'm frustrated with this problem, I'm no where near to giving up though, I WILL discover the cause/s of this problem, eventually.

[QUOTE=badbud65;400230]Again sorry, I'm frustrated with this problem, I'm no where near to giving up though, I WILL discover the cause/s of this problem, eventually.[/QUOTE]

Have you tried bringing another MB into the test set? Based on what you're reporting, this seems (to me at least) the most likely problem.

Just because you "examined the motherboard back and front and I haven't noticed anything that looks out of the ordinary" doesn't mean there's not something wrong with it.

[QUOTE=chalsall;400232]Have you tried bringing another MB into the test set? Based on what you're reporting, this seems (to me at least) the most likely problem.

Just because you "examined the motherboard back and front and I haven't noticed anything that looks out of the ordinary" doesn't mean there's not something wrong with it.[/QUOTE]

I've replaced every single part, even all the leads.

Sorry, but please read the starting thread it's all there. If you want any particular information just ask.

[QUOTE=badbud65;400235]I've replaced every single part, even all the leads.

Sorry, but please read the starting thread it's all there. If you want any particular information just ask.[/QUOTE]

Sorry... In what post did you say you replaced the MB?

[QUOTE=chalsall;400237]Sorry... In what post did you say you replaced the MB?[/QUOTE]

In the opening thread, I've just checked to make sure.

Also, how is your mains power supply? Like the juice coming out of the wall? Is it stable or shaky?



Speaking of power supplies, I recently got a computer working in the office after replacing the power supply not once, but twice, after testing various components.

Yes, after re-reaplacing the memory getting a UPS surge protector is the next on my list.

[QUOTE=badbud65;400239]In the opening thread, I've just checked to make sure.[/QUOTE]

Just reviewed the OP. No mention of a replacement MB.

Please quote the exact paragraph where you say that you tried a different MB.

[QUOTE=chalsall;400242]Just reviewed the OP. No mention of a replacement MB.

Please quote the exact paragraph where you say that you tried a different MB.[/QUOTE]

I've replaced every single part, if I didn't quote that then I apologise. Also using acronyms can be misinterpreted, I thought you ment memory, not motherboard, mobo.

[QUOTE=badbud65;400244]I've replaced every single part, if I didn't quote that then I apologise. Also using acronyms can be misinterpreted, I thought you ment memory, not motherboard, mobo.[/QUOTE]

Have you tried a different motherboard?

Yes or No?

Problem Finally Solved, it was memory after all.
 

Problem finally solved, it was memory after all.

I initially had 4 sticks, one of the pairs failed from day 1 which was then replaced. Both pairs then passed memtest86 for 10 runs. After that I continued to get BSOD's, then after changing all components with new components and still not finding the cause of the problems, I retested the memory again and one stick of each pair now failed. Which is rather infuriating, just goes to show that just because a test passes the first time it doesn't mean it actually passed. As you can't prove that a pass is actually a pass, but one failure is enough to determine a fail.

[QUOTE=chalsall;400248]Have you tried a different motherboard?

Yes or No?[/QUOTE]

Which part of "[B]I've replaced every single part[/B]" doesn't include the motherboard?


"[B]I've replaced every single part[/B]" means all components, which includes the motherboard, the CPU, leads, drives, even keyboard and mouse.

"Ive replaced every single part" rarely means exactly that. In this case it apparently did not include the memory. The questions you took offense to were only trying to figure out what your instance of "every single part" meant.

[QUOTE=owftheevil;400306]"I've replaced every single part" rarely means exactly that. In this case, it apparently did not include the memory. The questions you took offense to were only trying to figure out what your instance of "every single part" meant.[/QUOTE]

It did include Memory, they were the fist components that were changed. I was just extremely unlucky to get 3 sets of memory with 1 stick in each set being faulty. Also with the 2nd and 3rd sets passing memtest, hence why I went round the houses to finally find it was the first thing I'd previously tested that was the cause of the problem. In total, I'm now on to my 4th and 5th set of memory, different brand and different supplier this time, all seems good pressently.

[QUOTE=badbud65;400312]pressently.[/QUOTE]

The word of the thread right here

[QUOTE=badbud65;400289]Problem finally solved, it was memory after all.[/QUOTE]

In this typical not-overclocking-to-the-limits case, my experience is that 80+% of the time memory is the culprit.

Also, this reinforces my firm belief that memtest is poor at catching all memory errors. That is, if it passes memtest and you are still having problems then in my experience 50+% of the time memory is still the culprit.

[QUOTE=Prime95;400325]In this typical not-overclocking-to-the-limits case, my experience is that 80+% of the time memory is the culprit.

Also, this reinforces my firm belief that memtest is poor at catching all memory errors. That is, if it passes memtest and you are still having problems then in my experience 50+% of the time memory is still the culprit.[/QUOTE]

The lengths I've had to go through to eventually end up where I started from, (memtest) is infuriating to say the least.

However, and i've said this before, I've learned tonnes, 💩loads in the process. Which has given me more knowledge about overclocking and increased my confidence .

First step use the auto features if they work then tweak until it fails. I have no reason at all to overclock except to learn how. It'll be fun, I just need a stable system before I experiment and learn more.

stucked bsod

[QUOTE=fireball;432897]stucked bsod[/QUOTE]
Care to elaborate?
What are you trying to say?

i had several bsod's on my computer but i did fixed it now[URL="http://www.deskdecode.com/how-to-fix-stop-0x0000000a-irql_-not_-less_-than_-equal/"]
[/URL]

A little relevant titbit: it is impossible to test RAM in its entirety on a PC that uses that RAM. Some parts are reserved and cannot be tested. So there is always a chance that even though a test passes the RAM is still bad.

Switching sticks around sometimes helps to find these ‘hidden’ bad spots. But you got to have more sticks than channels. Like, if you have two channels and four sticks, try to switch the sticks on the same channels with each other. The exact switching around strategy depends on channel interleave and rank interleave settings...