Overclock.net › Forums › AMD › AMD - General › AMD Platform instability
New Posts  All Forums:Forum Nav:

AMD Platform instability

post #1 of 9
Thread Starter 
So, I have an issue with my rig that I'm finding entirely baffling.

Note that no component has ever been overclocked.

Over the past month, I've been experiencing some system instability, causing games to crash, intermittent bluescreens, and general headaches.

I turned to stress testing in order to determine which component was glitching, and got some interesting results. I tested using the following 4 programs:

Prime95 (small FFT and Blend)
Intel Burn Test
HCI Memtest
Memtest86+

Using Prime95 Blend, I experienced errors almost immediately, but was stable for one complete iteration of small FFT (looping back to 8k), before I aborted the test.

Using IBT, which I ran twice, I failed once on iteration 20, and once on iteration 32.

Using HCI Memtest, I accumulated 34 errors over 6 instances of the program (one per core). I ran the program through 100% coverage, but did not attempt a second iteration.

However, using Memtest86+, one complete pass was made, without detecting any error.

Throughout all this testing, CPU temps never rose above 40 Celsius, monitored using CPUID Hardware Monitor.

I'm somewhat flummoxed at this point, as everything seems to be pointing to a memory error, but Memtest86+ is throwing that out the window. I'm currently trying to source some additional DDR3 RAM in order to rerun HCI Memtest and IBT at the very least. If I'm still getting memory errors at that point, I'll have to examine the possibility of a CPU/motherboard problem.

I realize that my testing procedures haven't been the most thorough, but does anyone have any recommendations as to additional tests, tests to retry, or next steps, I'd greatly appreciate it.

AMD Phenom II 1055t @2.8Ghz
Gigabyte GA-890FXA-UD5
Corsair XMS3 DDR3 4x2GB, 1600Mhz, 9-9-9-24
2x XFX Radeon HD 5830 @800 Mhz
OCZ Vertex 2 128GB SSD
4x Seagate Barracuda 7200.11 1TiB (Raid 5)
Antec HCG 750W PSU
Cooler Master Hyper 212+
Cooler Master HAF X (stock fans)
post #2 of 9
Have you looker at Event Viewer?
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
post #3 of 9
Thread Starter 
I wouldn't even know what to be looking for :$

So no, I have not looked at event viewer.

E: I've swapped a known good set of RAM into my machine, and it's currently completed 80 of 100 IBT passes. After this, I plan to set up HCI Memtest again, and leave it running overnight.

I'm also currently running HCI Memtest on the system the good RAM came from, using my suspect RAM. It's getting plenty of errors, more than I got on my own machine, however it's getting all of them on the same instance of HCI Memtest. For those not in the know, this program operates by having the user manually assign RAM to multiple instances of the program, one per CPU core. One instance of the program is getting all of the errors. The CPU cooler on this second machine is not beefy enough to manage either Prime95 or IBT, so I don't think I'm going to do that.

However, if my machine can pass all my tests with the new RAM, can I safely say that the RAM is the only defective component in my system? IBT is only running using 2048 MB of RAM, but HCI Memtest should identify the RAM as clear, right?
Edited by user18 - 8/7/13 at 9:21pm
post #4 of 9
You will find Event Viewer very helpful in tracing down your issues.
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
post #5 of 9
Thread Starter 
I left the good RAM running HCI Memtest overnight, and the bad RAM running Memtest86+ overnight. The good RAM made it to 750% coverage with no errors, and the bad RAM returned several errors over several passes. It looks like it's definitely the RAM. Looks like I won't be getting a new SSD after all :/

Re Event Viewer: I've looked through it, but it doesn't seem to have any errors that match up with my bluescreens, and has been generally unhelpful. It seems like it would be more useful for a software error than a hardware error?
post #6 of 9
No such thing as "1600" ram on a Phenom or Phenom ][. Downclock that shiz to 1333. Mythical beast.
post #7 of 9
Thread Starter 
Quote:
Originally Posted by prjindigo View Post

No such thing as "1600" ram on a Phenom or Phenom ][. Downclock that shiz to 1333. Mythical beast.

Um, that's not the source of the problem. 1600Mhz RAM runs just fine. Multiple instances out there of this exact RAM, CPU and board working just fine together, as well as many more instances with 1600Mhz RAM used with Phenoms/PhenomIIs. It just gets detected in BIOS as 1333, a simple change to the settings runs it at 1600.

Just for the record, the same problems are still evident at 1333Mhz, and the other 1600Mhz RAM kit I tested with had no issues whatsoever.


Now for the most recent update:

I'm starting to wonder if I have a bad RAM slot on my motherboard (please god no)

I have 4 slots (0-3, closest to farthest from the CPU socket), and 4 sticks of RAM (A-D, originally closest to farthest from the CPU socket).
Using HCI Memtest, I've run the following tests to 500% coverage:
0A (no errors)
1B (many errors)
2C (no errors)
3D (no errors)

At this point, I thought I'd identified the problem, DIMM B. But just to double check, I ran two more tests (the second of which is finishing as I type).

1D (a couple errors)
3B (no errors)

Which makes it seem more like slot B is the problem. But the RAM (which has now had each individual stick test good at least once) didn't work in the other system....

I'm kind of at a loss as to what the next step is here :/

E: Ran 2B as well, a bunch of errors, but all in the same instance of HCI Memtest. This has happened a few times, and I'm not sure exactly what it means.....
Edited by user18 - 8/8/13 at 8:55am
post #8 of 9
boost the voltage on your ram by a tenth of a volt.
Edited by sparkeyjames - 8/9/13 at 3:27pm
post #9 of 9
Thread Starter 
Quote:
Originally Posted by sparkeyjames View Post

boost the voltage on your ram by a tenth of a volt to 1.6.

I actually solved the issue last night, this did end up being part of the solution, so +rep.

When I first got the RAM, I set it to its rated settings, as the board automatically set it to 1333 speeds, due to limitations of a Phenom II, as mentioned above by another user. As part of this, I bumped the voltage from 1.5 (the rated voltage for the RAM at 1333Mhz) to 1.65 (1600Mhz).

When I swapped the other set of RAM in to test, the voltage got switched back to 1.5V, the rating of the new kit. However, when I put the original RAM back in, I neglected to bump the voltage again, leading to the erratic results I obtained earlier.

After realizing my error and correcting the voltage, I ran each DIMM one at a time, and discovered that one particular DIMM was not working properly, even at loose timings and slow speeds. I've contacted Corsair and arranged for an RMA.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD - General
Overclock.net › Forums › AMD › AMD - General › AMD Platform instability