Need help testing the cause of my memory problem

[Satan_Prometheus]

So, I've been having an issue with my RAM for roughly the last year and a half and I'm trying to figure out what the hell is going on before I start spending money replacing parts.

 

This started around January 2020. Before this, I had been running a Ryzen 5 3600 on an Asus Prime X370-Pro motherboard with 16GB of Corsair Vengeance LPX DDR4-3000 (running at the default XMP DDR4-3000 settings). My PSU is a Corsair TX850M and my GPU is a Gigabyte Gaming OC RTX 2070 Super.

 

In January 2020 I decided that my rig was too large and I wanted to downsize to a smaller PC, so I ended up buying a micro-ATX case and a new motherboard, the Asrock X570M Pro4. I also bought a new boot drive at this time, a WD Black SN750.

 

I swapped everything over to this new system (and did a fresh install of Windows) and everything seemed to be running OK. I did have a moment when the side of my case slipped off the edge of the shelf it was sitting on and caused the whole PC to slam down about 1 inch on that side (not the other side). My problems started around this time so I am wondering if this is the original cause of the problem.

 

However, after about a month I noticed that I was having trouble upon re-boot - only one stick of my RAM was showing up. Upon boot, the Asrock splash screen would give me an error message that memory training had failed on one of the slots.

 

After this, I tried switching RAM to the other slots and attempting to boot at a lower speed (DDR4-2666 and DDR4-2400) with no luck. I also tried a different, brand-new RAM kit (G. Skill Ripjaws DDR4-3600) and this didn't change things either.

 

So, I reluctantly switched back to my old motherboard and case. On this board, both sets of RAM worked as expected. Since this worked, at this point I assumed that the issue was that the motherboard had either been damaged when I bought it or was damaged when the PC dropped. However, during this re-build I also bent a pin on the 3600 when I was moving the CPU back into the X370 board. Luckily, I was able to bend it back and had it boot with no issues.

 

However, after a few months, I started noticing that the system was still having some problems. It wouldn't always manage to boot at DDR4-3200 or DDR4-3000 (using XMP profiles for both) and fell back to the base DDR4-2133 speeds frequently. I also started to notice rampant crashing in memory-intensive situations, such as when playing AC Valhalla and walking through crowded areas, or when attempting to load a larger file in Photoshop. This didn't make any sense if the problem had been the X570 board, as before I moved to the X570 system, I had no problems running the Ryzen 5 3600 on the X370 board with DDR4-3000 at the proper speed. (I haven't tried to manually tune the RAM, but I don't feel that running DDR4-3000 with XMP and fairly loose timings should be a problem for a 3600 even on an X370 board.)

 

One day, I had an issue with the X370 motherboard and it died, so I switched back to the X570 board (yet again) and was surprised to discover that it could actually run my RAM just fine, with both sticks showing up. This worked...for a while, until just about a week ago, the old problem with the board only seeing one stick returned again. So now I'm stuck with a 3600 running a single channel of 8GB of RAM.

 

It's worth noting that before the problem with only one stick being visible re-appeared, I ran MemTest86 overnight and didn't get any errors (with the G. Skill memory). I also ran AIDA64 on CPU, GPU, RAM, and SSD for hours and didn't experience any problems or crashes.

 

All this makes me figure the problem has to be one of several issues:

 

1. The memory is damaged, though this seems unlikely since the problems re-occur with two different sets of memory, and I ran MemTest86 and didn't get any errors.

2. The IMC on the CPU is damaged (or possibly the bent pin is causing the problem)

3. The motherboard is damaged (though again, this seems unlikely since I was having memory problems on both motherboards)

4. The power supply itself is damaged and isn't providing enough power over the 8-pin or 24-pin cables to properly power the memory slots (though I did test the power supply with a multimeter and found that it was providing the correct voltage out of every plug).

 

So, at this point, I'm looking at possibly needing to replace CPU and/or motherboard and/or power supply.

 

Are there any further tests I can do to try to determine the cause of the problem before I start spending money to swap out parts?

 

If you made it this far, thanks!

[Bombastinator]

Seems to be repetitive instances where multiple things happen at once. Adding something besides the one thing often doubles or quadruples troubleshooting difficulty.   Most of troubleshooting is figuring out how to test just one thing at a time.  Simplifies matters immensely.  The x570 stick only seeing one thing on again off again thing is particularly annoying.  Could even be cracked solder joints which are generally hell.  The issue there is things can vary with temperature expansion which is Uber annoying. 

[OCNewbee]

10 hours ago, Satan_Prometheus said:

So, I've been having an issue with my RAM for roughly the last year and a half and I'm trying to figure out what the hell is going on before I start spending money replacing parts.

 

This started around January 2020. Before this, I had been running a Ryzen 5 3600 on an Asus Prime X370-Pro motherboard with 16GB of Corsair Vengeance LPX DDR4-3000 (running at the default XMP DDR4-3000 settings). My PSU is a Corsair TX850M and my GPU is a Gigabyte Gaming OC RTX 2070 Super.

 

In January 2020 I decided that my rig was too large and I wanted to downsize to a smaller PC, so I ended up buying a micro-ATX case and a new motherboard, the Asrock X570M Pro4. I also bought a new boot drive at this time, a WD Black SN750.

 

I swapped everything over to this new system (and did a fresh install of Windows) and everything seemed to be running OK. I did have a moment when the side of my case slipped off the edge of the shelf it was sitting on and caused the whole PC to slam down about 1 inch on that side (not the other side). My problems started around this time so I am wondering if this is the original cause of the problem.

 

However, after about a month I noticed that I was having trouble upon re-boot - only one stick of my RAM was showing up. Upon boot, the Asrock splash screen would give me an error message that memory training had failed on one of the slots.

 

After this, I tried switching RAM to the other slots and attempting to boot at a lower speed (DDR4-2666 and DDR4-2400) with no luck. I also tried a different, brand-new RAM kit (G. Skill Ripjaws DDR4-3600) and this didn't change things either.

 

So, I reluctantly switched back to my old motherboard and case. On this board, both sets of RAM worked as expected. Since this worked, at this point I assumed that the issue was that the motherboard had either been damaged when I bought it or was damaged when the PC dropped. However, during this re-build I also bent a pin on the 3600 when I was moving the CPU back into the X370 board. Luckily, I was able to bend it back and had it boot with no issues.

 

However, after a few months, I started noticing that the system was still having some problems. It wouldn't always manage to boot at DDR4-3200 or DDR4-3000 (using XMP profiles for both) and fell back to the base DDR4-2133 speeds frequently. I also started to notice rampant crashing in memory-intensive situations, such as when playing AC Valhalla and walking through crowded areas, or when attempting to load a larger file in Photoshop. This didn't make any sense if the problem had been the X570 board, as before I moved to the X570 system, I had no problems running the Ryzen 5 3600 on the X370 board with DDR4-3000 at the proper speed. (I haven't tried to manually tune the RAM, but I don't feel that running DDR4-3000 with XMP and fairly loose timings should be a problem for a 3600 even on an X370 board.)

 

One day, I had an issue with the X370 motherboard and it died, so I switched back to the X570 board (yet again) and was surprised to discover that it could actually run my RAM just fine, with both sticks showing up. This worked...for a while, until just about a week ago, the old problem with the board only seeing one stick returned again. So now I'm stuck with a 3600 running a single channel of 8GB of RAM.

 

It's worth noting that before the problem with only one stick being visible re-appeared, I ran MemTest86 overnight and didn't get any errors (with the G. Skill memory). I also ran AIDA64 on CPU, GPU, RAM, and SSD for hours and didn't experience any problems or crashes.

 

All this makes me figure the problem has to be one of several issues:

 

1. The memory is damaged, though this seems unlikely since the problems re-occur with two different sets of memory, and I ran MemTest86 and didn't get any errors.

2. The IMC on the CPU is damaged (or possibly the bent pin is causing the problem)

3. The motherboard is damaged (though again, this seems unlikely since I was having memory problems on both motherboards)

4. The power supply itself is damaged and isn't providing enough power over the 8-pin or 24-pin cables to properly power the memory slots (though I did test the power supply with a multimeter and found that it was providing the correct voltage out of every plug).

 

So, at this point, I'm looking at possibly needing to replace CPU and/or motherboard and/or power supply.

 

Are there any further tests I can do to try to determine the cause of the problem before I start spending money to swap out parts?

 

If you made it this far, thanks!

If this problem is intermittent, it is probably your CPU seating problem. To be more precise, your CPU cooler is not mounted properly hence the IMC part of CPU is slightly lifted causing a failure of contact with the motherboard. Make sure proper pressure is on the CPU.

[Bombastinator]

23 minutes ago, OCNewbee said:

If this problem is intermittent, it is probably your CPU seating problem. To be more precise, your CPU cooler is not mounted properly hence the IMC part of CPU is slightly lifted causing a failure of contact with the motherboard. Make sure proper pressure is on the CPU.

There was a cpu seating problem?  I thought ZIF systems basically removed that as a possibility. Ryzen uses PGA which like LGA is ZIF. PGA is pretty hard to screw up but I suppose it can happen.  Might even make some sense for intermittent connection. A barely attached or otherwise messed up pin on the CPU.  If it’s a cooler issue that would go along with wild heat variations where the cpu would overheat and shut down. That could be called a cooler seating issue rather than a cpu seating issue.

Edited May 31, 2021 by Bombastinator

[OCNewbee]

13 minutes ago, Bombastinator said:

There was a cpu seating problem?  I thought ZIF systems basically removed that as a possibility. Ryzen uses PGA which like LGA is ZIF. PGA is pretty hard to screw up but I suppose it can happen.  Might even make some sense for intermittent connection. A barely attached or otherwise messed up pin on the CPU.  If it’s a cooler issue that would go along with wild heat variations where the cpu would overheat and shut down. That could be called a cooler seating issue rather than a cpu seating issue.

whatever issue it is called, just reseat the CPU. Make sure your CPU is perfectly flat. You can test this by spinning the CPU on a flat surface. If it rotates like a top, you have an even surface which would cause issues. 

 

Just as more info., when CPU temps approach 85C+ it starts to warp the PCB and CPU is no exception. So over the years, if you were running the CPU a bit hot, it could have warped hence causing the poor contact with the IMC. Regardless, a firm pressure on the CPU with the cooler should solve the problem.

 

If this fails, boot with a single stick of ram on slot 2 and after a successful boot, turn it off. Now add 2nd stick on slot 4. 

[Bombastinator]

20 minutes ago, OCNewbee said:

whatever issue it is called, just reseat the CPU. Make sure your CPU is perfectly flat. You can test this by spinning the CPU on a flat surface. If it rotates like a top, you have an even surface which would cause issues. 

 

Just as more info., when CPU temps approach 85C+ it starts to warp the PCB and CPU is no exception. So over the years, if you were running the CPU a bit hot, it could have warped hence causing the poor contact with the IMC. Regardless, a firm pressure on the CPU with the cooler should solve the problem.

 

If this fails, boot with a single stick of ram on slot 2 and after a successful boot, turn it off. Now add 2nd stick on slot 4. 

I like the spin test.  Hadn’t heard that one but it makes sense for high spots.  Wouldn’t test for low spots. An uneven top with an indented center would still resist spinning. 
 

remembering that acronyms often suck.

ZIF is “Zero Insertion Force” where the electrical connection is friction fit.  Including from a spring system (lga) or from applying tension after the device is placed (pga) 

 

this may be a me problem where reseating the cpu and reseating the cooler are confused.  You can’t reseat the cpu without also reseating the cooler so it’s something that can be easily mixed back and forth.

Edited May 31, 2021 by Bombastinator
Oh wow autocorrect is evil. Also structural changes so it wasn’t the only problem this time.

[OCNewbee]

1 minute ago, Bombastinator said:

I like the spin test.  Hadn’t heard that one but it makes sense for high spots.  Wouldn’t test for low spots. An uneven top with an indented center would still resist spinning. 

true. Never considered that. But most silicon dies are in the center so an indented center is super unlikely but possible I suppose.

[Bombastinator]

3 minutes ago, OCNewbee said:

true. Never considered that. But most silicon dies are in the center so an indented center is super unlikely but possible I suppose.

Would require a problem with attaching the ihs in the first place.  Also there’s the possibility of a “doughnut” type issue where there is one or more dents around the silicon but not on it.

[OCNewbee]

11 minutes ago, Bombastinator said:

I like the spin test.  Hadn’t heard that one but it makes sense for high spots.  Wouldn’t test for low spots. An uneven top with an indented center would still resist spinning. 
 

remembering that acronyms often suck.

ZIF is “Zero Insertion Force” where the electrical connection is friction fit and either of curse from a spring system (lga) or from applying tension after the device is placed (pga) 

 

this may be a me problem where reseating the cpu and reseating the cooler are confused.  You can’t RDAs eat the cpu without also reseating the cooler so it’s something that can be easily mixed back and forth.

Yeah I do understand. Since AMDs are PGA, a poor connection due to pressure can be ruled out. So it could be dust in the mobo CPU sockets or damaged pin. Or both. But my hunch is that there is some kind of an IMC problem as evident from the intermittent ram issue. 

[Bombastinator]

5 minutes ago, OCNewbee said:

Yeah I do understand. Since AMDs are PGA, a poor connection due to pressure can be ruled out. So it could be dust in the mobo CPU sockets or damaged pin. Or both. But my hunch is that there is some kind of an IMC problem as evident from the intermittent ram issue. 

Also actual connections of individual pins.  They’re a bit like tiny little thumb tacks that are soldered on and have been know to fully or partially pull off. 

[OCNewbee]

3 minutes ago, Bombastinator said:

Also actual connections of individual pins.  They’re a bit like tiny little thumb tacks that are soldered on and have been know to fully or partially pull off. 

Yup, I know the the PGAs pretty well. First computer in 1994. 

[Bombastinator]

1 minute ago, OCNewbee said:

Yup, I know the the PGAs pretty well. First computer in 1994. 

Ah.  Name confused me.  The possibility that you were like 12 or something was there.

[Satan_Prometheus]

10 hours ago, OCNewbee said:

whatever issue it is called, just reseat the CPU. Make sure your CPU is perfectly flat. You can test this by spinning the CPU on a flat surface. If it rotates like a top, you have an even surface which would cause issues. 

 

Just as more info., when CPU temps approach 85C+ it starts to warp the PCB and CPU is no exception. So over the years, if you were running the CPU a bit hot, it could have warped hence causing the poor contact with the IMC. Regardless, a firm pressure on the CPU with the cooler should solve the problem.

 

If this fails, boot with a single stick of ram on slot 2 and after a successful boot, turn it off. Now add 2nd stick on slot 4. 

Thanks, I appreciate the feedback. That said, these problems have persisted over multiple re-installations on several different motherboards, so if it were a mounting error I would have had to make the same mistake with the mounting three times.

 

The cooler is NH-D15 and I don't run the CPU at maximum for extended periods of time (this is mainly a gaming system, with some light Photoshop and audio editing work). Also, the problems first emerged only a few weeks after first installing the X570M Pro4 motherboard, and the system is not overclocked. So I'm not inclined to blame heat for the problems.

 

My inclination is also to believe that it's an IMC problem, but I wanted to see if it were possible to eliminate the motherboard and PSU as the potential source of the problem before I spend money on a new CPU. I would be very happy if I could simply drop in a new CPU and fix everything.

 

@Bombastinator do either of you have any suggestions for how to test the motherboard or power supply in isolation? I've already tested the RAM with Memtest and it brought up no errors so I believe that the DIMMs themselves are fine.

 

Basically, I'm trying to avoid having a situation where I having to keep spending money replacing components until I find the one that fixes the problem.

 

Thanks to both of you for your help!

[Satan_Prometheus]

Also, is there a way to clean out the socket itself? I am worried I might have gotten thermal paste in it.

 

I am a VERY SMART PERSON

[OCNewbee]

3 minutes ago, Satan_Prometheus said:

Thanks, I appreciate the feedback. That said, these problems have persisted over multiple re-installations on several different motherboards, so if it were a mounting error I would have had to make the same mistake with the mounting three times.

 

The cooler is NH-D15 and I don't run the CPU at maximum for extended periods of time (this is mainly a gaming system, with some light Photoshop and audio editing work). Also, the problems first emerged only a few weeks after first installing the X570M Pro4 motherboard, and the system is not overclocked. So I'm not inclined to blame heat for the problems.

 

My inclination is also to believe that it's an IMC problem, but I wanted to see if it were possible to eliminate the motherboard and PSU as the potential source of the problem before I spend money on a new CPU. I would be very happy if I could simply drop in a new CPU and fix everything.

 

@Bombastinator do either of you have any suggestions for how to test the motherboard or power supply in isolation? I've already tested the RAM with Memtest and it brought up no errors so I believe that the DIMMs themselves are fine.

 

Basically, I'm trying to avoid having a situation where I having to keep spending money replacing components until I find the one that fixes the problem.

 

Thanks to both of you for your help!

The only way is to borrow parts.

 

As you mentioned, you can rule out ram by doing memtest on each.

PSU cannot be the problem because your window boots up fine. Btw, what is your PSU?

The only problem is determining whether it is the CPU or the motherboard.

The only way is to borrow a CPU and a motherboard from someone else or just go to a computer shop. 

 

[Satan_Prometheus]

4 minutes ago, OCNewbee said:

The only way is to borrow parts.

 

As you mentioned, you can rule out ram by doing memtest on each.

PSU cannot be the problem because your window boots up fine. Btw, what is your PSU?

The only problem is determining whether it is the CPU or the motherboard.

The only way is to borrow a CPU and a motherboard from someone else or just go to a computer shop. 

 

My PSU is a Corsair TX850M. I've had it for about four years so I think it should still be OK.

 

I may just bite the bullet and buy a 5600X/5800X and see if dropping that in solves the problems. If not I'll have to buy a new board.

[OCNewbee]

5 minutes ago, Satan_Prometheus said:

My PSU is a Corsair TX850M. I've had it for about four years so I think it should still be OK.

 

I may just bite the bullet and buy a 5600X/5800X and see if dropping that in solves the problems. If not I'll have to buy a new board.

I think its the bent pin man. Poor contact with the IMC part.

[Bombastinator]

1 hour ago, Satan_Prometheus said:

Also, is there a way to clean out the socket itself? I am worried I might have gotten thermal paste in it.

 

I am a VERY SMART PERSON

I used to think I was a very smart person.  Tests said so.  Things change.  People get old.  There are people here who think I am a complete moron though they tend not to say so directly.

[Satan_Prometheus]

6 minutes ago, Bombastinator said:

I used to think I was a very smart person.  Tests said so.  Things change.  People get old.

Me at age 18: acing college calculus exams

 

Me now at age 33: using a calculator to add 2-digit numbers

[Bombastinator]

1 hour ago, OCNewbee said:

I think its the bent pin man. Poor contact with the IMC part.

If the cpu slots all the way in the bent pin man is defeated I thought.

[Satan_Prometheus]

So I bit the bullet and went ahead and replaced the 3600 with a 5600X. Problems gone instantly and the RAM now runs at full DDR4-3600 speeds with no issues. I guess it was the IMC/bent pin after all.

[Bombastinator]

11 minutes ago, Satan_Prometheus said:

So I bit the bullet and went ahead and replaced the 3600 with a 5600X. Problems gone instantly and the RAM now runs at full DDR4-3600 speeds with no issues. I guess it was the IMC/bent pin after all.

Maybe.  Jayz2cents recently did a video on his experiences with ryzen over time.  He was experiencing problems with ryzen2 that just didn’t happen with ryzen3

[Satan_Prometheus]

1 minute ago, Bombastinator said:

Maybe.  Jayz2cents recently did a video on his experiences with ryzen over time.  He was experiencing problems with ryzen2 that just didn’t happen with ryzen3

That might be part of it, though in my case the 3600 worked perfectly fine for the first 6 months I had it before I started swapping things around so I'm inclined to believe the issues were caused by damage the CPU incurred while I was swapping it between boards.

 

Thanks for all your help!