Missing RAM??? (previously Dead motherboard/GPU)

[tuffjuff]

This is a fine way to start my weekend.

 

I was playing a game last night when my 980 Ti randomly displayed a black screen. I was on Skype at the time, and could continue talking for maybe 30-45 seconds before everything went out. Several hours of attempted resuscitation later and no dice. In the morning, I tried swapping my graphics card from the top PCIe 16x port on my Asus X99-A to the second slot. I was able to get display (previously, POST/the BIOS wouldn't even show up on my monitor through HDMI or DP) but I noticed in the BIOS that instead of the 8x the graphics card should run at in this port, it was running at 1x. I did try to swap out a Windforce 970 I have in a different system, but the HDMI cable I had available was too thick so it didn't fully plug in and I didn't get a display. Bringing that same card back to it's host system, I now have the same problem.

 

Is my motherboard dead, and is it taking graphics cards with it?

[AbstractView]

What a shitty experience but from what I can tell it can be a multitude of things but what I think the main problem is your PCI slot. You tried switching your 980Ti to another slot and that didn't work yet your computer boots up. You also switched the graphics card to the Windforce one, did you put this card in the same slot as the 980Ti? Have you put the 980Ti into another build to see if it works? Are you getting any post code errors on your motherboard?

[tuffjuff]

The system shows the POST/BIOS (and presumably, Windows) screens in the second PCIe slot down, but it's running at 1x instead of 8x (when I originally built the system, I thought the fit would be too tight with the Noctua on the CPU to fit the GPU in the top/main PCIe slot, and it ran at 8x in the second slot) so I didn't continue further. So the GPU doesn't seem fully dead - I just can't tell if it's the GPU causing the issue or the motherboard. The Windforce card was tried in both the top and second from top (total of 4) PCIe slots, but because of how the back plate is orientated, the HDMI port comes too close to the case metal to fully plug in an HDMI card. I was contemplating picking up another graphics card (that I may be able to use later) like a 750 Ti or something, but I don't want to unnecessarily kill a 750 Ti if that's what my motherboard is doing.

 

POST codes on the motherboard come back normal - that was the first thing I checked.

 

Because the 980 Ti is water cooled (AIO eVGA unit) it's been too big a hassle and I've had too little free time to try that but it was going to be one of my next steps.

[tuffjuff]

Update:

 

I found an old GeForce 210 I had lying around, and it doesn't work in that top PCIe slot, so my guess is the board is jacked up. I have a replacement coming in the morning. Now I just wonder if that killed the 970 I tried in that system. Haven't had time to put the 980 Ti in a different system, yet.

[AbstractView]

Well I'm glad you found something wrong. Please let me know what happens

[tuffjuff]

I installed the new motherboard, and initially the system didn't have video, but after some messing with the DP cable and turning the system off/on, I'm now writing the post from the system.

[tuffjuff]

I now have a new problem. Loading up Windows, I noticed that my 32GB of RAM was reading as 28GB. CPU-Z shows I'm in quad channel mode (4 8GB sticks) and I don't exactly know how one can have half a stick of RAM be working... now I'm completely lost. Could it be a messed up 5820k, now that a bunch of stuff is on-board the CPU? Could it be a faulty stick of RAM? This is the only DDR4 I own in desktop format, so I don't exactly have a way to test...

[-rascal-]

Does the UEFI (BIOS) recognize all 32 GB??

 

A few things to try (since you got a replacement motherboard).

 

1. Turn off your computer and re-install your RAM sticks (take them out and put them back in).

 

2. Check the CPU socket for any bent or damaged pins.

 

3. Check the CPU itself for any sings of damage -- particularly the under-side.

[tuffjuff]

Repeated the 5820k and RAM after checking things. No discernable difference  (still at 28GB), running memtest now at 81% with no errors.

[HrutkayMods]

Windows takes up some ram for system processes or graphics sometimes, it might be Windows, I have around 31GB useable of my 32GB I'd check the memory section of task manager to see if Windows is using 4GB of ram up, it should tell you if it is

 

edit: okay never mind you've apparently done that already, one of your modules might be damaged, swap them around and see if the 4gb one switches slots if it does you need a new module, if it doesn't it might be a bad slot? Or maybe a BIOS problem (aka you need to update it).

[tuffjuff]

What's odd is memtest only sees 28GB addressable RAM, which I assume is why it didn't through up any errors. Prior to all this, Windows showed the full 32GB. I wonder if this could be a processor problem, what with various controllers being moved to the CPU?

[HrutkayMods]

39 minutes ago, tuffjuff said:

What's odd is memtest only sees 28GB addressable RAM, which I assume is why it didn't through up any errors. Prior to all this, Windows showed the full 32GB. I wonder if this could be a processor problem, what with various controllers being moved to the CPU?

Maybe but it still might be a damaged module 

[tuffjuff]

Alright - sometime today I'm going to try swapping sticks out. It's a tad obnoxious with the Noctua CPU cooler and a water cooled video card right next to it all, but I have enough airflow where I can run the system without CPU fans for 5 minutes at a time to test it out. I'm reaaaally hoping it's the RAM.

[tuffjuff]

13 hours ago, -rascal- said:

Does the UEFI (BIOS) recognize all 32 GB??

 

A few things to try (since you got a replacement motherboard).

 

1. Turn off your computer and re-install your RAM sticks (take them out and put them back in).

 

2. Check the CPU socket for any bent or damaged pins.

 

3. Check the CPU itself for any sings of damage -- particularly the under-side.

 

What's confusing is that, when this first happened, the BIOS showed all 32GB, but after repeated restarts it now only shows 28GB. Memtest only recognized/addressed 28GB, too. Pins are pristine, as I'm on the replacement motherboard, although they've gotten so small over the years that if one was messed up (I visually checked, and they looked fine) it would probably wouldn't see it. CPU is clean as well.

[KWelz]

On my system I only see around 28GB in CAM and other monitoring software.  I am also running a 5820K   BIOS always shows the full 32.   Windows and other background apps seems to take up about 4Gb of Ram so take that as you will.  

[tuffjuff]

I don't expect 32GB to be free, but it should (and prevooudly, was) available for use by the OS, etc.

[Snadzies]

Maybe try booting with a single stick at a time and see if each stick shows the correct size in the bios and Windows.

[HrutkayMods]

22 hours ago, tuffjuff said:

What's odd is memtest only sees 28GB addressable RAM, which I assume is why it didn't through up any errors. Prior to all this, Windows showed the full 32GB. I wonder if this could be a processor problem, what with various controllers being moved to the CPU?

yes I didn't think of that, that'd be the best option

[tuffjuff]

Alright, last update (I hope!)

 

Received and installed the replacement RAM kit, and now I'm back up to 32GB. I'm guessing the motherboard, when it died, jacked up some of the RAM with it. Now that both are replaced, everything seems in order.