MSI DRAM error light and moving RAM slots

[Kongar]

Running into an issue on an MSI board and my google fu is failing me. For no reason, the PC wouldn't post, and the DRAM error light came on. I googled, and found an obscure post on a forum in the back corner of the internet where some dude said "oh - this is an issue with MSI boards. The solution is stupid simple, but weird. You just move the ram to different DIMM slots and reboot. The mobo unscrews itself up, and you can then move them back to the optimal slots."

I think to myself - "no way" but I'll be dammed - it worked. So I ran a while with the RAM in the non optimized slots, the mobo complained on boot "you're in the non optimized slots" but it ran. THEN, it errored out again, and now no moving around of ram modules works.

Is there a trick here I'm missing, before I start ripping things apart and doing the part swap "does this vid card work, does this other ram work, does this swapped part work, etc."

Such a pain

 

[pitingres]

I'd guess that either the RAM is on the way out and needs replaced, or dirty contacts are involved. Try taking a (clean!) eraser to the dimm contacts, and/or using a good contact cleaner.

 

[grumperfish]

If cleaning the contacts works you may want to run at least two passes with a memtest USB install to confirm that the modules are not reporting any errors: https://www.memtest.org/
Also verify that you are running the RAM at the correct timings and voltages in case the system is detecting them incorrectly.

 

[Kongar]

If cleaning the contacts works you may want to run at least two passes with a memtest USB install to confirm that the modules are not reporting any errors: https://www.memtest.org/
Also verify that you are running the RAM at the correct timings and voltages in case the system is detecting them incorrectly.

I did try cleaning the contacts - that was my first thought as to why moving them "worked" (it didn't have anything to do with the slot and more with making a connection after physically reseating them) No dice. Can't even get into the BIOS to check the timings - never even comes alive. But ya they've been good for about a year and a half.

Strange problem. I'm just going to dismantle my son's PC and start swapping parts until I find the problem area. Doesn't look like the RAM - my money is on the mobo, but we'll see. No avoiding the work anymore I guess...

 

[Furious_Styles]

I did try cleaning the contacts - that was my first thought as to why moving them "worked" (it didn't have anything to do with the slot and more with making a connection after physically reseating them) No dice. Can't even get into the BIOS to check the timings - never even comes alive. But ya they've been good for about a year and a half.

Strange problem. I'm just going to dismantle my son's PC and start swapping parts until I find the problem area. Doesn't look like the RAM - my money is on the mobo, but we'll see. No avoiding the work anymore I guess...

Make sure there are no bent pins in the CPU socket if it's an Intel board. But I agree it is likely the mobo if that memory works fine in your son's PC. Updating the bios can sometimes help, what board is it exactly?

 

[pendragon1]

what "msi board" are you talkin aboot?
if its 2yr+ check the cmos battery.

 

[kirbyrj]

Make sure there are no bent pins in the CPU socket if it's an Intel board. But I agree it is likely the mobo if that memory works fine in your son's PC. Updating the bios can sometimes help, what board is it exactly?

It definitely sounds like the motherboard. I would double check the memory in a different computer if you have the ability though. Or swap in memory from your computer just to see if it's the board or the ram.

 

[Macho]

LGA stands for LAND GRID ARRAY. If you bought a brand new LGA 1151 Socket, they are already balled with balled legs on the bottom.

I have watch quite a few people who attempted to remove their LGA Socket from the motherboard. They failed a few times de soldering the tiny capacitors
in the center of the LGA cut out. That has to be covered first to protect it. Tin foil around everything else. The whole motherboard has to be heated up to 100 C. in the oven.
Bottom heat is needed and not too much to melt the capacitors connected. 200 degrees C. is very hot to get it to be removed with an air gun. The plastic starts smoking.
Then it can be lifted off.

With copper wick to pull in the excess solder dipped in flux to wick off the solder. First scrape it off. They have tinning solder paste with a little solder mixed in.
The pads should be tinned first.

Watch YouTube for this issue of how they attempted but damaged their motherboard first. The botton of the CPU Socket is Ball Grid Array, and we lay the CPU on what must be considered the Land Grid Array. Pin Grid Array is when the CPU had pins decades ago.

A very hard job for an expert with http://www.GBAeasy.com machine or one simular, even one made by the expert as a D.I.Y. BGA machine.
Only $5.00 to buy a LGA 1151 already balled. If you have bent pins this would have been a good choice if it was possible to change. A heat shield has to cover what you do not want heated up, and about when the plastic starts to smoke that is the heat needed. nearly 500 degrees F. I just sent away for a IR heat temperature gun. Not going too hot is recommended. Also everyone will warn you that this is another fraternitiy other than SuiperUsers like US.

 

[pendragon1]

was all this bot rambling just to pimp a site?!
edit: Kongar , you ever get this sorted?!

 

[lopoetve]

LGA stands for LAND GRID ARRAY. If you bought a brand new LGA 1151 Socket, they are already balled with balled legs on the bottom.

I have watch quite a few people who attempted to remove their LGA Socket from the motherboard. They failed a few times de soldering the tiny capacitors
in the center of the LGA cut out. That has to be covered first to protect it. Tin foil around everything else. The whole motherboard has to be heated up to 100 C. in the oven.
Bottom heat is needed and not too much to melt the capacitors connected. 200 degrees C. is very hot to get it to be removed with an air gun. The plastic starts smoking.
Then it can be lifted off.

With copper wick to pull in the excess solder dipped in flux to wick off the solder. First scrape it off. They have tinning solder paste with a little solder mixed in.
The pads should be tinned first.

Watch YouTube for this issue of how they attempted but damaged their motherboard first. The botton of the CPU Socket is Ball Grid Array, and we lay the CPU on what must be considered the Land Grid Array. Pin Grid Array is when the CPU had pins decades ago.

A very hard job for an expert with http://www.GBAeasy.com machine or one simular, even one made by the expert as a D.I.Y. BGA machine.
Only $5.00 to buy a LGA 1151 already balled. If you have bent pins this would have been a good choice if it was possible to change. A heat shield has to cover what you do not want heated up, and about when the plastic starts to smoke that is the heat needed. nearly 500 degrees F. I just sent away for a IR heat temperature gun. Not going too hot is recommended. Also everyone will warn you that this is another fraternitiy other than SuiperUsers like US.

No one here has suggested replacing the socket.

 

[Denpepe]

was all this bot rambling just to pimp a site?!
edit: Kongar , you ever get this sorted?!

He seems top keep mentioning different sites, starting to think he wants to repair motherboards for a living from as little as I can make out of his ramblings.

 

[Macho]

I just started a few days ago. I see that this has been a problem for at least 2015. Some people say to determine which component needs to be re balled, to see if the solder joint is "cracked," you can heat it up so it expands. In a laptop, there is the graphics chip and the CPU. It only takes him several minutes to re-ball and get the laptop up and running.
What a nightmare, putting the cooling heat sink on a component and giving pressure, it will most likely work until you remove the heat sink and put the heat sink back on again. Now the cracked or fractured balled connection may fail within a few minutes. I see visually that there is a small space between the LGA CPU socket and the motherboard, so it is on a leggos. Maybe an authorized dealer retrofit would be some shims underneath.

First, never use irate servicemen to be sent out, and try not to work for anyone so irate as you guys. Fix it yourselves and have your own nightmare without anyone else who has to hear you.

Just putting liquid solder under the component and reflowing the solder would be a good fix. Re-balling is what can be done, then if you are capable you can.
The solder joint is not cracked at first until someone cracks it. Some of your games cause a GPU to go bad. It must be a crack first though. huh ?

You guys are some nightmare that goes with this problem happening.

 

[lopoetve]

I just started a few days ago. I see that this has been a problem for at least 2015. Some people say to determine which component needs to be re balled, to see if the solder joint is "cracked," you can heat it up so it expands. In a laptop, there is the graphics chip and the CPU. It only takes him several minutes to re-ball and get the laptop up and running.
What a nightmare, putting the cooling heat sink on a component and giving pressure, it will most likely work until you remove the heat sink and put the heat sink back on again. Now the cracked or fractured balled connection may fail within a few minutes. I see visually that there is a small space between the LGA CPU socket and the motherboard, so it is on a leggos. Maybe an authorized dealer retrofit would be some shims underneath.

First, never use irate servicemen to be sent out, and try not to work for anyone so irate as you guys. Fix it yourselves and have your own nightmare without anyone else who has to hear you.

Just putting liquid solder under the component and reflowing the solder would be a good fix. Re-balling is what can be done, then if you are capable you can.
The solder joint is not cracked at first until someone cracks it. Some of your games cause a GPU to go bad. It must be a crack first though. huh ?

You guys are some nightmare that goes with this problem happening.

Cells.

 

[michalrz]

Russ, do you like our owl?

 

[lopoetve]

I just started a few days ago. I see that this has been a problem for at least 2015.

No, it hasn't.

Some people say to determine which component needs to be re balled, to see if the solder joint is "cracked," you can heat it up so it expands.

No, you don't, and no, they don't.

In a laptop, there is the graphics chip and the CPU. It only takes him several minutes to re-ball and get the laptop up and running.

ROFL, no.

What a nightmare, putting the cooling heat sink on a component and giving pressure, it will most likely work until you remove the heat sink and put the heat sink back on again.

No. That's not even close to how this works. You're doing it wrong.

Now the cracked or fractured balled connection may fail within a few minutes.

No. I've swapped coolers on dozens and dozens of systems - heck, adding MEMORY to one of my systems requires removing a ~massive~ noctua HSF. Never had an issue.

I see visually that there is a small space between the LGA CPU socket and the motherboard, so it is on a leggos. Maybe an authorized dealer retrofit would be some shims underneath.

Shims would be a bad idea, showing clearly that you don't know what you're talking about.

First, never use irate servicemen to be sent out, and try not to work for anyone so irate as you guys. Fix it yourselves and have your own nightmare without anyone else who has to hear you.

No?

Just putting liquid solder under the component and reflowing the solder would be a good fix.

That would short out the entire socket and likely start a fire, if OCP didn't kick in fast enough.

Re-balling is what can be done, then if you are capable you can.

No, you really can't. Not at that level of detail, unless you do it professionally, and none of us are going to bother. Motherboards aren't THAT expensive.

The solder joint is not cracked at first until someone cracks it. Some of your games cause a GPU to go bad. It must be a crack first though. huh ?

No. And the game in question (new world) broke the GPUs a VERY different way. This has nothing to do with solder links failing.

You guys are some nightmare that goes with this problem happening.

I'm pretty sure you're a bot. Or you need to check your meds.

We don't have this issue. Really, ever. I've had TWO motherboards die in the last 10 years - one was an Intel Atom system with the Atom timing failure (not an issue with solder), and the other was basically DROPPED, was bought used, was already 9 years old when I got it, had used CPUs and memory in it, and wasn't in good shape to start with - and I still got 3 years of use out of the thing. For $50.

If what you were saying was true, we'd be replacing boards every few weeks. It isn't. I'm not sure what your background or experience is, but I strongly suspect you're a turing bot or a troll, and either way, you're SO far off base that we have to respond to make sure people don't try to follow your advice.

 

[Kongar]

was all this bot rambling just to pimp a site?!
edit: Kongar , you ever get this sorted?!

So in the end, it had nothing to do with the RAM or the motherboard. It was a failing/ed video card. I have no idea what failed, or why it lit up the DRAM LED error light and not the VGA light. But it's the video card. (powercolor 5700XT). Pulling the card allows the pc to boot. Putting in another video card allows the PC to boot as well. Putting my "bad" video card into my son's similar, but not the same, MSI mobo causes HIS mobo to light up the DRAM error light (go figure).

So unlucky me - failed video card when video cards cost 2k. Lucky me it was still under warranty, so I sent it in for an RMA. Unlucky me - Powercolor doesn't seem to be very responsive or fast - multiple threads here about how bad they are. So I'm hoping, they'll actually fix it and send it back sometime this century. If and when prices and availability come back to normal, I'll probably build a completely new rig from scratch (been upgrading this one for 13 years now) Time for something new and flashy - but not at today's prices.

 

[Furious_Styles]

So in the end, it had nothing to do with the RAM or the motherboard. It was a failing/ed video card. I have no idea what failed, or why it lit up the DRAM LED error light and not the VGA light. But it's the video card. (powercolor 5700XT). Pulling the card allows the pc to boot. Putting in another video card allows the PC to boot as well. Putting my "bad" video card into my son's similar, but not the same, MSI mobo causes HIS mobo to light up the DRAM error light (go figure).

So unlucky me - failed video card when video cards cost 2k. Lucky me it was still under warranty, so I sent it in for an RMA. Unlucky me - Powercolor doesn't seem to be very responsive or fast - multiple threads here about how bad they are. So I'm hoping, they'll actually fix it and send it back sometime this century. If and when prices and availability come back to normal, I'll probably build a completely new rig from scratch (been upgrading this one for 13 years now) Time for something new and flashy - but not at today's prices.

Wow that's a weird one. And why it always helps to have a set of spares around to help diagnose the problem.

 

[Kongar]

Wow that's a weird one. And why it always helps to have a set of spares around to help diagnose the problem.

yup - I always keep a computer+ of old components around for exactly this reason. The pain is ripping everything out of the case and playing "let's troubleshoot by changing parts one at a time" Not really my favorite game, but sometimes absolutely necessary - the error you get isn't always helpful in finding the true problem. I SUSPECT, that moving the RAM around the first time did absolutely nothing, and that it was a coincidence that lined up with an intermittently failing video card. Otherwise, I can't explain why I got it working again temporarily. Had to be a fluke that led me down an incorrect troubleshooting path. But ya - since the problem followed the card into a 2nd PC, and my PC works fine in any configuration without that card - it has to be the card...

 

[Furious_Styles]

yup - I always keep a computer+ of old components around for exactly this reason. The pain is ripping everything out of the case and playing "let's troubleshoot by changing parts one at a time" Not really my favorite game, but sometimes absolutely necessary - the error you get isn't always helpful in finding the true problem. I SUSPECT, that moving the RAM around the first time did absolutely nothing, and that it was a coincidence that lined up with an intermittently failing video card. Otherwise, I can't explain why I got it working again temporarily. Had to be a fluke that led me down an incorrect troubleshooting path. But ya - since the problem followed the card into a 2nd PC, and my PC works fine in any configuration without that card - it has to be the card...

Did you take your son's card and throw it in yours to see for sure it's not a PCI-e slot problem?

 

[Kongar]

Did you take your son's card and throw it in yours to see for sure it's not a PCI-e slot problem?

Yup - I did that too just to make sure it wasn't the slot. I did:

1. Run the PC with no video card (CPU vid) - works fine
2. Run the PC with my son's video card in the same slot - works fine
3. Run the PC with a really old video card in the same slot - works fine
4. Run the PC with functioning cards in my other slot - works fine
5. Run both motherboards, in any slot, with the bad card - PC does not work, and both MSI boards pitch ONLY the DRAM error LED

 

[Kongar]

You know, and for anyone stumbling across this in the future (since there seems to be a bot in here). I concur with lopoetve, I've had one mobo fail on me in almost 30 years of building PCs - it's not very common in my experience for them to fail. I've had more CPUs fail than mobos believe it or not (2 if you're wondering). But I can't count the number of times I've had RAM, hard drives, or video cards fail.

 

[Furious_Styles]

Yup - I did that too just to make sure it wasn't the slot. I did:

1. Run the PC with no video card (CPU vid) - works fine
2. Run the PC with my son's video card in the same slot - works fine
3. Run the PC with a really old video card in the same slot - works fine
4. Run the PC with functioning cards in my other slot - works fine
5. Run both motherboards, in any slot, with the bad card - PC does not work, and both MSI boards pitch ONLY the DRAM error LED

Well done, that makes me very confident it's a faulty gpu.

 

[grumperfish]

You know, and for anyone stumbling across this in the future (since there seems to be a bot in here). I concur with lopoetve, I've had one mobo fail on me in almost 30 years of building PCs - it's not very common in my experience for them to fail. I've had more CPUs fail than mobos believe it or not (2 if you're wondering). But I can't count the number of times I've had RAM, hard drives, or video cards fail.

The last motherboard I had that actually failed was a socket 775 board and I really had to go out of my way mangling a HSF installation for that to happen. I have never actually seen a failed CPU, but plenty of dead or dying power supplies.

 

[lopoetve]

So in the end, it had nothing to do with the RAM or the motherboard. It was a failing/ed video card. I have no idea what failed, or why it lit up the DRAM LED error light and not the VGA light. But it's the video card. (powercolor 5700XT). Pulling the card allows the pc to boot. Putting in another video card allows the PC to boot as well. Putting my "bad" video card into my son's similar, but not the same, MSI mobo causes HIS mobo to light up the DRAM error light (go figure).

So unlucky me - failed video card when video cards cost 2k. Lucky me it was still under warranty, so I sent it in for an RMA. Unlucky me - Powercolor doesn't seem to be very responsive or fast - multiple threads here about how bad they are. So I'm hoping, they'll actually fix it and send it back sometime this century. If and when prices and availability come back to normal, I'll probably build a completely new rig from scratch (been upgrading this one for 13 years now) Time for something new and flashy - but not at today's prices.

Oddly enough, my first 6800XT dying did the same. And half-killed my Zenith II Extreme Alpha.

 

[lopoetve]

You know, and for anyone stumbling across this in the future (since there seems to be a bot in here). I concur with lopoetve, I've had one mobo fail on me in almost 30 years of building PCs - it's not very common in my experience for them to fail. I've had more CPUs fail than mobos believe it or not (2 if you're wondering). But I can't count the number of times I've had RAM, hard drives, or video cards fail.

If I extend to my entire history building systems, I've had the following motherboards fail:
AMD 760 (first DDR board for Athlon systems) - after 11 years of use and no maintenance for the last 4 (was passed on to a family member). Just quietly died one day. Probably full of dog hair.
Supermicro dual socket somethignoranother - this was the dropped one
The Intel Atom timing bug
Early S754 Albatron board, with a MASSIVE zalman full-copper HSF that yes, warped the board eventually. That system was also carried to a half-dozen lan parties though, WITH that HSF in it... You know those big big suckers? Yeah. That's on me.
One pos compaq system I bought from costco (did some contract work for them) in the Slot-A days - someone had bought it, stripped parts, returned it, and they sold it to me for $100. I fixed it up and it ran for 6 years. They weren't kind taking the parts out either, so no concerns there.

I had supremely early failures with some NF3 boards from Gigabyte, but that was eventually tracked down to those being HUGE problems because they didn't follow spec on memory spacing/distance from socket.

I've had 3 CPUs fail, generally after YEARS of use, often overclocked (never within the first 5 years).