System details:

• MSI X470 Gaming Plus
  
• AMD Ryzen 7 3700x
  
• 32GB Corsair 3200MHz DDR4 RAM (2x 16GB)
  
• nVIDIA RTX 2070 Founders Edition
  
• Samsung 860 EVO 1TB SSD
  
• Corsair CX750M 750W modular PSU
  

This system has (barring brief experimentation with overclocking) been rock solid from when I built it in August until tonight. I finally got around to transplanting the guts from my old Antec Lanboy Air into the Rosewill rackmount server chassis (and its corresponding 4u of space in the rack) I have lying around, and this must have given the gremlins a chance to set in upon it. When a specific RAM configuration is used, it gives me 3 long beeps, which Google and the MSI forum say is a memory issue.

From what I can gather from the experimentation I did, it does not want to POST when the same DIMM slot on both channels is occupied (say, if I have slots 1A&B occupied, or slots 2A&B occupied). This means it's RAM, motherboard or CPU.

Both DIMMs check out individually in each of the 4 slots on the board, and I can get it to POST no problem with both DIMMs in, as long as it's not dual channel. I've cleared the CMOS 3-4x.

So, RAM is good, leaving the motherboard or CPU. I never removed the CPU from its socket, nor did I remove the heat sink, which makes me suspect the motherboard.

I *was* able to get it to run for a little while (maybe 15-20 minutes) in its original configuration after initially throwing the beep code at me by clearing the CMOS, but then it bluescreened on me with a system_service_exception error and would not come back up even after another CMOS clear.

My thoughts (in no particular order):

• CPU got nudged off kilter somehow when I pulled it out of the new case (the standoffs that came with the case were utter trash and I had to test fit it 3 or 4 times)
  
• I accidentally bumped something around the RAM slots while transplanting it
  
• Somehow I angered the ESD gods (memory controller or any IC that might be between it and the DIMM slots).
  

Is there something that you think I might have missed? It's up for the moment - I'm typing on it now, but I would really like to have my properly working system back (read: that extra 400MHz per channel; plus it kinda kills my plans to put another 32GB of RAM in it down the line).

Thanks in advance for any help you might be able to offer.

(Edit: forgot to mention PSU)

quote:

MSI X470 Gaming Plus

I think you did this, but from the manual:
"If 3 long beeps are heard, remove all
memory modules and try to install only
one memory module in the DIMMA2 slot
first and then restart the computer."

The only suggestion I have is to reseat the CPU and verify nothing is damaged.

You might find something here if you haven't came across it:
https://forum-en.msi.com/index.php?topic=305675.0

Since you did some overclocking, try setting the BIOS to default.

Also verify your RAM is on MSI's compatibility list:
https://www.msi.com/Motherboar...ING-PLUS#support-cpu

If you know someone that has compatible RAM you could try to swap their's in and see what happens.

Are you installing the memory modules both on the same dimm channel? The way you worded your memory configuration is a little confusing to me. For two sticks of ram, the fist stick should go in Dimm slot A2 and the second stick should then be in Dimm slot B2. Is that configuration what is giving you the memory fault?

If so and the memory modules check out individually, then I agree and reset the motherboard to the original settings see if it boots. Perhaps the memory is overclocked too much. It could have been on the verge of not liking the overclocked memory setting you had initially and now that you moved things around it doesn't like that setting.

quote:

I finally got around to transplanting the guts from my old Antec Lanboy Air into the Rosewill rackmount server chassis (and its corresponding 4u of space in the rack) I have lying around

This implies the working system is the guts referred to, and it implies that you didn't install or move anything on the assembled MB moving the components from old enclosure to new enclosure.

So it also sort of implies that the DIMMs were in those slots before and POSTed OK in the old enclosure. OTOH, if anything had been changed on the MB before this all started, those changes would be a focus for diagnostics.

Assuming the MB wasn't tweaked, I don't know why unplugging power/headers, moving the MB to the new enclosure, and reconnecting everything exactly as before on a MB with no physical changes would suddenly cause RAM-related POST failures. If you'd static-zapped something while moving the components, you'd probably be seeing more global/serious issues.

Does the MB have diagnostics that you could run, when booting in a working-but-not-your-end-goal config?

quote:

Both DIMMs check out individually in each of the 4 slots on the board, and I can get it to POST no problem with both DIMMs in, as long as it's not dual channel. I've cleared the CMOS 3-4x.

It sounds like you've gone one by one on the DIMMs on all the slots and they individually all check out on every slot.

If it was working before (assumption, per above) then one thing I would NOT recommend at this point is updating any firmware or changing any settings. Why introduce another change to the mix during diagnostics? May be that after figuring this out, your fix may require tweaking/updating, but it doesn't sound like that's where you are.

quote:

Originally posted by xantom:

quote:

MSI X470 Gaming Plus

I think you did this, but from the manual:
"If 3 long beeps are heard, remove all
memory modules and try to install only
one memory module in the DIMMA2 slot
first and then restart the computer."

The only suggestion I have is to reseat the CPU and verify nothing is damaged.

You might find something here if you haven't came across it:
https://forum-en.msi.com/index.php?topic=305675.0

^^^this ^^^ remove the memory and install on at a time (attempting to boot after each change) in different slots and try to isolate the issue.

Also IF it is a CPU seating issue (which should not have happened if it as in properly first and never removed) then be carful and inspect the socked pins with a magnifying glass to make sure they are not bent. This can cause memory issues.

My many years of system building say it is related to the memory or the slots themselves.
But that is just an educated guess, test on.

I would first try some other memory in dual channel if you have any. Then I would try to go into BIOS and try slower memory speed settings (non-XMP) and standard voltages with the possibly bad memory. Then I would try booting to BIOS and re-flashing BIOS (current or newer). If you pull the CPU, be careful with the AMD CPU's when pulling the heat sink. The heat sink goop dries up and the heat sinks tend to pull the AMD cpu right out of the socket which makes it VERY easy to bend pins. Just popped the spring pressure lever on a Wraith AMD cooler and it pulled the CPU out of the socket from the opposing spring tension on the other side of the cooler. Luckily didn't bend any pins, but could just have easily done so. If you are going to try pulling and re-seating the CPU, run it for a little while in BIOS just before trying so the CPU warms up and softens the heat sink compound a little.

quote:

Originally posted by xantom:

quote:

MSI X470 Gaming Plus

I think you did this, but from the manual:
"If 3 long beeps are heard, remove all
memory modules and try to install only
one memory module in the DIMMA2 slot
first and then restart the computer."

The only suggestion I have is to reseat the CPU and verify nothing is damaged.

You might find something here if you haven't came across it:
https://forum-en.msi.com/index.php?topic=305675.0

Did that. BIOS is at default configs. No overclocking at all any more; it's not worth the hassle. Likewise, XMP has never given me anything but trouble.

If Slot 1A and Slot 1B are occupied, I get 3 beeps. Same if Slot 2A and 2B are occupied. If Slots 1A and 2A are occupied it POSTs no problem. Same with 1B and 2B. Each DIMM tests out fine on its own, in each of the 4 slots on the motherboard.

I did not remove the CPU or cooler. It did work quite well in the old chassis.

I do agree that if I zapped it I'd likely see issues across the board. I'll look for some diagnostic tools.

I do have the Wraith, so I'll check that tonight. Thanks for the tip.

Another thing to try is check for any firmware updates for the motherboard and updates for the drivers as well and update as needed. Perhaps a motherboard firmware update may solve the problem.

quote:

Originally posted by vulrath:

If Slot 1A and Slot 1B are occupied, I get 3 beeps. Same if Slot 2A and 2B are occupied. If Slots 1A and 2A are occupied it POSTs no problem. Same with 1B and 2B. Each DIMM tests out fine on its own, in each of the 4 slots on the motherboard.

I'm not sure what the problem is according to your description. According to the manual I downloaded for your motherboard, Slot DIMMA2 and DIMMB2 is the proper way to put in "two" sticks of memory for dual channel. Not DIMMA1 and DIMMA2 or DIMMB1 and DIMMB2. Basically, there should be a empty DIMM slot between the memory.

Dual channel memory means half of memory on one channel, and half of memory in other channel. So one stick in DIMMA2 slot, and one stick in DIMMB2 slot. Your motherboard memory is operating as it is supposed to according to the manual. DIMMA1/2 is memory channel A, DIMMB1/2 is memory channel B.

Page 29 of the manual.

quote:

Originally posted by vulrath:
[If Slot 1A and Slot 1B are occupied, I get 3 beeps. Same if Slot 2A and 2B are occupied. If Slots 1A and 2A are occupied it POSTs no problem. Same with 1B and 2B. Each DIMM tests out fine on its own, in each of the 4 slots on the motherboard.

I did not remove the CPU or cooler. It did work quite well in the old chassis.

Since it does work but only in certain slots:
1. How was it 'slotted' before the move?
2. Did you verify the compatibility against tested their site?
3. Did you change anything else such as BIOS?
4. Again did you verify the pin health.

I build computers for a living (although 100% Intel and not saying it is an AMD problem) but I have seen this in two scenarios:
1. BIOS issues - requiring a new update. Sometimes you need to speak to a factory tech to get this bit of info out of them.
2. Physical damage to the CPU (socket)pins. You need to remove carefully and inspect.

Otherwise the easy thing may be to live with just 32gb (diff from 32 to 64 is not really a big deal or improvement for most installations).
I have been through many frustrating trouble shooting sessions on hardware issues and know how you feel.

Good Luck

These are the memory configurations shown in the manual to run dual-channel.




It sounds like they're not running in this dual channel configuration in the middle figure, correct?

quote:

Originally posted by bobtheelf:
These are the memory configurations shown in the manual to run dual-channel.




It sounds like they're not running in this dual channel configuration in the middle figure, correct?

Correct. Initial configuration was slots 2A and B (if you look at a photo of the motherboard, it's the two red slots). I didn't touch the ram until I started to encounter issues, then I powered up and down 10 times with all the different possible ram configurations. It does not work in dual channel, but if I have 1 dimm or both dimms on the same channel I don't have any issues - everything is detected and shows the correct speed for single channel.

Unfortunately I don't have access to any more DDR4 to test with.

Ram is on the compatible list.

I'm not at home right now but when I do I'm going to try a few things like pulling it from the chassis and reseat the processor.

Make sure you don’t have a left over chassis stand-off In the new case. The previous motherboard standoff configuration could have been different.

Does the system work properly outside of that chassis on the test bench?

quote:

Originally posted by pbramlett:
Make sure you don’t have a left over chassis stand-off In the new case. The previous motherboard standoff configuration could have been different.

Does the system work properly outside of that chassis on the test bench?

Excellent suggestion. Had that happen to me once.

I'm actually going to be finding that out shortly.

Could be a corrupted BIOS image. Might be worth trying to re-flash the BIOS. That will do more than just a cmos reset.

Memory training is certainly an arcane process, but if you reset the BIOS that should mean XMP (or I guess AMD likes to call it AMP) is disabled and all but the most god-awful sticks of RAM should train at those stock speeds. Barring that, it seems like the cpu is the most likely culprit.

Since you tried sticks in all the slots, it is unlikely to be a physical problem with the board like a short or gunk in a ram slot. It works with. It’s sticks in one channel, so it isn’t likely a power issue either. That pretty much leaves BIOS software, or the CPU its self.

- Bret

quote:

Originally posted by sadlerbw:
Could be a corrupted BIOS image. Might be worth trying to re-flash the BIOS. That will do more than just a cmos reset.

Memory training is certainly an arcane process, but if you reset the BIOS that should mean XMP (or I guess AMD likes to call it AMP) is disabled and all but the most god-awful sticks of RAM should train at those stock speeds. Barring that, it seems like the cpu is the most likely culprit.

Since you tried sticks in all the slots, it is unlikely to be a physical problem with the board like a short or gunk in a ram slot. It works with. It’s sticks in one channel, so it isn’t likely a power issue either. That pretty much leaves BIOS software, or the CPU its self.

- Bret

This is correct. Likely your over-clocking adventure fucked the memory timings. Those are stored in BIOS. In BIOS, you can also turn off the over-clocking you tweaked. Just restore defaults in BIOS and restart.

Again, no overclock. Stock timings, stock multipliers, stock everything.. Been stable for months that way. Even when I was messing with that, I tried as hard as I could to stay away from messing with ram. I understand CPU settings; I don't understand ram settings well enough to do it safely.

So I got it pulled out of the chassis and up on a box, and it fired up without any issues in the original configuration, went straight to BIOS. From what I could see without having the keyboard attached it picked up all 32GB too. I did notice one squirrelly standoff when I pulled it out (it came out with the board), but it was at the bottom left corner of the board. Still, I know grounding issues can manifest in weird ways. Going to rebuild it and see what happens.

Is the PSU one a single or two different, one each case?

0-0 el