GA-Z68X-UD3H-B3 Rev 1.3 RAM woes

bernjc3

Honorable
Mar 13, 2012
10
0
10,510
So I recently put together a new system (probably my 10th system build or so) and I have to say that I have never had this many memory problems before. First for reference purposes:

Motherboard: GA-Z68X-UD3H-B3 rev 1.3 (Bios: F12)
RAM: G.SKILL Value Series (2 x 4GB) DDR3 1333 (PC3 10600) 9-9-9-24 at 1.5V
CPU: Intel i5-2500k
PSU: CORSAIR CX600 V2 600W ATX12V v2.3 80 PLUS Certified

So I started with 2x4GB of the above mentioned memory and had problems. Found out the one of the modules was bad after troubleshooting. Order another set of 2x4gb and RMA'd once the new set arrived. New set had 1 bad stick as well. New pack arrives after RMA, both sticks are fine, RMA for 2nd pair sent out. Noteworthy: I was able to boot with 16gb of RAM but had errors from the bad sticks, but it would boot.

Fast forward to this weekend, the latest RMA arrives (4th pack of memory if you've been counting). I install the new RAM for a total of 16gb and the system will not post just 2sec. boot loops.

So this week will be trouble shooting week here's what I'm thinking and have done. At first I thought that the new modules were just completely bad because I had been able to at least boot with errors with 16gb before. However after reading around a bit, it seems that I may need to adjust the voltages to accomodate the 4 sticks.

Known
1. I have 2 known good RAM modules
2. I have been able to boot with errors with 16gb of RAM before
3. Trying to boot now with new modules (16gb total) causes 2sec boot loop with no beep or post.

Possible Problems
1. New modules are dead
2. Bent CPU/MB pin causing memory controller problems
3. Voltage too low/timings too high for 4x4gb DDR3 (all memory is ddr3 1333 9-9-9-24 1.5v both from double 4gb packs)
4. 3/4 DDR slots are bad

Troubleshooting Ideas
1. Test 2 new modules each alone in first memory slot to ensure they are working.
2. Try using 2 known good RAM modules in DDR slots 3 & 4 to see if they are bad. (MB manual recommends using slots 1&2 or 3&4 order goes 1_3|2_4 on MB layout)
3. Up voltage to 1.6v for memory slots and attempt 3x4gb then 4x4gb
4. If new modules are confirmed to be working and above steps don't solve issue, assume bad memory controller or bad slots 3 & 4 on MB.

I've updated the BIOS from F10 to F12 but haven't tested with it yet. Today I'm planning on starting the troubleshooting ideas above to try and figure this out.

Sorry for the long post but wanted to get as much information as possible in here to try and figure this out.

Any ideas?

*Edit* Added PSU
 
Good info, and your next proposed steps are a good line of thinking. Except, you do not need to up your Voltage to accommodate 4 modules. If you do have to do this, it's because of a fault and you're merely band-aiding the situation.

Your first next-step above is probably going to be crucial and telling. If that fails, then you know you have yet again another bad set of RAM. Unfortunately this would not surprise me as GSkill doesn't exactly have the best QA record in the world - personally I think they're crap (I know I'll probably get flamed for saying that).
Step 2 above is a good step to verify that the board is ok (which there is a significant possibility the board is bad - not because of brand, but because mobos are complex and run a significant probability of mfg defects.)

One other additional comment, throw memtest into this testing cycle. Use it to validate the RAM is good, instead of testing "if the system will boot". Memtest is very thorough and telling for RAM errors. One side note, unfortunately memtest can take a few passes to expose small errors - but 1 RAM error is 1 error too many and must be fixed (replaced).

You can pull the CPU to insure there aren't any bent pins, but I would expect errors all over the place if this happened. I would suspect RAM then mobo as potential problems and then very unlikely the CPU is at fault.

Again, don't increase the voltage for this testing cycle. It's for overclocking, which I don't think you're wanting to do.
And if you still need help, can you reply (or edit your OP) with the details of your PSU?
 


NP, bernjc3. Sounds like your plan of attack is appropriate. Just one more thing jumps to mind - do you have a massive GPU (or GPUs) in this system? If you went crazy SLI or something like that, I could see the PSU being a little too small. But I think I'm grasping at straws with this last comment. :)
 


Running a single Radeon 6870, 1 SATA HD, 1 SATA DVD-RW, 4x case fans so I'm 99% sure it's not for a lack of power...
 


After this response, I agree too. :) Just a topic that wasn't covered in the OP.
 
So I tested the new RAM modules and each passed with no errors. I tested each slot individually with 1 stick of RAM and all worked fine. I tested pairs in slots 1 & 2 | 3 & 4 and they worked fine.

Initially add a 3rd stick of RAM caused the system to not turn on. The fans spin up for about 1 sec. and then the it would shut off and do this over and over again. I removed the 3rd module and upped the voltage in BIOS to 1.55V and left the VTT at 1.05V. In could then boot with 3 modules fine.

I added a 4th and had the same boot looping issue (1 sec. fans spin up then shut off, on and on). I then tried raising the voltage to 1.65V and the VTT to 1.15V with no luck. I'm at loss for what to do now, so I think I'll end up just putting the extra 2 sticks of RAM in the extra PC parts draw and chalk it up as loss.

I'm thinking that my mother just isn't providing enough power to the memory slots when all 4 modules are installed...

Sidenote: The bottom clip that holds the memory in on the 4th slot broke off at the end of all this testing :fou:
 
I doubt that the board isn't providing enough power. The controller is spec'd to be able to load 32GB of RAM @ 1.5v. And if you're using LP DRAM modules, it can do this at 1.35v.
You may in fact have a motherboard problem, but power (V) isn't the problem.

I would make sure that you let memtest run for more than 1 pass. Sometimes it needs up to 10 - 16hrs of testing (several passes) to expose errors. If you can verify that the RAM is OK, then my next suggestion would be to replace the mobo.

(with the mobo out of the case, I'm sure you can get the clip reconnected to the socket well enough to pass an RMA)
 


I think the memory is compatible. I think that link just shows how low quality GSKill is. None of GSKill's standard 1333 memory lists the z68 - I think they just haven't bothered to update their website. The z68 is more than capable of running 1066, 1333, 1600. As well, Asus does list a very close cousin of this part in their compatibility list.
So, I don't think it's as "simple" as that weblink.

I do think it's as simple as not buying GSKill though. :) But unfortunately for our friend, I do think his motherboard is in question.
 
Very possible it's the mobo... been there. It's just that ram issues (the one's I've experienced) are so difficult to track down (especially those actually on the mobo but affect ram) I've gotten in the habit of checking compatibility first (one less possible issue). As far as brand, G.Skill is no worse or superior to any other mainstream brand although I am not familiar with their econo line. All three of my computers are running G.Skill, all perform as expected - all purchased from QVL or QML.
 
Buying from a QVL is always ideal, and if you get explicit support from the RAM mfg AND the mobo mfg, then you're best off.
Bernjc3 does have a sudo-QVL from Asus supporting these modules. They're not the exact same SKU, but the letters at the end are more for packaging purposes, not technical. And while I don't really want to get into a debate over quality, GSkill does not have the best quality track record, especially when it comes to their value line.

At the end of the day, if you get the modules running and burned in for 30 days without a problem, you're not going to have a problem with them for at least 10yrs. Those first 30-days and those "hard to track" RAM problems are the reasons you buy from higher quality RAM mfgrs (or at least buy GSkills higher quality lines, e.g., Ripjaws)