F3-10666CL7D-8GBRH on Gigabyte GA-H67A-UD3H - Random Crashes

[maddogmdd]

I'm trying to get F3-10666CL7D-8GBRH (2x4GB) to be stable on a Gigabyte
H67A-UD3H board with an i5 2500k. This RAM is listed on Gigabyte's supported list. Using the latest BIOS and everything at stock speeds, and using built in Intel SandyBridge graphics.

Windows 7 x64 randomly crashes about twice per day on average. Can happen under
load, just surfing the net, or completely idle. It goes to a scrambled/flickering screen and the only way out is to reset the machine.

The RAM passed memtest86+ for about 3 passes (running more passes now).

I am using XMP (profile1) which seems to be setting the correct main timings 7-7-7-21 and a command rate of 2. I'm at 1.5V. Could there be a problem with the other timings? What numbers should I use?

I read somewhere that there is a different XMP for 1155 boards than 1156 boards.

Thanks

[Tradesman]

May want to try inputting your base timings, voltage and CR manually..and could you post what voltages you are currently at

[maddogmdd]

Ok I put in these timings manually:

CAS 7
RCD 7
RP 7
RAS 21
CMD 2

Left the rest on auto (NOT XMP profile). The following settings differed between auto and XMP, in case it matters:

SETTING Auto XMP
RC 33 29
RFC 108 120
FAW 20 22
Round Trip Latency 32 34

My voltages:
DDR: Set at 1.5V, reading as 1.536V in BIOS
VTT: Set at 1.05V, reading as 1.076V in BIOS

Also, my BIOS has something called "Performance Enhance" with the options: "Standard" "Turbo" and "Extreme"..... it defaults to Turbo so I left it alone. Not sure what it does but it's in the RAM settings page.

Does everything look right? Anything to try changing in case it crashes again? Thanks

[maddogmdd]

Update: just crashed again.

I am now trying running the ram at 9-9-9-24 just to to see if that makes any difference.

[GSKILL TECH]

Performance Enhance "Standard". tRC should at least be 33.

Do you have the latest BIOS? Test one module at a time.

Thank you
GSKILL TECH

[maddogmdd]

Update: I tried replacing the 8GB kit (F3-10666CL7D-8GBRH) with a working 4GB kit from another computer (F3-12800CL9D-4GBRL). I changed the BIOS timings back to "Auto" and made no other changes.

With the 4GB kit, the random crashing has stopped... I've been stable for 2 days.

The 8GB kit had no errors in memtest for 7 passes, but it now seems that they may be the cause of the crashes. Is this even possible?

Assuming so, I'm planning to RMA the 8GB kit to Newegg for refund and maybe try the F3-10666CL7D-8GBXH, since it is the same price and timings but is "designed specifically for Intel Sandy Bridge CPUs, H67 and P67 Platforms."

Is this worth a shot? Is it the same kit just different branding, or is there actually a difference?

Thanks!

(Btw, yes, I am running the latest BIOS and tried the settings you mentioned to no avail. Also tried running at auto 9-9-9-24 and that didn't help)

[GSKILL TECH]

Did you try one at a time to see if it may be a problem with one module?

The XH is a different kit, so it may be more compatible with your H67 chipset.

Thank you
GSKILL TECH

[maddogmdd]

Thanks for the info. Actually I just ordered the XH kit, but continuing to troubleshoot just out of curiosity while I wait for it to arrive.

I'm trying one stick right now. No crash so far after about 1+ day. Will switch to the other stick probably later tonight.

[baobab]

I hit the same problem.

My system hit the same problem. I have the same memory kit and motherboard from newegg. This problem is really annoying.

[maddogmdd]

Definitely an annoying problem. Does yours do the crazy screen thing too?

Well I tried each stick by itself for about a day each. Neither one caused a crash. Would have tested longer but my RMA was due.

So I've returned those and I just put in the Ripjaws X F3-10666CL7D-8GBXH kit, which is designed for SandyBridge but is otherwise the same specs and price. So far so good at 7-7-7-21 2T 1.5V, but it's only been about 20 hours so far. if it lasts through the weekend without crashing I'll consider the problem solved. Cross your fingers!

[baobab]

It crashed with the windows blue screen and then dumped the core.
BTW, does newegg charge any restocking fee for memory returning?

[maddogmdd]

They do, but if you e-mail them explaining that it was defective they'll waive it (at least they did for me).

They do make you pay for return shipping though. $9 for me. Fair enough.


Be sure to try all the standard troubleshooting stuff first if you haven't already. At least with BSOD's you have something to work with. This forum and also Tweaktown's Gigabyte thread are useful places to go.

Are you running the latest Intel graphics drivers?

[baobab]

Thanks very much for the information.

Actually I'm not using GPU embedded in Sandy Bridge CPU. I'm using a Galaxy PCIe video card with Nvidia chipset.

[baobab]

some update ...

When I used XMP profile1, standard Performance Enhance, memtest86+ hung randomly. Then I disabled XMP, and still used standard performance enhance. memtest86+ became better this time, but still hung after running some time. After that, I decreased the mem frequency to 1066, memtest86+ passed.

Currently I'm using the system with 1066 mem frequency, haven't hit the crash yet.

So it looks this memory model cannot work with GA-H67A-UD3H with 1333 frequency.

[maddogmdd]

Never tried running them at 1066, although memtest never hung for me, even at stock settings. Ran it for 6 or 7 passes. The sticks just didn't play nice together in Windows for some reason.

I can tell you that the 8GBXH kit has been working flawlessly for 48 hours at this point. You might to give it a go, especially since it looks like you narrowed the problem down to the RAM freq.

I'll post back if it happens to crash again.

Good luck. This setup rocks once you get the memory worked out!