After Adding Cooler, PC Won't Post With 16gb But Does With 8gb

suntower

Distinguished
May 16, 2009
73
0
18,630
I have an ASUS Z78 mobo with an Intel i7 K2600 and 16GB of Mushkin DR3 1333 memory (4 sticks all matching.)

I finally got tired of the CPU fan noise so I swapped the stock CPU fan with a Coolermaster Hyper 212 EVO. This required me to remove the mobo from my Antec mini-tower. And it's been downhill since.

The replacement -seemed- to go smoothly, but when I powered up, I got no video. So I tried the old 'remove the battery' trick and only using 1 stick of RAM at a time. And here's where I'm at.

I have tried every combination of sticks. ANY 2 stick in the first 2 slots will boot with 8GB. But the moment I add a 3rd or 4th stick, it won't post. The DRAM memory light stays solid red. I tried hitting the MemOK button, but it reboots and then blinks endlessly... never finishes.

Now before I swapped fans, I had used the AI Tweaker and so the mobo was OC'd -slightly- but I hadn't paid attention to it because it was set to 'Auto'. I thought that maybe this was causing the problem so I've tried resetting the BIOS (with 2 sticks) to 'Defaults' and 'Easy' but nothing seems to help. The moment I add back a 3rd stick the DRAM light comes on and it won't POST.

So... It looks as though the RAM is OK, but -something- happened. How do I troubleshoot this? Could I have somehow messed up the mobo in such a specific way? Or is the overclocking still somehow affecting things?

Help!

 
The CPU may not be making proper contact with some pins. Loosen CPU cooler mounting screws and then make sure the CPU and it's cooler are mounted evenly, and don't over-tighten mounting screws.

If problems persist, remove CPU and check for damaged or missing pins, and check that there is no debris in the socket or on the CPU.
 
Nope. Next? Remove and re-seat the mobo?

It's just weird that -everything- else works. I have several PCI cards and USB devices and they all work fine. Just only at 8gb.



 


I got all 4 DRAM sticks at the same time when I bought the parts to assemble (it was like 3 years ago now.) They are all identical sticks.

I've never adjusted DRAM voltages before (knowingly)... although I've seen lots of posts about this. I even saw one from a guy who had a similar problem and the answer was to reset the DRAM voltages to more 'conservative' numbers, but I don't know what that means. I thought that by using the AI-Tweaker Auto-OC I may have inadvertently changed these settings in some way and now I can't get it back to the true defaults.

Can you point me to a step-by-step with settings for changing these to more 'conservative' values?

Mushkin enhanced Silverline 996770 (2x4gb) PC3-10666 9-9-9-24 1.5vdc

Thanks in advance.
 
Is your power supply adequate?
Did you plug the new cooler fan into the FAN connector and not another one?
Have you checked the CPU temps to see that the cooler is working properly?

there is a piece of clear plastic on the bottom of the cooler to protect it but it needs
to be removed before placing it on the CPU. If you don't the CPU will run very hot and
you will get weird errors such as you have.

You can download CPU-Z and it will tell you a lot.
 

Well, it worked fine for 3 years.



Yes



88 degrees F. Seems fine.



Did that.



What specifically am I looking for?

 


I can't see how to change these values. Although they are currently set to 9-9-9-24 2 and the DRAM is 1.5.

Where do I change these settings?

 
Frustrating update: I removed the mobo and re-assembled EXTRA carefully from scratch. Same results.

Does this lead one to think the mobo got damaged somehow?

I'd like to think it's just a memory configuration issue, but frankly, I never had to adjust -anything- with the memory when I first assembled this PC. I just plugged everything in and the BIOS seemed to automatically sense everything it needed and I was good to go.

Ideas?

The frustrating thing is that if it really -is- a memory config issue, I wish the BIOS would at least show some message on video so I can adjust it. All I get is the solid red DRAM error LED.
 
1. The manual cover says P8Z68-V LX

2. Basically, putting -anything- in either slot 3 or 4 kills it. IOW: I can use any of the 4 DIMMS in slot 1 or slot 1 & slot 2. But putting any of the 4 DIMMS in either slot 3 or 4... whether there is a DIMM in slot 1 or 2 or not... makes the red DRAM LED light up and no POST.

It's almost as though there is some switch that shorts out the mobo the moment anything touches slot 3 or 4.

I don't mind replacing if necessary, but I've never had a mobo misbehave so -specifically- before. IOW: whenever I've had a problem, there are random nasty things. As I said, this seems so 'programmatic'. Like I flipped some switch somewheres.

OR... it could be completely random. 😀






 


Sadly, no. But as I wrote before, I've tried (literally) every combination of DIMMs in all 4 slots and the results are always the same. So doesn't that mean that all the DIMMs are functioning properly?

 
Check that there is no debris within nor damage to DIMM slots 3 and 4.

Slots 3 and 4 may be faulty (for whatever reason). But could try increasing DRAM voltage a bit as mentioned above. Refer to pages 2-13 and 2-16 of the motherboard manual, and change DRAM Voltage from Auto to Manual (?) and increase DRAM voltage a little (to around 1.55 V). If still have problems could increase DRAM voltage more, but don't exceed 1.65 V.
 
Loosened, re-tightened cooler.

Tried increasing DRAM voltage to 1.65. No dice.

I've dusted the slots so much I'd eat off 'em.

One thing which may not matter: there's a setting for overriding the DRAM Frequency. It has these unexpected numbers. Instead of 1333, 1600 as in the manual, the choices are like 1373, 1648, etc. Is that normal?

Any other ideas?





 
No damage to or any missing slot 3 and 4 pins? Check this closely.

EDIT: A remote possibility is that the CPU may be faulty. Can another CPU be used as a test?

The frequencies of memory such as 1333 and 1600 MHz are "nominal" values. Actual frequency the memory runs at is dependent on the actual base clock frequency and multipliers running on particular motherboards. So it is not unusual to see slight variation in "actual" memory frequencies.

At this stage it appears that memory slots 3 and 4 are faulty. So if wanting to use all available (4) memory slots, then it seems that a replacement motherboard will be needed.
 
I am very curious about your trouble and did some googling and found two answers on Toms that showed it was a defective CPU.
http://www.tomshardware.com/forum/286448-30-desperate-boot-boot
http://www.tomshardware.com/answers/id-1768227/computer-boot-ram-slots.html

HOWEVER you stated that everything was fine until you changed the CPU cooling fan.
So everything points to the CPU but we know it was working before you swapped fans so
that leads me to think that the CPU may be loose in its socket. Perhaps it did not lock in
place when you put on the new fan?? Or maybe you accidentally got some thermal compound
on the CPU pins so they are not working? It seems to me that something has caused the CPU
to not work properly. I would look at it carefully.
I hope you find it! Good luck!
 
I just got off the phone with ASUS and they referred me to another forum (Hexus) where other users have complained that this particular mobo has had issues with 'flex'... ie. the weight of the new cooler backplate may be bending the mobo just enough to cause a short. They suggested I try reverting back to the old cooler.

At this point? I am so fed up I'm thinking that I'll leave it as is and get 2 8GB sticks so I can get back to 16gb. The problem is that this machine is used in a recording studio with a TON of add-on cards, dongles, USB gizmos and software drivers and tweaking ANYTHING is asking for trouble. IOW: once I get a machine working I tend to LEAVE IT.
 
An update. I took the ASUS tech advice and removed the mobo and the Coolermaster backplate. And I think I see the issue.

There is a noticeable burr on one corner of the Coolermaster backplate which dug into the back of the mobo and appears to have nicked just deep enough to scratch 1 or 2 of the copper traces. It -may- be a red herring, but my totally WAG is that these traces are what connect DIMM slots 3/4 to the CPU.

Normally, I'd just chalk it up to me being careless, but the corner of the Coolermaster backplate is -clearly- burred. My guess is that I didn't notice it because the side you see (which doesn't contact the mobo) is perfectly smooth.

I'm hoping it's such a clean 'incision' that nothing else is messed up. If that's the case, I'll just suck it up, buy 2 8gb sticks and move on. In the back of my mind, though, I'll always be wondering a little.

Also, I feel like Coolermaster should pay me for the cost of 2 new sticks, but I'm sure there's little chance of -that- happening. (sigh).
 
I took photos and called Coolermaster, figuring I could start a conversation. And here is their policy:

1. Create an on-line account.
2. Fill out an RMA with a short description.
3. If they decide it's worth pursuing, they will issue me an RMA.
4. I then send them both the Cooler -and- the motherboard.
5. They will examine and file it as an 'insurance claim'. The guy made it clear that I would need to sign a form indicating that they are the arbiter in all cases.
6. If they determine I have a case, then they would replace the cooler and reimburse me for the mobo, but it may not necessarily be returned. (Which I found unbelievable.)

I didn't expect to get anywhere really, but this is so bald faced crappy. I've filed insurance claims before and you simply send them pictures. You don't send them the -gizmo- itself... which causes us downtime. They're -trying- to make it total PITA.