Multiple BSOD's on new machine

[Danny72170]

The link for the download site brings up the following error:

pnErrorResult=3 pnErrorDetail=Database unavailable ==END==

 

[usasma]

I get the same error. Just Google for Prime95 and select a trustworthy site to download it from: prime 95 - Google Search

FWIW - make sure you get the 64 bit version...
Good luck!

 

[Danny72170]

Ok downloaded the program. It found an error in one of the CPU's in about a minute. It's still running though. it mentions a stress.txt file. where will i find that?

 

[Danny72170]

Another question: Could the errors be due to overclocking? I'm not sure if my computer is overclocked (it's running at 3.2MHz) but I assume if the BSOD's are due to overclocking I'll need to talk to the manufacturer about stopping the overclocking.

 

[Danny72170]

Ok the last CPU produced an error about 1 hour 15 minutes after starting the test. Still not sure where to find the stress.txt file mentioned in the test.

 

[usasma]

The stress.txt file is found in the folder that you extracted Prime95 to.
It's a lot of technical info - but basically, the Blend test tests RAM, CPU, and mobo (RAM interface)
Now try the Small FFT's test and the Large FFT's test - and let us know how long each one took to show errors.

More info in this DRAFT of my explanation:

DRAFT Prime95 testing procedure
This is in case you are getting errors in Prime95 and want to nail down the cause

Run HDD diag and MemTest prior to Prime95

Run Blend test
Run Inplace Large FFT's test
Run Small FFT's test

Run CPU Stress Tests
- http://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=19182&lang=eng
- 7Byte : Hot CPU Tester Pro
- 7Byte : BurnIn64
- CPU Stress test - Overclock.net - Overclocking.net
- http://www.fossiltoys.com/cpuload.html
- CPU Stability Test description, System Resources Tune-Up. Downloads List By All Time Popularity | PCWorld | PCWorld
- CPU Stress test - Overclock.net - Overclocking.net
- |MG| CPU Stability Test 6.0 Download
- LinX - A simple Linpack interface - XtremeSystems Forums

Run Video Stress Tests
- FurMark: VGA Stress Test, Graphics Card and GPU Stability Test, Burn-in Test, OpenGL Benchmark and GPU Temperature | oZone3D.Net
- МИРNVIDIA / Утилиты / VMT
- МИРNVIDIA / Утилиты / Artifact Locator

Swap RAM sticks around per Steps 7 through 10 of this link and run Prime95 (in Blend mode) on each stick/slot: Memory Diagnostics
This is because of issues found in this topic: http://www.sevenforums.com/crashes-debugging/124155-blue-screens-again-2.html#post1076674
************************
DISCUSSION:
Small FFT's test CPU cache and very little memory.
Large FFT's access' a lot of memory - but in-place option keeps it limited
Blend is not run in-place, so it cycles the tests through a lot of RAM

All tests stress the CPU
In-place large FFT's and Blend also test CPU/RAM interface (mobo in all but Athlon64 and Intel i-series)
- This would then be testing the memory controller onboard the CPU in the Athlon64 and i-series processors
In-place large FFT's stress CPU more than Blend
MemTest is better at testing RAM

Prime 95 tests 3 things:
- Memory
- CPU/RAM interface (mobo in all but Athlon64 and Intel i-series)
- CPU

FWIW - the distinction between RAM and different types of memory in the system is one that I've made. The documentation that I've read speaks of RAM issues and CPU cache - and doesn't mention the other aspects. BUT, this is just speculation on my part, so we'll have to see how it works for us.
Link to explanation about i-series CPU/RAM interface: Intel Core - Wikipedia, the free encyclopedia

Common features of all Nehalem based processors include an integrated DDR3 memory controller as well as QuickPath Interconnect or PCI Express and Direct Media Interface on the processor replacing the aging quad-pumped Front Side Bus used in all earlier Core processors.

 

[usasma]

Overclocking can be stopped by going into the BIOS and resetting the system to defaults.
If you haven't done it, it's not likely that it's overclocked.

You can also use CPU-Z (free here: CPUID - System & hardware benchmark, monitoring, reporting ) to see if it's overclocked

 

[Danny72170]

Ran the large FFT test and the first CPU got an error in 2 minutes and the last one at 34 minutes. Currently running the small FFT with no errors so far after 3 hours of testing.

 

[Danny72170]

Small FFT test has run 6 hours so far without an error

 

[usasma]

This is a DRAFT of my Prime 95 speech:

DRAFT Prime95 testing procedure
This is in case you are getting errors in Prime95 and want to nail down the cause

Run HDD diag and MemTest prior to Prime95

Run Blend test
Run Inplace Large FFT's test
Run Small FFT's test

Run CPU Stress Tests
- http://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=19182&lang=eng
- 7Byte : Hot CPU Tester Pro
- 7Byte : BurnIn64
- CPU Stress test - Overclock.net - Overclocking.net
- http://www.fossiltoys.com/cpuload.html
- CPU Stability Test description, System Resources Tune-Up. Downloads List By All Time Popularity | PCWorld | PCWorld
- CPU Stress test - Overclock.net - Overclocking.net
- |MG| CPU Stability Test 6.0 Download
- LinX - A simple Linpack interface - XtremeSystems Forums

Run Video Stress Tests
- FurMark: VGA Stress Test, Graphics Card and GPU Stability Test, Burn-in Test, OpenGL Benchmark and GPU Temperature | oZone3D.Net
- МИРNVIDIA / Утилиты / VMT
- МИРNVIDIA / Утилиты / Artifact Locator

Swap RAM sticks around per Steps 7 through 10 of this link and run Prime95 (in Blend mode) on each stick/slot: Memory Diagnostics
This is because of issues found in this topic: http://www.sevenforums.com/crashes-debugging/124155-blue-screens-again-2.html#post1076674
************************
DISCUSSION:
Small FFT's test CPU cache and very little memory.
Large FFT's access' a lot of memory - but in-place option keeps it limited
Blend is not run in-place, so it cycles the tests through a lot of RAM

All tests stress the CPU
In-place large FFT's and Blend also test CPU/RAM interface (mobo in all but Athlon64 and Intel i-series)
- This would then be testing the memory controller onboard the CPU in the Athlon64 and i-series processors
In-place large FFT's stress CPU more than Blend
MemTest is better at testing RAM

Prime 95 tests 3 things:
- Memory
- CPU/RAM interface (mobo in all but Athlon64 and Intel i-series)
- CPU

FWIW - the distinction between RAM and different types of memory in the system is one that I've made. The documentation that I've read speaks of RAM issues and CPU cache - and doesn't mention the other aspects. BUT, this is just speculation on my part, so we'll have to see how it works for us.
Link to explanation about i-series CPU/RAM interface: Intel Core - Wikipedia, the free encyclopedia

Common features of all Nehalem based processors include an integrated DDR3 memory controller as well as QuickPath Interconnect or PCI Express and Direct Media Interface on the processor replacing the aging quad-pumped Front Side Bus used in all earlier Core processors.

Please see that the Large FFT test tests RAM and the CPU/RAM interface
So now's the time to run the additional tests above to test the CPU (in order to rule it out).
Further memory tests will help to rule out the memory - and if those are both ruled out, you're left with the mobo as the source of the issues.

This is not a 100% diagnosis - but rather it's a best guess using the available technology. I suspect that other failing components can also cause issues (I had a 56k modem that caused problems with the RAM after a lightning strike years ago).
I'd suggest stripping the system down to the bare essentials and then running Prime95 again.
Here's a link to my "stripdown" post: Hardware Troubleshooting Via System Stripdown

 

[Danny72170]

Should I be running all the CPU stress tests you listed? Or pick one? Also, I downloaded Furmark but not sure which test to run with it. The website where I downloaded it shows a video card damaged by Furmark so I don't want to do the wrong test.

I got another BSOD tonight, the info is attached.

 

[usasma]

Run the BurnIn test in Furmark and keep an eye on the computer.
If the temp gets over 95ºC, then shut it down.
If you can't see the temps - and the video starts acting up - shut it down.

I can't recall if you ran Prime95 yet? If not, then you'll have to run all 3 tests (Blend, Small FFT, Large FFT) in order to isolate the problems (which test shows errors first, which shows errors last)

Then run 1 or 2 of the CPU stress tests - once again, watching the system for bad behavior. If the system starts acting badly, then shut it down.

 

[Danny72170]

Yes I ran the prime 95 test a few days back and posted the results.

Blend: first cpu error after a minute and last after 1 hour 15 minutes
Large FFT: first cpu error after 2 minutes and last after 34 minutes
Small FFT: No errors after running more than 9 hours

I'll start on the cpu and furmark tests now

 

[Danny72170]

I ran Furmark for about 2 and a half minutes before I got nervous and shut it off.

Temps started to level off at about 84C and max temp was 88C before I shut it off.

There was no instability from what I could see I just got nervous about the fan running so fast. It never ran that fast before so I wasn't sure how long to run the test.

 

[Danny72170]

Ran the test from Hot Cpu Tester Pro.

I've attached the log files from this test.

One says there were no errors from the testing but the other shows an error about 15 minutes into the test so I'm not sure what to think

I will try another test from the list and try the Furmark test again tomorrow

 

[Danny72170]

I was looking more into the Hot CPU tester pro program and one of the items on the left is system info. When I clicked it it said I had 4 GB of memory (4,194,303 KB to be exact). But when I check in the control panel it says I have 6 GB which is what I should have. Is this a problem or is it something else?

 

[Danny72170]

I downloaded Speedfan yesterday and it showed 3 temps above it's norm (showing a flame symbol). GPU, Remote Temp and Temp 2 were the high temps indicated. Now this was after the computer was on for several hours and not long after I was playing Rift so I checked again this morning immediately after turning the computer on.

Remote temp and temp 2 were already above normal and the GPU temp was normal for a few minutes then went high after all I did was start my browser.

Also, speedfan shows 4 fans but fan 2 and 4 indicate 0 RPMs, either those fans don't exist or they aren't running or speedfan isn't detecting their speed

One question I have in interpreting these results I have is what does remote temp mean and why is it so high even when the computer has only been on for a few minutes and what parts of the computer are temp1, temp2 and temp3 and why is temp2 also high right away?

Another question is does any of this matter in determining why I'm getting the BSODs I'm getting?

 

[usasma]

It's the results that tell us where/what the problem is. Unfortunately it's not as specific as we'd like - so we do have to interpret it.

So, for Prime 95:
- the Blend test tests a bit of everything with lot's of RAM being tested
- the small FFT's test the CPU cache and very little RAM
- large FFT's test the CPU/RAM interface (mobo/CPU depending on the model) and the RAM
So, we're looking at it most likely being a CPU problem - but that's not definite yet.

The error that you noted was a File Exception error - and that's not directly related to the CPU test (rather it appears to be an error within the testing program). The program may not be able to recognize more than 4 gB of RAM if it's written as a 32 bit program (even if you're running it on a 64 bit system). I wouldn't worry about either of those issues right now.

I would next run MemTest86+ overnight to make sure you're not getting more errors. This'll help to rule out RAM as the cause - which will then point us towards a CPU issue.

I'm not a bit fan of the little "flame" indicator in Speedfan. Please let us know what the numbers are. Typically we work with degrees Celsius ( ºC )

The temps in Speedfan are from the sensors that it recognizes. In most cases the sensors are standard, so the program names them correctly. But if the sensors aren't used in a "standard" way, then it becomes confusing (such as the remote temp). Does the remote temp ever change it's reading? Or, does it stay the same even when the other temps are changing?

Depending on the temps, I usually recommend the "finger test" to see what's overheating. This involves getting the system up to temperature, then remove the side of the case and unplug the system from the wall. Then poke around (CAREFULLY !!!) to feel where the heat is coming from. What you're looking for is heat that is uncomfortable to the touch (so only get close, don't actually touch anything!!!)

I'd also run FurMark again - and let it get up to 95ºC before shutting it off. The fan's supposed to be working hard as we're stressing the GPU and it's ability to stay cool. As before, if the fan makes unusual noises - or the video displays odd behavior - shut if off no matter what the temp.

Right now we're looking at either overheating or a bad CPU as the source of the BSOD's. Also, be aware that the CPU may be stable under normal temps - and as it gets hotter, it starts to spit out more errors.

 

[Danny72170]

The remote temp is a constant 72C and doesn't very, The GPU temp is currently 56C after the computer is on for several hours and Temp2 varies from 60-65C. I went into the BIOS to look at the temps there and in there the IOH temperature was in the low 60's so I'm thinking that is Temp2. There are temperatures labeled Core0 and Core1 in Speedfan that when I run the CPU test get warmer but in normal operation are running in the low to mid 30s and have a blue downward arrow so I would guess the CPU temps are ok.

When I ran Furmark the first time the fan ran very fast but didn't sound unusual. I'll run it again though.

I'll run memtest86+ overnight to see what I get for results

 

[usasma]

I'm unsure about temp 2 or the IOH - but suspect that it's a bit high.
Try pointing a house fan (on high) at the mobo and see if that lowers the temp 2.