Sign in with
Sign up | Sign in
Your question

The BSOD - New System unstable (Prime95 Error)

Last response: in CPUs
Share
September 26, 2008 10:51:37 PM

Hello,

i have had problems with the BSOD ever since i put together my new computer about 1½ month ago.
It happens very often and seems totally random.

My new computer consists of the following:
- CPU: Intel Core 2 Quad Q9550
- Motherboard: Asus P5Q-E
- CPU Cooler: Scythe Ninja 2
- PSU: Corsair HX520
- Graphiccard: Asus EN9600GT
- Memory: OCZ Reaper PC8500 2x2 GB
- Harddrive: 2x 500 GB Seagate 7200.11

I have installed Windows Vista 64 Ultimate and have my 2 harddrives set up in Raid 1 (one of them i sent back for a replacement currently).

Everything is currently set to default in the bios, except i have changed the DRAM Voltage to 2.20 V to be sure that the Memory gets the right voltage.


I have been trying to find out what is causing the problems that makes the system give me the Blue Screen of Death all the time, but unlucky so far.

The error messages i get is either IRQL_LESS_OR_EQUAL (0xA), PAGE_FAULT_IN_NONPAGED_AREA (0x50) or MEMORY_MANAGMENT.

By looking at the stop codes it seems to be either defect hardware in form of memory in either the RAM, CPU or Graphiccard, or a Driver error.


I have tried with only one RAM dimm in the motherboard at the time, and in different slots. Also i have tried underclocking the RAM. All of it with no success.

I have run MemTest86 without any errors.


Yesterday i then decided to try to run Prime95 (25.6).
I ran with 4 threads because of the 4 cores.

First time it only ran for 6 minutes then one of the threads stopped with the following error:
FATAL ERROR: Rounding was 0.4970703125, expected less than 0.4
Hardware failure detected, consult stress.txt file.

I tried to run it 3 times afterwards. All of the times one or more (1-3 threads / cores) of the threads stopped with the following error, within 2-6 minutes:
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

I didnt get a BSOD any of the times the threads stopped because of an error.

I had the HWMonitor and CpuZ open all the time and the CPU didnt get hotter than 56-57 Degress on one of the cores (which should be okay), it was lower on all other cores.

I have downloaded IntelBurn, but didnt dare to try it out as it should stress the CPU even more than Prime95.



I have no idea what to do now.

Can someone help me ?



Martin
a b à CPUs
September 26, 2008 11:04:07 PM

Two things to do. Reinstall your 9600GT graphics card driver. Get the lastest Vista 64 driver for the 9600GT from Nvidia.com. Also, PC8500 requires 2.0-2.1v, your BIOS defaults to 1.8v. In BIOS, set the memory voltage to 2.1v. This is correct driver for your VC.

http://www.nvidia.com/object/winvista_x64_178.13_whql.h...
September 26, 2008 11:08:39 PM

You may need to update your bios if the voltage change doesn't work. I just did a build with one of the P5Q series and it had some big memory compatibility problems even after changing the settings. The bios update fixed it though.
Related resources
September 26, 2008 11:57:50 PM

badge said:
Two things to do. Reinstall your 9600GT graphics card driver. Get the lastest Vista 64 driver for the 9600GT from Nvidia.com. Also, PC8500 requires 2.0-2.1v, your BIOS defaults to 1.8v. In BIOS, set the memory voltage to 2.1v. This is correct driver for your VC.

http://www.nvidia.com/object/winvista_x64_178.13_whql.h...


I have both tried installing the new drivers from Asus site, and the new drivers from Nvidias site.

It says 2.1 volt on their site, but a little silver sticker on the RAM says 2.2 Volt.
I currently have the RAM (DRAM in bios) set to 2.20 Volt.
I have also tried 2.10 Volt.
September 26, 2008 11:59:34 PM

LAN_deRf_HA said:
You may need to update your bios if the voltage change doesn't work. I just did a build with one of the P5Q series and it had some big memory compatibility problems even after changing the settings. The bios update fixed it though.


I have the newest BIOS (v. 1306).
It seems all their latest changes to the bios have been due to Memory compatiblites.
September 27, 2008 12:58:01 AM

I have just ran 3D Mark Vantage (Trial version) without any problems.

I tried Intel Burn Test, first with the lowest settings and then medium and the 1 pass with Max.
I didnt experience any problems
(But i should probably have 20 pass on max). Just wanted to test if it might be a heat problem.

On core #0 it only got up to 61 degress at max.
September 27, 2008 1:11:08 AM

Everything reported so far points to bit-flipping in RAM.

I would run Memtest more thoroughly with the tightest timings possible, to encourage failure.

If it passes, suspect faulty CPU cache unless you believe your RAM voltage to be inadequate under heavy load.

Lastly, reduce the CPU multiplier to vary internal CPU speeds to further isolate the failure modes.

September 27, 2008 1:29:40 AM

I have run 24 passes in MemTest.


Which timings should i try ?

The spec of the RAM is 5-5-5-18 (with 2.20 Volt it says on a little silver sticker on the RAM), but in the BIOS they default (auto) to 5-5-5-15.
September 27, 2008 1:41:04 AM

MartinFrom said:
I have run 24 passes in MemTest.


Which timings should i try ?

The spec of the RAM is 5-5-5-18 (with 2.20 Volt it says on a little silver sticker on the RAM), but in the BIOS they default (auto) to 5-5-5-15.


Repeating my post for confirmation: use the tightest timings to induce failure: 5-5-5-15 (@2.2v)

Let Memtest run for hours and hours, if it thusly passes....you can treat the RAM as good (with the bonus of being known good in the suspect system) and direct focus on the other parts as described, further attempting to isolate the cause: CPU cache error? CPU multiplier error? Bus signalling error?

You can not even discount your HDDs and HDD controllers, perhaps they are wonky and bit-flipping...that is a subtle diagnosis, but you reported a drive failure...any failure must be factored into the "component failure probability matrix". Fun, eh?


September 27, 2008 3:41:43 AM

Yes, i cant rule out the HDD controller.
I experienced the system freezing and the Raid being broken afterwards in the start.
It happened a couple of time, every time with less time in between.

It was always the same HDD which was broken.
At last i couldnt even repair it anymore, as it could simply not use the drive.
After i have removed the faulty HDD, i havent experienced those problems.

Raid Matrix is still crashing though if i go to advanced and right click on the working drive.
September 27, 2008 4:36:50 AM

My CPU voltage is changing between 1.10 V and 1.20 V depending on the load. Is that to low ?
September 27, 2008 5:08:17 AM

The error messages i get is either IRQL_LESS_OR_EQUAL (0xA), PAGE_FAULT_IN_NONPAGED_AREA (0x50) or MEMORY_MANAGMENT.

Almost always driver issues, and or install of the drivers.
September 27, 2008 5:17:29 AM

I just finished a build with the Asus P5Q-E as well and I attempted to use the Q9300. It would not run stable and Prime95 always failed the 3rd and 4th core. I even ordered a replacement Q9300 and exact same thing.

I then put in 1st a E6600 and worked great. Just today I installed a E8500 and again runs great. It appears that some of these boards have issues with the new quad-core CPUs. I say this because in my personal machine, I'm also using the P5Q-E and I was able to use the Q9300 without issue and I'm running it now as I type.

I'm wondering if some boards have a weak voltage regulator or something preventing the quad cores from running properly.
September 27, 2008 5:49:16 AM

The broken hdd may have damaged the raid controller, or the cache on the other seagate.
Try installing your os on a different drive (without the raid).
September 27, 2008 12:53:03 PM

Tonight, while sleeping i have ran 12 passes of MemTest with everything in the bios set to default / auto (5-5-5-15 - tighter timmings as bf2gameplaya asked me to do), and the DRAM Voltage to 2.20 V.

It gave me 210 errors !


When i woke up i then tried setting it manually to 1066 mhz, 5-5-5-18 and still 2.2 V, and within 1-2 minutes it already got the first error.


It is pretty weird though as i have earlier run 24 passes without any problems with one dimm at the time. Also no problems under Intel Burn Test or running 3D Mark Vantage (trial version).

So i better test the RAM dimms on at the time again and see what happens.



I am wondering if the bios could be the reason for the RAM errors and should i try going back to etc. the 0610 version ?


The CPU cant be the reason for the memory errors in MemTest ?
September 27, 2008 9:51:10 PM

I have done some further Testing of the RAM.

With one dimm inserted at the time, i still got errors both when setting everything to Auto, setting them manually, or underclocking them.

I didnt get these errors last time i tested with MemTest.


I am wondering if the CPU can cause the RAM to fail when running MemTest ?

Or can i be totally sure that it is the RAM when tested with MemTest ?

Should i Try downgrading the bios to an old version ? Is there any chance that the RAM errors is caused by the Bios ?
a b à CPUs
September 30, 2008 1:17:31 PM

^It's probably your RAM, CPU falure is rare.
!