Gigabyte GA-X58A-UD3R & Core i7 980X - RAM Problem

[SV5FRZ]

Hi everyone,
I've been reading several posts on this problem - in this forum and elsewhere on the internet, and I'd like to add my specific problem to see if there's anything that could be done before RMA.

Here is a link to the Maximum PC forum where I started discussing this problem, I would be happy to fill in anything that's been left out.

http://www.maximumpc.com/forums/viewtopic.php?f=36&t=111403

Thanks for the help.

[Dark Mantis]

Hi and welcome.

It would be a great help if you could also post a cut down version of your problem here on the forum so people can find it when doing a search for the subject.

I would advise checking for bent pins in the CPU socket firstly.

[Aussie Allan]

  Just to add what DM said....

  If he stated in the build he wanted full CAD support.......the builder did the right thing with the GTX-480.......minimum for FULL CUDA/MERCURY/PHYSIC-X support.....and 1536 mem excellent for rendering/redraw (3D)

  What makes you think The second memory channel is fried.? ?

  check it is in fact a 64 bit install..........32bit only support 3.3/3.6Gb of mem

  Pins....DM.....good chance

 Test for a dead stick.....memtest......one at a time........slot one ... you said...yup tested mem!.....but did you test 1 at a time

  I don't think this memory comes in a Triple channel kit......is it possible the sys builder just chucked in 3 sticks he had on the floor ?

 Adata Memory..... High failure rate out of the box

 Aussie Allan

[SV5FRZ]

- Yes, I tested all the sticks one at a time. I tested them in each slot aswell. They all work fine in both Channel 1 and Channel 3 slots.
As soon as I plug ANYTHING in either slot of Channel 2, it's over.
For example: I grabbed stick 1, and I plugged in in A1 and tested. Then A2, then B1, B2, C1, C2. Each time running MemTest and going in Windows. And the same thing for the other two sticks.
That took a looong time I have to admit, but I did it.

- I was never able to get memory errors in Memtest on this PC. It either works perfectly or the BIOS doesn't let it boot.
("Recovering Lost DRAM Size" and then it reboots)

- It is in fact 64bit windows and it does use all 4Gb available to it right now.

- I thoroughly photographed the CPU socket and all RAM slots with my macro lens and they are all perfect.

- Perhaps the builder just grabbed 3 individual sticks and tossed them in there... I have no way of knowing.

- I have tried all the available BIOS versions for this Revision 1.0 board and nothing changes.

[SV5FRZ]

Here I will post a synopsis of the original thread for SEO purposes.

Post 1
-------
Hi all,
A customer of mine bought a new PC from a well known retailer, and it has a couple of problems.
It's not an OEM PC, the retailer custom builds computers using retail parts.
The PC is used mainly for AutoCAD, and the usual office use (word, excel, firefox etc.)

Configuration:
Gigabyte GA-X58A-UD3R  Rev 1.0
Intel Core i7-980X
6Gb Triple channel DDR3-1333 kit - AData AD3U1333C2G9-B
Corsair Nova V128 SSD
Palit GTX480 1536M
Premier SL-X850EPS PSU
1Tb Seagate HDD (i don't remember the model)
Samsung BD-ROM drive
Coolermaster Cosmos Pure
Windows 7 Ultimate x64

A few notes before I move forward.
1. Poor workmanship in the installation of the parts.
2. That's no PSU for a PC like this. The same goes for the RAM.
3. GTX480 for AutoCAD?
4. They had the hard drive controller set to IDE and not AHCI... do they know nothing about SSDs? Obviously they Ghosted the windows install...

Moving right along.
I fixed the IDE problem right way. I changed the registry bit to 0, changed to AHCI and rebooted, it's all fine now.
The main problem that I noticed right away was that only 4Gb of the RAM were usable. I took a picture from the BIOS for you to see.
Even in Windows 7, it lists "6Gb (4Gb available)".



All that was with the F5 BIOS installed. So I tried flashing to F6 and F7g.
And here comes the second problem. The flash went smooth both times, but as soon as it tries booting into Windows, it shuts off, and back on by itself.
I'm guessing this issue has something to do with the way they installed Windows on this machine, but there it is.

Recap:
1. Only 4Gb of RAM available.
2. Cannot boot to windows with BIOS newer than F5.


Thanks for the help.

Edit:
The F6 and F7g updates do NOT fix the RAM problem.
The above screenshot was actually taken from the F7g BIOS. But nothing changes between versions anyway.
-----------------------------------------------------------------------------------------------------------------------------------------------------


Post 2
--------
Sorry I forgot to mention this, but I was perfectly able to get from F5 to F6 to F7g and back and forth at will... that was not a problem at all.
The computer is currently working just fine with 4Gb available RAM and the F5 BIOS.

I was just trying to test possible solutions with the newer versions of the BIOS.

Thanks for the response though.
---------------------------------------------------------------------------------------------------------------------------------------------------


Post 3
--------
I haven't tried the CPU reseat or different RAM.
I did run memtest which ran just fine.

What's even more weird though is that it tells you that it does have 6Gb, but that it can only use 4...
Anyway, I'll give that a try and let you know.

I'll use the macro of my camera to take very close shots of the socket for inspection.

Another thought. What's a good liquid to use for wiping the contacts under the CPU? I've been using alcohol so far.
---------------------------------------------------------------------------------------------------------------------------------------------


Post 4
--------
I've finally found the source of this problem. The second memory channel is fried.
I was at the customer's yesterday and I tried a few things.

Note: The ram was in fact installed on the white slots. I'd checked that way back.

First, I tried the CPU thing. I removed the i7, I took very close pictures of the socket and I inspected it - everything was perfect. So I cleaned the old thermal paste off it and the heatsink, I applied some Arctic Silver Ceramic and I reseated the CPU as carefully as possible.
Meanwhile I thought I'd check and reseat the RAM sticks so I did that aswell before I plugged it back in.

Neither of these made any difference, and that's when I started checking different possibilities.

I upgraded to BIOS F7g and I started working with just one stick of RAM. Everything was fine. Windows worked perfectly. I tried all three sticks of RAM each one by itself in slot 1, and they all worked fine. So I knew then that the RAM was not defective.

Then, I thought I'd try Dual and Triple channel configurations and see how it goes. First I tried 1 and 3 for Dual channel and it started rebooting again. I tried 1 and 5 and it worked fine in Dual channel. I tried 3 and 5, and again, it rebooted.
I tried installing just one stick, any one of the three, on either slot of the second channel and it wouldn't work at all.

Now, I can't decide whether it's the CPU or the motherboard that has the problem. The CPU has the controller but the motherboard provides connectivity to the sticks... I took close pictures of the DRAM slots, they're not damaged. The only way to check for sure is to take the CPU and try it on another motherboard, or another CPU on this motherboard. Problem is I don't have one to check. I'll let the people who built it honor their warranty and fix it.

Note:
I noticed something that I hadn't noticed previously although I'm sure it was there. Starting at version F6 and F7g, there is a better self-check mechanism in place that actually checks the RAM further, and seeing as the second memory channel is defective, it wouldn't let the machine boot into any OS. Obviously that check must not have been in place at BIOS F5, so it would let the OS boot.
There was a message: "Recovering Lost DRAM size..." right after "Verifying DMI Pool Data", and after that, the PC would actually shut down for 2 seconds and then back ON again.


Right now, the computer is working with 2 sticks installed on 1 and 5, with BIOS F7g. It works fine at Dual channel until the guy can contact the company and get it replaced.

Thank you all for your help.
---------------------------------------------------------------------------------------------------------------------------------------------------


Post 5
-------
Hi guys, I'm back on this topic for some more help.

Given the fact that this motherboard's 2nd memory channel is fried, the customer decided to buy another pair of memory modules to occupy the remaining 2 slots (1 slot in channel 1 and 1 in channel 3). Note that the new pair is identical to the existing. (2x2Gb)
I plugged in the new pair, but the computer wouldn't work, it had the same exact problem as before (when we plugged anything in channel 2). It would only recognize the existing pair, for a total of 4Gb RAM.
I tried the existing pair in either slot of each channel (1 and 3) and they work fine, so I know there's nothing wrong with those ports. It's just when there's more than 1 pair of sticks that we have a problem.

The customer hasn't returned the motherboard to the retailer for replacement due to the fact that this computer is essential for the operation of this business and cannot be absent AT ALL. We live on an island so the PC would have to be shipped to the city for service (which means a week downtime).

I'm continuing my search for a possible cause and solution to this problem, I'm just wondering if noone has come across this one before.

Thanks again for all your help.
--------------------------------------------------------------------------------------------------------------------------------------------------------


Post 6
--------
I haven't made myself clear as to what works and what doesn't.

Here's a little text drawing:
The ports of the motherboard are these from left to right: A2 A1 - B2 B1 - C2 C1

So here are the combinations that work:
1. A1 - 2Gb (single stick)
2. A2 - 2Gb (single stick)
3. C1 - 2Gb (single stick)
4. C2 - 2Gb (single stick)
5. A1 C1 - 4Gb (two sticks) <---- this is what he's been using for a while now (and still is)
6. A2 C2 - 4Gb (two sticks)

And these are the combinations that do NOT work:
1. B1 - 2Gb (single stick)
2. B2 - 2Gb (single stick)
3. A1 B1 C1 - 6Gb (three sticks)
4. A2 B2 C2 - 6Gb (three sticks)
5. A1 A2 C1 C2 - 8Gb (four sticks)
6. A1 A2 B1 B2 - 8Gb (four sticks)
7. B1 B2 C1 C2 - 8Gb (four sticks)
8. A1 B1 - 4Gb (two sticks)
9. A2 B2 - 4Gb (two sticks)
10. B1 C1 - 4Gb (two sticks)
11. B2 C2 - 4Gb (two sticks)
------------------------------------------------------------------------------------------------------------------------------------------------



Post 7
--------
The screenshot is wrong... part of the fault in this board.

You'll notice in the screenshot that it accurately lists the placement in "DIMM Slot" 1 3 5 (A1, B1, C1) but then under it, it lists slot 4 as enabled. Slot 4 (B2) didn't have anything plugged.
I was initially trying to occupy A1, B1, C1 (strangely they have the slots backwards on this board - A2 A1 - B2 B1 - C2 C1)
The BIOS lists RAM slots 1 2 3 4 5 6 and they actually correspond to A1 A2 B1 B2 C1 C2. So you can't visually match the physical placement of the sticks to that list in the BIOS.

I NEVER tried to run 4 sticks in Triple Channel. I tried running 4 sticks in Dual Channel. Due to the fact that channel 2 is dead, I figured I'd occupy all of channel 1 and all of channel 3 (A1 A2, C1 C2).

The ports are clean and unbent, that was my first checkpoint.
I removed the CPU, photographed the socket with my macro lens, and all the pins are just fine.

I was reading online about a little over-voltage trick that's supposed to fix this problem even if the voltage is set to default afterwards.
I've never had that even suggested before and I'm a little skeptical.
-------------------------------------------------------------------------------------------------------------------------------------------



Post 8
--------
Yes, it really is Rev 1.0 and not 2. I have a photo of the bottom left corner on file.

Nope, the ram sticks are 1.5V, there's nothing to suggest they need more voltage.

I do understand what the manual says about the possible memory configurations and believe me, I have tried them all. But as soon as a stick touches either of the banks of the middle channel, it's over. It doesn't work.
Right now the computer runs fine at Dual Channel mode (which I verified with CPU-z) with a pair of sticks on slots A1 and C1. Contrary to what the manual lists, if you use just 1 pair of sticks, it does run at dual channel between channels 1 and 3. Perhaps the fault in this board makes it happen? I can't know for sure. As soon as I introduce a second pair of sticks to double the capacity (but still use just channel 1 and 3) the board doesn't like it.

As far as RMA goes, I've tried to find a way to get an advance return but they don't do that here. I'm afraid that if there's no way to fix this board where it stands, it will never be fixed. Sometimes simple facts determine our course of action and in this case, the fact is that this computer cannot be absent from this business AT ALL. The most I have is an hour after business hours to do a simple swap, that's it.
-------------------------------------------------------------------------------------------------------------------------------------------

[Dark Mantis]

It is quite possible that the memory modules are not matched and not great RAM anyway. How about posting the photograph of the perfect pins ? It could also be a duff CPU memory controller. Normally if the pins are ok the problem lies with the CPU not the board.

Another thing to check is your MaxMem setting.
MSConfig maximum memory setting

Start/run/msconfig/boot/advanced options/maximum memory

[Aussie Allan]

 Agreed and Agreed!.......still could be slot 2 (Board) but looking less and less likely

  Aussie Allan

[SV5FRZ]

I had deleted the pin photos, I'll go and take new ones as soon as I can.

MaxMem is unchecked. Besides, Windows lists the same reduced RAM size that the BIOS does.
(And that's in the older BIOS versions that weren't "smart" enough to stop the boot process and reboot. In the newer versions I can't even get in Windows to see the reduced size)

[Tinker]

It's normal for the later bios to recycle if a memory problem exists trying to find  missing
memory the one I tried was beta. That was an EX58. Used same memory on X58A & no
problems memory wise. Win 7 is misleading at times as it can report memory size even
if not working.


Regards tinker.

[Dark Mantis]

Win 7 is misleading at times as it can report memory size even
if not working.

Yes, I guess that is because it just reads the SPD.

[SV5FRZ]

Hi again guys,
I am dumbfounded at the outcome of this problem.

This PC suddenly works just fine!
I really don't know why, the only thing that changed is one version of BIOS.

I went to the customer's office to fix something else and I figured I'd take some photos like we discussed. I didn't have my camera with me so I used the customer's camera which unfortunately didn't have a macro lens. But that's really irrelevant at this point because I tried the triple channel once more, just for the hell of it and IT WORKED!

Just for reference:  DO NOT trust the manual. Good God, half the RAM settings in that manual didn't work at all. I had to go through trial and error to see what actually works.

Here's a photo of what we had up until now:  (relax, they were NOT unclipped when they were in use)


And these are the trials that worked:




Look at the 8Gb configuration. In the manual it says we MUST use 1 2 3 5 in order to work triple channel with 4 sticks. But that's just crap, the only configuration that works is the one you see. 1 3 5 6

And the 1 2 3 4 config they have for dual channel - 4 sticks, that doesn't work either.

Either way, I am NOT going to touch that PC again (I wanted to take those photos of the CPU pins) as long as it works. I won't try to fix what's not broken anymore.

Thanks again for all your help, I hope we can make sense of this somehow.
Tim

[Dark Mantis]

Hi again guys,
I am dumbfounded at the outcome of this problem.

This PC suddenly works just fine!
I really don't know why, the only thing that changed is one version of BIOS.

I went to the customer's office to fix something else and I figured I'd take some photos like we discussed. I didn't have my camera with me so I used the customer's camera which unfortunately didn't have a macro lens. But that's really irrelevant at this point because I tried the triple channel once more, just for the hell of it and IT WORKED!

Just for reference:  DO NOT trust the manual. Good God, half the RAM settings in that manual didn't work at all. I had to go through trial and error to see what actually works.

Here's a photo of what we had up until now:  (relax, they were NOT unclipped when they were in use)


And these are the trials that worked:




Look at the 8Gb configuration. In the manual it says we MUST use 1 2 3 5 in order to work triple channel with 4 sticks. But that's just crap, the only configuration that works is the one you see. 1 3 5 6

And the 1 2 3 4 config they have for dual channel - 4 sticks, that doesn't work either.

Either way, I am NOT going to touch that PC again (I wanted to take those photos of the CPU pins) as long as it works. I won't try to fix what's not broken anymore.

Thanks again for all your help, I hope we can make sense of this somehow.
Tim

Well great to hear that it is finally working but it is not good news to hear that the manual couldn't be trusted. This however is not the first time I have found errors in the manual but I suppose it was only written by humans after all 

Shame your pictures didn't display properly.

[SV5FRZ]

dammit, my photos didn't work.
I'll get to it in a sec

[Dark Mantis]

Well I am glad to hear that it is all workiong as it should now but it isn't the first time that I have found errors in the manual. I guess it was only written by humans after all! 

[SV5FRZ]

Something's wrong with this topic.
I was in the middle of editing my post when you replied to it and in so doing, locked it.
When I hit Save, my post disappeared.

Is there something you can do to fix it please?

Thanks