DDR3 1866 boot failure, 1600 unstable, 1333 OK

[Peroxyde]

Hi,

My new built suddenly have stability issues: boot loop, computer hang up. I think it's may be a RAM, but I am not sure. Here the details, all components are brand new:

Motherboard: GA-970A-D3 (rev 3.0) + latest BIOS version FD (Jan 2013)
RAM: 32GB, 2x (8GB x2) Patriot Viper3, Black Mamba, DDR3 16GB , 1866. Each pair is mount in the matching color slot on the motherboard.

CPU: AMD FX-8320 + Cooler Master Hyper 212 EVO

PSU: Corsair VX550X

No heat issue: Good fans, no overcloking, I even put a home fan blowing directly inside the case during the test to make sure heat is not the cause of the instabilities.

Hard drive, Video and other components are not worth mentioning. This computer is used to play around with Hyper-V server 2012.

The very first thing I did was to test the RAM using Memtest86+ 4.20. The test ran overnight, zero error. However, I noticed that I was never able to use XMP Profile1. The motherboard refuses to boot and a screen pops up to suggest to change BIOS settings.

I then disabled XMP and let everything in Auto (CPU, RAM). For the RAM, I saw the multiplier is 8 which gives a frequency of 1600 MHz. Then I could install HyperV Server 2012 and the computer ran fine for about 1.5 months. I left it on and went in vacation for a week. Then I found the machine powerd off. From that point on, the server became unstable: boot loop that I could not solve. I end up re-installing HyperV from scratch, but couldn't because even a re-installation is affected by unpredictable reboot.

I then mount the hard drive of the HyperV server on my desktop computer and could complete the re-install of HyperV 2012. Some more hard drive and various components swappings showed that what run OK on the desktop computer could not run reliably on the AMD FX-8320 + 32GB Ram described above. I guess that may be the Motherboard or the RAM could be the cause. The easiest thing to test first is to lower the speed of the RAM. I reduced the RAM frequence multiplier from 8 to 6.7 to make it 1333 Mhz, all other timings settings remain automatic. Then the computer seems to run fine.

Question 1: What is the reason for the motherboard to refuse to boot when the RAM is set to use XMP Profile 1? And yet the RAM is XMP compatible and is rated at 1866.

Question 2: Is there a better way to use Memtest86+ 4.20 to test the RAM? Here is how the RAM are recognized by Memtest86+: https://app.box.com/s/xogkjoe951nw9cdd3qta

Question 3: Why did the RAM pass Memtest86+ but become instable when running software? Do you have any recommendation to help me to investigate further?

Question 4: RAM voltage. The BIOS screen show different values of voltage. What is the correct reading? Could the voltage be the cause of the instabilities? Photo of BIOS screens: Screen 1 (M.I.T) https://app.box.com/s/8fsd6cvdwdnx965aaz3h
Screen2 (Advanced Voltage Settings) https://app.box.com/s/uwfgb0mcq478bizj5vyu


Sorry for the long post. Any help would be greatly appreciated. Thanks

 

[dma0991]

1. There could be incompatibilities due to the weakness of the RAM itself or the BIOS version isn't suited for the XMP setting. I've used Kingston HyperX 8GB rated at 1600MHz @ 1.65V only with XMP enabled. With XMP enabled, on some BIOS revisions, it is the same problem with boot loops, etc. I never got to run at its advertised specs, it defaulted to 1333MHz 1.5V without XMP.

2. Refer to Overclocking CPU/GPU/Memory Stability Testing Guidelines.

3. Its the same with CPUs. I could overclock mine to be nearly stress testing stable but the minute I run a program, it BSODs. Simply put, stress testing only covers some portions of its stability and does not guarantee 100% stability.

4. Screen 1 is showing you the voltage it is currently running, in real time. Screen 2 shows the maximum voltage that is defined. You can change the voltage settings in Screen 2 but in Screen 1 its just showing you.

 

[Peroxyde]

Hi dma0991,

Thanks for your input. I'll try Prime95 and HCI Memtest as you suggested. What is strange is that the computer had worked OK for 1.5 months.

Installed yesterday WinXP + BurnInTest, ran OK with RAM at 1600 Mhz. But could only test 4GB. Now I installed Windows 2012 R2 Datacenter Preview. The installation went OK, I could log in OK. But couldn't do much as the boot loop cycle comes back again. The Windows Memory Diagnostic ran OK. The only thing I can do is to boot in safe mode.

 

[Peroxyde]

UPDATE: looks like it's a software issue. May be the Nvidia drivers version 320.49, 64 bits for Win7 & Win8 is the culprit. The video card is an Asus GT210 Silent, 512MB.

In my previous re-install of Win2012R2 even when booting in Safe mode to uninstall the Nvidia driver I could not get rid of the boot loop. I have re-installed Windows, without NVidia drivers and for now BurnInTest seems to work OK (which I couldn't run previously when NVidia driver was installed).

Looking up the internet, using keywors "NVidia driver boot loop", then I found plenty of people having the same issue: Blue Screen or Boot loop appear suddenly after a Windows update.

Have any of you had experienced instabilities issue because of NVidia drivers? And how did you solve it?

 

[Nec_V20]

Question 1: What is the reason for the motherboard to refuse to boot when the RAM is set to use XMP Profile 1? And yet the RAM is XMP compatible and is rated at 1866.

Because marketdroids and salescritters are lying scumbags and in their opinion if the system is able to pass POST with "XMP" then it is compatible.

Always remember that "compatible" will not necessarily mean it will "work".

 

[denis280]

UPDATE: looks like it's a software issue. May be the Nvidia drivers version 320.49, 64 bits for Win7 & Win8 is the culprit

it is.in fact on nvidia site they mention a problem about the drivers.they also made a new one to correct it.but still not to good.i am running the one from march 314.?? and it is fine.just waiting for thems to get the right one.and for the rams did you make sure they run on the right voltage.

 

[Peroxyde]

UPDATE #2: It's the PSU, nothing to do with NVidia driver.

Without any NVidia drivers installed. I could run the RAM Tested OK using HCI Memtest (I stopped at 650% coverage after 15 hours). Prime95 also ran OK for about an hour. I then rebooted the computer and the boot loop cycle came back. Booting in safe mode, I saw an EventID 41, Log Name: System, Source: Microsoft-Windows-Kernel-Power, Level: Critical.

May be the PSU? Because the 3 last components I didn't test are Motherboard, CPU and PSU. I borrowed the PSU from another computer and all boot loop disappeared. Every tests succeeded. Even the old hard drive havinh HyperV server 2012 which boot loop at every login now runs OK.

What a weird issue, I have no idea as why the PSU decides to cause Window to reboot but not the BIOS screen. I hope this time I got the right faultly part.