Constant unexpected shutdowns with RAM in Dual Channel config

[element6]

Recently threw together an inexpensive HTPC build:

• 
  ■SeaSonic S12G-650 PSU (Originally purchased a Corsair CX600 and returned it)
  ■Gigabyte GA-78LMT-USB3
  ■AMD FX-8350 Black Edition
  ■Mushkin Enhanced Silverline 16GB (2x8GB) DDR3-1333 (PC3-10600) 9-9-9-24 Model 997018
  ■MSI R9 270x 2GB
  ■Pioneer BDR-209DBK
  ■Samsung 840 EVO 250GB SSD (boot)
  ■2x Hitachi Deskstar 4TB 7.2k 64MB 3.5" HDD in Software RAID 0
  ■Arctic Freezer Heatsink
  ■Lian Li PC-A04B case

Course of actions:

• ■I put it all together and installed Win 7 and Win 8 on it and started noticing problems. It would just randomly reboot itself after being powered on for a few minutes. It never lasts longer than 30-40 minutes. Typically, it will run for about 20 minutes or less and will lock up very briefly and unexpectedly and then you just get a black screen, *beep*, and BIOS splash screen as it boots back up again.

Windows error logs and dumps show no reason for shutdown. All I get is an entry that says "unexpected shutdown" whenever it goes down.

• 
  ■My first thought was that it was a power issue (at the socket, I thought that maybe it was maxing out my Tripp-Lite UPS with two high-powered PC's hooked up at the same time). So I hooked it up in another room with my APC 900VA UPS. Same problem. Then I figured it might be the power supply itself being flaky. So I ordered the SeaSonic 650W and returned the Corsair 600W. Same problem.
  ■The next course of action was to remove the PCIe graphics card and use integrated graphics. Same problem.
  ■Next - removed all peripheral devices, HDD's, and Opt. Drive. - Same problem.
  ■Then I sent in the system board on an RMA to Gigabyte and they bench tested it and told me there was nothing wrong with it. Shipped it back to me and I received it back again last week - still having the same issue.
  ■Currently down to no attached devices, no HDD's, no opt. drive, no PCIe graphics and have been running memtest86+ v5.1.0 many times for the past 24 hours. It is still unexpectedly shutting down during memtest86 diags. I was only running it in dual channel mode last night (both sticks installed). It would never last longer than 25 minutes without an unexpected reboot. There was only one instance out of 8 attempts where memtest reported an error on the screen before it rebooted itself. I had to set up a tripod and camera to record it so I could review where it crashed each time.
  ■Today I ran memtest86 with only one DIMM in slot 1 on the motherboard. It ran for over an hour every time with no errors and no reboot or hang. I alternated the two DIMMs and let memtest run for over an hour each time. No problems at all. No crashes.

So I guess at this point I have a few possibilities:

1. Did Gigabyte thoroughly test my system board with Dual Channel RAM to ensure the Northbridge or something else is not defective? I did not send my Mushkin RAM in to them with the board because I was instructed not to.

2. Could my RAM in fact be defective even though when testing each DIMM individually the system is stable?

3. Could my CPU somehow be to blame for this issue?

Any thoughts would be greatly appreciated.

I really wanna start loading this PC up with my HyperSpin library after staring at it for over a month now and dealing with RMA's 🙁

 

[Tradesman1]

Try raising the CPU/NB voltage to 1.2 and the DRAM voltage to + 0.05 over spec (i.e. if DRAM is rated 1.5 volt go 1.55) and try

 

[element6]

Further info:
■RAM timing/voltages are exactly as they should be set when I allow the motherboard to use automatic settings. I even set it to manual mode and used the same settings to make sure nothing was changing.

When I set up my camera and recorded the system running memtest86+ I was reviewing what it was saying right before it shut down and I made a text file of what was on the screen during each unexpected shutdown. I had to review the camera footage and type everything in by hand since I was running memtest from the boot CD.

What occurs is this - memtest86+ Free V5.1.0 set to run all tests (0-10) at 99 passes. System will run tests until unexpected shutdown with RAM in Dual Channel config. When it gets to the point where it reboots unexpectedly, everything on the screen locks up and stays there for about 2-5 seconds. Then the screen goes black, BIOS splash screen shows up, and system reboots like nothing happened.

Here is the text output of what was on the screen before each shutdown with a brief explanation next to each crash instance (there are 5 total that I recorded with memory in dual channel config).

http://pastebin.com/pzX781SQ

 

[element6]

Try raising the CPU/NB voltage to 1.2 and the DRAM voltage to + 0.05 over spec (i.e. if DRAM is rated 1.5 volt go 1.55) and try

Ok, in this BIOS most voltage adjustments are under the M.I.T. (MB Intelligent Tweaker) Menu.
Just raised DDR3 Voltage to 1.550V
CPU Voltage control left at [Normal] and shows 1.4250V
CPU NB VID Control left at [Normal] and shows 1.2025V
Nonadjustable voltages listed below these show:
Normal CPU Vcore - 1.4250V
Normal CPU Vcore MB - 1.1625V

Above the DDR3 Voltage setting I have the following adjustable setting:
NorthBridge Volt Control - currently= [Normal] - When selected I can choose between +0.1V, +0.2V, or +0.3V. I left it alone for now.

Also maybe worth mentioning - Under RAM config - Rank interleaving and Channel Interleaving are both enabled,

Just started memtest86+ again with both DIMMs installed and DDR3 voltage set to 1.550V

Will report back.

 

[mamasan2000]

Did you try 'Load failsafe defaults' in Bios?
Try to put as much of the options at auto as you can. Let the Bios figure it out =)
That includes Rank and channel interleave.
Also, are you using XMP-profile in Bios? Try with it on and off, see if theres a change.

Last thing to check is your motherboard manual. Check what it says about how to install 2 sticks running dual channel. Which slots to use.
Reading the manual on page 5 and 11 it seems you should use slot 4 and 2. The 2 slots closest to CPU. If that doesn't work, try slots 1 and 3.

 

[element6]

Did you try 'Load failsafe defaults' in Bios?
--Everything was set to defaults to begin with
Trry to put as much of the options at auto as you can. Let the Bios figure it out =)
That includes Rank and channel interleave.
Also, are you using XMP-profile in Bios? Try with it on and off, see if theres a change.
--Yes, as described in one of the posts above, I have let it use auto settings and have manually set timing and other settings.
Have you tried running memtest with just 1 RAM-stick inserted? Test both sticks separately.
--Yes, also described in detail above

After increasing DDR3 Voltage to 1.550V I let memtest86 run again. It has not locked up or rebooted yet. I believe the voltage being set to 1.500V may have been to blame. I will let it run overnight and reply back.

 

[Tradesman1]

That may well take care of it, the 8350 is rated for up to 1866 at 1 stick per channel - however they based this on testing with 4GB sticks, so often the little extra voltage to the DRAM and the MC (memory controller) helps with 8GB sticks or going to 4 sticks

 

[mamasan2000]

I have also noticed that RAM volt by default on my mobo is 1.48 (according to HWinfo). Doesn't matter if I run 1333 or 1600MHz. So I bump it to 1.53-1.55v. HWinfo reports the voltage at 1.5 roughly after bump.

 

[Tradesman1]

Could be a fluctuation, or a glitch with the voltage regulation, as long as it's stable it's fine 😉

 

[element6]

Over 14 hours of memtest running with RAM in Dual Channel -
0 Errors
0 Crashes

Thank you Tradesman1
Thanks mamsan2000 for replying as well

Copying my HyperSpin library over now

 

[mamasan2000]

Glad you got it sorted

 

[Tradesman1]

Looking good! 😉

 

[tepublico]

Hi from Spanish-tan 😀

This issue break my head arroun 2 months, thanks a lot, i will try it 😀
in all this time when i use 2 ram slot sistem crash
when i use just one ram slot system work nice!

Here is the pc hardware:

AMD Socket AM3 FX Series FX-8350 4.0Ghz 8X Black Edition
Gigabyte GA-990FXA-UD3 Rev 4.0
Ripjaws DDR3 1333 PC3-10666 16GB 2x8GB CL9
Nvidia GeForce GT 730 2GB GDDR3
TacensMars Gaming 800W
Seagate Barracuda 7200.14 1TB SATA3
Samsumg ssd 500 gb
NOX Coolbay SX Devil USB 3.0
Cooler Master Seidon 120V R2

 

[mamasan2000]

Something I haven't mentioned yet. Command rate for RAM. Try set it to 2T. 1T can be too agressive for stability.