[BanEmAll]

Hi there. I have been experiencing some computer trouble lately and getting blue screens. This is a rather new build with Windows 7 64 bit. The specs include:
AMD 1090t (stock cooler)
G.Skill 2x2GB 1600mhz
MSI 890FXA-GD70
Nvidia GTX 260
two rather old seagate 250gb hds in RAID 0

It seems the blue screens normally have different errors, the one mentioned memory_cache, which leads me to believe that one of, or both memory sticks are bad. I ran memtest86+, one stick got a error but when I ran the test repeatedly, it didn't have any errors show up. I then used the windows memory test which didn't show anything. I tried stressing the CPU and RAM using Prime95. Each time, a few minutes into the test, one of the workers will stop because of an error. All it says is "Hardware failure detected, consult stress.txt file." Not very helpful when I can't find it, although I read that the stress.txt doesn't have anything useful online elsewhere. I use core temp to monitor my cpu temperatures, it idles at 26c and maxes at 60c. So nothing stange there. My graphics card drivers are updated.

When I look at the XML view of the Windows event viewer, I see this for the blue screens:

Code:

<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
- <System>
  <Provider Name="Microsoft-Windows-Kernel-Power" Guid="{331C3B3A-2005-44C2-AC5E-77220C37D6B4}" /> 
  <EventID>41</EventID> 
  <Version>2</Version> 
  <Level>1</Level> 
  <Task>63</Task> 
  <Opcode>0</Opcode> 
  <Keywords>0x8000000000000002</Keywords> 
  <TimeCreated SystemTime="2011-07-16T05:03:36.592835000Z" /> 
  <EventRecordID>24046</EventRecordID> 
  <Correlation /> 
  <Execution ProcessID="4" ThreadID="8" /> 
  <Channel>System</Channel> 
  <Computer>Taylor-PC</Computer> 
  <Security UserID="S-1-5-18" /> 
  </System>
- <EventData>
  <Data Name="BugcheckCode">52</Data> 
  <Data Name="BugcheckParameter1">0x50853</Data> 
  <Data Name="BugcheckParameter2">0xfffff880031537e8</Data> 
  <Data Name="BugcheckParameter3">0xfffff88003153040</Data> 
  <Data Name="BugcheckParameter4">0xfffff80002d217b2</Data> 
  <Data Name="SleepInProgress">false</Data> 
  <Data Name="PowerButtonTimestamp">0</Data> 
  </EventData>
  </Event>

Any ideas on what could be wrong or what I can do further to narrow down the problem? Thanks!

[trents]

We get this all the time. First of all if your are running your 1600-rated memory try lowering it to 1333. Generally, 1600 means 1333 that will overclock to 1600 on Intels and on some AMD systems when a little more voltage is applied to the ram or the timings are loosened. If you don't believe me then run CPU-z and look at the "SPD" tab, the farthest right column as compared to the other JEDEC speed columns. You will probably see that to get it to run at 1600 you need more volts or relaxed timings or both and that the column may be headed something like "XMP 1600", the XMP referring to overclocked or "extra" mode. The other fact is that the current generation of AMD CPUs have an integrated memory controller rated only at 1333 mhz.

60 C is probably past the stability threshold for current AMD CPUs which generally become unstable once you exceed mid 50s C.

[Rigit]

I agree with everything trents said. You didn't say of the 1090t was stock speed or not. 60C is usually the max stability wise for any AMD CPU. If you're oc'n the 1090t then 60C will not work at all. Every time I've oc'd when I got to or even close to 55C BSOD. OC'n makes the CPU even less tolerant to heat. Back off the memory timings just so we're safe. Then do what you need to do to get that CPU temp down to 55C or below at stock clock. Then repeat the test. If you plan on oc'n then you'll have to be that much cooler. I'd advise you dump the stock cooler for something better. Once you get the temps down and you're stable then start pushing the memory and/or oc the CPU.

[trents]

Originally Posted by Rigit

I agree with everything trents said. You didn't say of the 1090t was stock speed or not. 60C is usually the max stability wise for any AMD CPU. If you're oc'n the 1090t then 60C will not work at all. Every time I've oc'd when I got to or even close to 55C BSOD. OC'n makes the CPU even less tolerant to heat. Back off the memory timings just so we're safe. Then do what you need to do to get that CPU temp down to 55C or below at stock clock. Then repeat the test. If you plan on oc'n then you'll have to be that much cooler. I'd advise you dump the stock cooler for something better. Once you get the temps down and you're stable then start pushing the memory and/or oc the CPU.

Don't just back off on the memory timings back off on the frequency.

[BanEmAll]

I planned on ocing in the future, however, I wanted to fix this first. The computer is currently all stock with no overclocking. The processor is at 3.2 Ghz. I planned on upgrading the cooler to a Hyper 212 plus in he near future. Going through my bios. it looks like I can't change any of my voltages or the memory frequency. Nor can I find a setting to change them. I was able to change my timings however to match the ones in the SPD tab under XMP-1600 column as you suggested. You can see what I'm looking at in the screen shot below.

Thanks again for all your help guys.

edit: I ran Prime95 and didn't receive any errors/warnings.

[trents]

You certainly can change the frequencies voltage settings in bios. That's an expensive board and surely it has all the bios bells and whistles imaginable. Take them off of "Auto" so you can see the manual choices.

[trents]

Here's one of your problems right here. See what I have circled.

[BanEmAll]

Fixed the 1T and made it 2T for the command rate. I was also able to set the frequency to 1333. Now in the Memory tab for DRAM frequency, it shows 666.7 Mhz. Is that normal? I honestly wasn't sure what to set the voltage to. On the memory stick, it mentions 1.5 volts. I would imagine thats recommended. So should I up it you think?

Edit: I have a zip containing some of the minidumps of from my BSODs if your interested. You would probably understand them more than me.

[trents]

No, don't "up it". For CPU-z to show 666 mhz when you set the ram to 1333 in bios is correct. CPU-z shows the memory speed as the DDR2 bus speed whereas in bios you are working with the speed expresed in terms of DDR3 quad-pumped speed. In other words the DDR2 bus speed of DDR3 1333 ram is actually 666. Just double it in your head to convert to DDR3 equivalent. Now set the memory voltage to 1.55.

Also, please (at least temporarily) disable Cool N Quiet and C1E in bios. Go into Windows Control Panel and change Power Options change from power saving to Performance. I want to see what your CPU-z "CPU" tab shows to CPU core voltage to be without the down-throttling, power saving technologies being active.

Rrerun the Prime95 bend stress test for 2 hrs. with HWMonitor open on the desktop as you do this and see if you still get errors. You still may have a core temp-related stability problem. By the way, what are your room temps running lately?

Please post back with pics of CPU-z tabs: "CPU", "Memory" and "SPD" and of the HWMonitor interface after running the test.

[BanEmAll]

During the day, today it was also in the 100s outside, but my room temp was at 80, at night, it is now 75. I set the dram voltage to 1.55, disabled cool n quiet and c1e in the bios and changed it to performance for power options. I ran the test for 12 hours because I ended up working. No crashes, no workers in prime 95 stopped. Here is the screen shot I took right before I stopped the test. http://i.imgur.com/VWHEF.jpg

I just put a link to it as the attachment made the picture full sized on the screen which was huge.

[trents]

For some reason your core temps are showing "0". Those are the critical ones. Do you have ACC or (whatever core unlocker terminology your bios uses) enabled? If so, disable it because that prevents core temps from being monitored. Besides, you don't have any cores to unlock anyway. CPU temps are high at 65 C but they aren't as important as core temps. Where do you live? Sounds like its pretty hot there.

[BanEmAll]

Heres the new screen shot. Still no trouble with the stress test.

[trents]

Your core temps are plenty high! We usually advise not to exceed mid 50s c with core temp because instability usually sets in at that point when overclocking but a non-overclocked computer is more tolerant from a heat standpoint. For the safety of the CPU, however, I would not let yourself exceed 65 c.

Is this computer overclocked? Lets see a pic of CPU-z "CPU" tab, please.

How have you been stress testing this system. Prime95? How long are you running the stress tests?

[BanEmAll]

I stressed test for 2 hours. The system is not overclocked. The cooler is also stock. The cpu tab can be found in this picture. http://i.imgur.com/VWHEF.jpg

I am using prime95.

[trents]

Are you still having trouble with random blue screens? I see you have lowered your memory frequency and that was my main suggestion.

[BanEmAll]

No, it hasn't recently since the changes. I'll keep you updated though.

[trents]

You mentioned stress testing the ram with Memtest86+ but have your stress-tested the entire system with Prime95 or some similar general purpose stress tester that loads the CPU to 100%?

[BanEmAll]

I have done the blend testing with prime95 to put the cpu at 100% load.

[BanEmAll]

What should I do about my Cool N Quiet and C1E being disabled in bios? I haven't had any crashing recently.

[trents]

You can re-enable them and see if it causes problems. Usually they only cause problems at higher overclocks.