Random Reboots with 4 Modules (2x F3-12800CL7D-4GBECO Kits) on a H55-GD65

[mazumoto]

Hi :-)


I recently upgraded my system from 4GB RAM to 8GB. For that I got a second Kit of G-Skill F3-12800CL7D-4GBECO, so now all 4 memory slots have the same modules (of course the second set is manufactured later). With that setup I experience random reboots (about once a day) - no matter how I set the timings / voltages. The XMP-Profile does not work with 4 modules installed, manually setting the timings does work though. But even on "Auto" settings (CL9, 1333MHz and 1.5V) I experience the reboots.
Each set on its own (2 modules installed) works flawlessly (multiple days of continuous uptime).

I'd be very glad for any pointer to what could be the problem and how to solve it. Stability is my utmost concern (the system is running on 4GB at the moment), but I also need the larger memory.


Model of the motherboard: MSI H55-GD65 (MSI H55M-P33 shows similiar behavior)
BIOS-Version: Latest available: 1.5
Model of the memory: 2x G-Skill F3-12800CL7D-4GBECO
Model of the CPU: Intel Core i5-650
Problem: Random reboots about once every 24h


Thanks and regards,
mazumoto

[Tradesman]

Might drop the DRAM voltage to 1.4 (1.5 is a little high w/ the 1.35 rating) and could you post your DRAM advanced timings and your other system voltages, sounds like it just needs a minor tweak

[mazumoto]

Well, pictures are probably the easiest way ... these should show the important settings:

General settings, Timings overview and DRAM Voltage:


Manual DRAM Timings


SPD and XMP Information


The reboots apparently happen more often than I said before - I now had 2 reboots in about 1 hour. One more thing came to my mind: I think I didn't have reboots when running memtest86 (which also found no errors). If required I can of course retest this for a few hours.

Oh and about the voltage - as you can see I currently run the modules at 1.35V with conservative timings (1333MHz, CL9). But I don't think that 1.5 would be bad, I'm quite sure I've seen postings about these modules stating they easily eat 1.65V without taking damage.

Regards,
mazumoto

[Tradesman]

Things seem to look pretty good, I think I'd raise the vCore to 1.2 or slightly more, would also suggest disabling EIST and C1E

[mazumoto]

Thanks for your suggestions, Tradesman :-) I'm currently running on those settings (after the computer idled stable for 14h on the old settings).
Yet power consumption is another important factor for me (it's a 24/7 server system) and all three suggestions won't be beneficial in that point. Doesn't make much sense to use "ECO" RAM on 1.35V if I have to increase the CPU Voltage for using it and disable Speedstep. Well, I'll see if it is stable that way and then try to re-enable EIST/C1E.

Another question someone here might be able to answer: might this be solved by an updated BIOS ? Many BIOS-Updates for my board state they'd improve "Memory Compatibility" ... so does it make sense to contact my motherboard vendor (MSI) ?

Thanks & Regards,
mazumoto

[Namor]

Usually I'd associate random reboots as a CPU core problem.

But I didn't even read the first post, so disregard what I'm saying.

[mazumoto]

Well, diabling EIST, C1E and upping the VCore to 1.226V didn't help ... just had my reboot again.

Namor: the system runs perfectly stable (at least half a year) with only 2 modules installed, so I highly doubt it's a CPU problem. It even runs stable overclocked to a Baseclock of 160 (3840MHz CPU speed instead 3200MHz) at stock voltages (Ram still reduced to 1.355V).


[EDIT] I'm now testing two modules in memory slots 2 and 4 - first result is that overclocking isn't possible at all in those two slots (neither CL7 at 1333MHz nor CL9 at 1600MHz). It booted now with CL9 at 1333MHz and I'll test that for stability. Don't know if that test is sane at all, but at least it's giving some results ;-)

[GSKILL TECH]

Updating BIOS would be the next step to take especially if there is a memory update.

If problem persists, it may be a compatibiliy issue between the two kits, so send them in for RMA to see if two new kits would work together.

Thank you
GSKILL TECH

[mazumoto]

Hm, well, it seems to run stable on memory slots 2 and 4 with CR2T, CL9 and 1333MHz.

@GSKILL TECH:
I do run the latest BIOS, but I could of course contact my motherboard manufacturer (MSI) and see what they can do ...
Should I really RMA both kits at the same time ? (I'd rather RMA only one kit - probably the older one - to be able to use the computer during that time)

Thanks & Regards,
mazumoto

[Tradesman]

Probably want to contact RMA direct, at rma@gskillusa.com or rma@gskill.com, they should be back in Monday, could reference your post:

http://www.gskill.us/forum/showthread.php?t=7456

[GSKILL TECH]

mazumoto,

I understand, but if it is a compatibility problem between the two kits, then sending in one kit only does not fully solve the problem. The new kit may not be compatible with the kit you keep, so you can have the same exact problem. Of course assuming that is the problem.

Thank you
GSKILL TECH

[mazumoto]

@Tradesman: Thanks for the adresses. I'm currently talking to my motherboard manufacturer and will see what they say before contacting G.Skill.


@Gskill Tech: YOu certainly got a point there :-) Today I tried to mix the kits and use one module of each kit in a 2-module setup. So far this works stable. Might that be a way to determine if the kits are compatible ?

There seem to be two rules for my board:
1. Using DIMM Slot 2 or 4 doesn't allow overclocking (DIMM Slots 1 and 3 are recommended for Dual Channel). Working is only 1333MHz at CR2 9-9-9-24 with 1.355V.
2. Using 4 modules leads to instability

[GSKILL TECH]

Try one kit in one color slots and the other kit in the other color.

Thank you
GSKILL TECH

[LeetMiniWheat]

I would try arranging the kits into different channels if you think they aren't playing well together. for example, kit 1 in slots 1&3, then kit 2 in slots 2&4 (like GSKILL said above), and if that doesn't work try kit 1 in slots 1&2 and kit 2 in slots 3&4

also if you want a faster way to test stability, try IntelBurnTest this should bring out any instabilities in a matter of minutes. (or a hour)