Gigabyte MA770-UD3 not booting with all RAM banks filled

TigerhawkT3

Reputable
Feb 22, 2014
18
0
4,510
I have an old motherboard (GA-MA770-UD3) that has worked fine for years with 2x2GB Gskill RAM modules. The CPU socket is on the left and the RAM slots are on the right. The two modules have been in the right-most slots (farthest from the CPU).

I recently bought some new generic Hynix RAM with 4x4GB modules. To make a long story short, all four modules are functional, but the PC only boots when there is RAM only in the right-most slots. If I also install modules in the RAM bank on the left, closer to the CPU, I get endless restarts and Windows BSODs with messages like "PFN_LIST_CORRUPT", "IRQL_NOT_LESS_OR_EQUAL", and "DRIVER_IRQL_NOT_LESS_OR_EQUAL". Leaving that bank empty would mean a maximum of 8GB of RAM rather than 16GB, as well as causing a pair of apparently perfectly good RAM sticks to sit in a drawer.

I have seen some threads here that suggest unseating and reseating the CPU cooler as a solution for this, but I'm worried about causing more problems (messing up the thermal paste or something) if I do that, so I'd like to make sure that this isn't a BIOS/software problem before I start messing with that.
 
Worth a try, don't fully remove the cooler, just loosen then snug down screw tght. Can also try going into BIOS and raising the DRAM voltage + 0.05 and set the CPU/NB voltage to 1.2. If no joy, let us know the CPU, mobo and actual model # of the DRAM
 


Lifted and lowered the CPU cooler latch, locking it closed very gently. Didn't work.

Went into the BIOS and increased the RAM voltage from 1.920V to 1.960V. Didn't work.

Tried putting the old RAM modules into the left-most slots, with new RAM modules in the right-most slots (12GB is better than 8GB). Didn't work.

During one of these (don't recall which), the Windows boot/load process complained about having a missing or corrupt C:\Windows\System32\drivers\ibbus.sys and that I should get the installation disk for a repair. Well, my 8.1 DVD has no such file on it (unless it's packed into an archive of some kind), and I couldn't even manage to boot from it with all four RAM slots filled. I'm looking in that folder right now, and no such file is present. Googling that file name, it appears to be some third-party thing anyway, and I'm not gonna start tossing random .sys files from the Internet into System32.

Should I contact Gigabyte? (Edit: I looked into filing a tech support ticket, but that requires a registration process which takes me to a fake/insecure site, so I guess that's not happening.) (Edit again: I emailed their warranty service center. The board is way out of warranty, but hopefully they can direct me to tech support.)

CPU: AMD Phenom II X4 940 Black Edition
Mobo: Gigabyte MA770-UD3
Old RAM: G.SKILL 4GB (2 x 2GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Dual Channel Kit Desktop Memory Model F2-6400CL5D-4GBPQ
New RAM: Hynix 16GB 4x4GB PC2-6400 DDR2-800Mhz 240pin (no model number available; Newegg link in original post)
 


I was going to, but with the way the BIOS kept randomly freezing (requiring restart button), and with the fact that the CPU is currently working properly, I decided not to mess with it.

Just in case.
 


Just had a thought: those other RAM slots were unused and empty until today. Might a speck of dust on a pin be causing this problem?
 


I'll try reducing the RAM clock speed, then.



No, it's 64-bit. I used to be running a 32-bit version and Windows could only access 3.25GB, but then I moved to 64-bit and it could access the whole 4GB, and now I'm using 2x4GB and it is running fine with (and using) 8GB. It's just when I use 4x4GB that it crashes.

I have not tried 3x4GB, which would (if I'm not mistaken) take the whole thing into single-channel, for 400MHz.
 


Gigabyte's email reply told me to call them, which I did today. Surprisingly helpful! After maybe three minutes waiting on hold, I spoke to an actual technician who didn't even ask the usual tech support stuff (is it plugged in, reboot it again, etc.), and went straight to the mobo model, which RAM slots I had used, and troubleshooting.

He had me remove both RAM modules from the slots farthest from the CPU (channel 1), and plug one module in to a slot closer to the CPU (channel 0). I tried using that module in each of those slots, and, with that single module, Windows loaded just fine each time. He then explained that if a single one of those slots was working and the other wasn't, that would be a defective RAM slot. However, since each individual slot is working, then it's a problem with the RAM's controller, which is... the CPU.

So now I get to contact AMD.
 


I looked for such software and my Googling suggests that the only way to test for a bad memory controller is to replace the CPU. I hope AMD has a miracle up their sleeve for me, because if they don't then that means I can only "solve" this by building a new computer, as I can't find a NIB AM2/AM2+/AM3 CPU to replace mine with.
 
It may be time for a new computer. DDR2 memory is pretty obsolete (as will DDR3 be in not too long). I am still rockin' with a Phenom II 1045T 6 core processor in my Pro Tools audio recording rig, and it is a great CPU. You may be able to find a used one on eBay or Craigslist, depending on where you live.
 


I'm well aware that it's time for a new computer. Unfortunately, it's not in my budget right now. It's mildly irritating that I bought more RAM than will actually work in my machine, but it was $43 shipped, which was within my budget.

One little problem that has appeared with the new RAM is that the PC doesn't wake from sleep properly. It'll power up, but just hangs there while constantly accessing the HDD. I hit the hard reset button and it boots from scratch, loads Windows... then restores the previous session correctly. I guess it's trying to use that hybrid standby/hibernate mode or something.

Anyway, I called AMD, and - long story short - they told me I should call Gigabyte because it's definitely the mobo's fault and not the CPU's. I think it will simply be an 8GB machine with slow wake times.