Issues with new Ryzen 3600 build; stuck in a bootloop?

[W1nt3r]

Hi team,

I recently (Wednesday just gone) just built a new machine (using some old parts, GPU and Drives) after moving to the US and my old 2500k finally dying, and it's been running fine for 5 days. However, yesterday I left it on when I went out as I was downloading some steam files, and when I got back it appeared to stuck in a boot loop. Most of the time it makes it past the BIOS splash page and then just black screens and restarts, other times it was throwing a BSOD with various errors (IRQL not less or equal, and PAGE FAULT IN NONPAGED AREA being the ones I can remember). That said, the BSODs seem to have disappeared today.

Specs
Ryzen 5 3600
MSI X570 Gaming plus
16 GB G-Skill Ripjaws V Series DDR4
MSI RX480 (old part)
Corsair Rm550x
Samsung 850 Pro 256GB SSD (Boot drive; old part)
Some random HDDs for storage

Troubleshooting so far

1. Windows Auto-repair (as during the week a number of updates were installed)
   1. Fails to complete with the machine often restarting
2. Flashed the newest BIOS to the motherboard.
   1. Flashed successfully with no changes, still looping
3. Barebones boot with CPU, 1 RAM stick (tried both independently), SSD, and GPU. (Eech RAM stick as tested in both the A2 and A1 slots)
   1. No changes, still looping
4. Barebones boot with CPU, 1 Ram stick, HHD (without a windows install, for a new install), and GPU.
   1. No changes, still looping
5. Barebones boot with CPU, 1 RAM stick, bootable USB, and GPU.
   1. No changes, still looping
6. Reinstall of Windows onto the SSD via a bootable USB.
   1. Fails, can't make it into the USB
7. CMOS reset via removing battery and unplugging machine overnight; also via JBAT1 reset
   1. No changes, still looping

The motherboard debug LED on the motherboard seems to indicate a CPU error. But what's weird to me is that the machine can happily sit in the BIOS menu with no problems whatsoever, and I would have thought if the CPU was the fault this wouldn't happen. I am currently running Memtest86 just to double check the memory, I know the first pass completed without error, but I had to leave before seeing the other three.

My only thought is that I don't have the additional 4-pin CPU connector plugged into the board, and that is causing issues? I have a Molex to 4 pin convertor on the way, but any thoughts/suggestions would be most welcome.

Thanks kindly.

[Jurrunio]

What memory speed?

 

If it's not the memory, try lock the CPU to a lower frequency (say 3GHz) and more than enough voltage (1.3V) and see whether it still bootloops. This will reduce performance and make it run unnecessarily warm (maybe not warmer than stock though), just a test on whether it's something else or the CPU.

 

9 minutes ago, W1nt3r said:

My only thought is that I don't have the additional 4-pin CPU connector plugged into the board, and that is causing issues?

Dont need that extra 4pin, it's hooked up in parallel to the 8pin so all you do is to reduce stress (which is nothing when thr 8pin can do over 300w easy) on the cable and connectors.

 

11 minutes ago, W1nt3r said:

I have a Molex to 4 pin convertor

This is a bigger problem. 4pin molex only has 1 12V wire while EPS 4pin has 2, so that single wire will have to handle double the current other 12V wires powering the CPU are

[W1nt3r]

26 minutes ago, Jurrunio said:

What memory speed? If it's not the memory, try lock the CPU to a lower frequency (say 3GHz) and more than enough voltage (1.3V) and see whether it still bootloops. This will reduce performance and make it run unnecessarily warm (maybe not warmer than stock though), just a test on whether it's something else or the CPU.

I'm still not a home, so I'm not sure if the how Memtest went, so that's still pending. As for speed it's 3600, but I believe in all my removing of parts/BIOS changes etc. the XMP profile is disabled so from memory it's running at 2133.

 

Interesting that you mentioned the 1.3V as a voltage to use, because from memory I am pretty sure that BIOS was reporting the default VCORE to be sitting around 1.4V. Could it be overdrawing and resetting because of that?

 

Quote

This is a bigger problem. 4pin molex only has 1 12V wire while EPS 4pin has 2, so that single wire will have to handle double the current other 12V wires powering the CPU are

I ordered the molex converter to just test and see if it makes any stability difference. If it does, I'll replace it with a proper EPS cable.

[Jurrunio]

45 minutes ago, W1nt3r said:

Interesting that you mentioned the 1.3V as a voltage to use, because from memory I am pretty sure that BIOS was reporting the default VCORE to be sitting around 1.4V. Could it be overdrawing and resetting because of that?

 

default voltage is higher because it's also running higher clock speed and need that voltage to stablize.

[W1nt3r]

So I just got home; Memtest completed (all four passes on 11 tests) with no errors, so it looks like the RAM is fine.

19 hours ago, Jurrunio said:

default voltage is higher because it's also running higher clock speed and need that voltage to stablize.

That makes sense. Seeing as the memory passed, I just tested your suggestion of reducing the CPU to 3.0GHz and on restart it seemed to lock up on a blank black screen. But after a CMOS reset, and a second attempt where I also manually adjusting the core voltage to 1.3V and it's made it to Windows and booted. So I'm not sure exactly what's going on.

 

Maybe it's hitting a thermal issue? But the temperatures reported by the BIOS as like mid 30's, and even during Memtest topped at I think 57.

 

I'm currently updating the Chipset drivers, and hoping that makes a difference, but we will see I guess.

 

EDIT ~ For posterity's sake, here is how I fixed it.

1. Down-clocked the CPU (via setting the ratio to 30) to run at 3.0GHz, and defined the VCORE max to 1.3V rather than an AUTO setting.
   1. This allowed the machine to be stable enough to make it into Windows.
2. After logging into Windows, I downloaded and installed the newest Chipset drivers from AMD, and updated all other Radeon software.
   1. I also made sure to be running on the AMD balanced power plan option.
3. Incrementally adjusted the CPU ratio (32, then 34, and finally 36) and the VCORE max (1.3V, 1.325V, 1.35V) until the processor was back to stock speeds.

The machine has been happily running for 10 + hours now, and passed some CPU stress tests from OCCT. So I'm fairly confident that everything is back to normal.

[Mark Kaine]

56 minutes ago, W1nt3r said:

So I just got home; Memtest completed (all four passes on 11 tests) with no errors, so it looks like the RAM is fine.

That makes sense. Seeing as the memory passed, I just tested your suggestion of reducing the CPU to 3.0GHz and on restart it seemed to lock up on a blank black screen. But after a CMOS reset, and a second attempt where I also manually adjusting the core voltage to 1.3V and it's made it to Windows and booted. So I'm not sure exactly what's going on.

 

Maybe it's hitting a thermal issue? But the temperatures reported by the BIOS as like mid 30's, and even during Memtest topped at I think 57.

 

I'm currently updating the Chipset drivers, and hoping that makes a difference, but we will see I guess.

 

 

 

Just FYI, R3600 voltage should not be higher than 1V while idling,  so you should probably check this after installing the chipset drivers which perhaps already fix this problem. 

 

 

ALSO NOTE: it's fine when it exceeds 1V *while idling* occasionally,  it just shouldn't be constantly above,  it should be under 1V.

 

Just something to keep in mind, might or might not be related to your issue. 

Spoiler

 

 

 

BTW I had very similar issues like the ones that you described, with a B450 board, now I'm on a B350 board,  have nearly no issues and absolutely no crashes, no bootloops,  whatsoever. 

 

What I'm saying, it could be your board too.  Always hard to troubleshoot these things of course. 

[W1nt3r]

50 minutes ago, Mark Kaine said:

Just FYI, R3600 voltage should not be higher than 1V while idling,  so you should probably check this after installing the chipset drivers which perhaps already fix this problem. 

 

 

ALSO NOTE: it's fine when it exceeds 1V *while idling* occasionally,  it just shouldn't be constantly above,  it should be under 1V.

 

Just something to keep in mind, might or might not be related to your issue. 

  Reveal hidden contents

 

 

 

BTW I had very similar issues like the ones that you described, with a B450 board, now I'm on a B350 board,  have nearly no issues and absolutely no crashes, no bootloops,  whatsoever. 

 

What I'm saying, it could be your board too.  Always hard to troubleshoot these things of course. 

This was really interesting, and I took a look through it all and the various tests. Idle voltage bounces a fair bit, but if I had to guess in CPU-z it would average between 0.3-0.5V, and in Ryzen Master it was about 0.4-0.5V, so it seems to be okay.

 

I suspect you may be right that the Chipset driver update has perhaps solved the problem.

 

EDIT ~ I got the CPU back up to 3.6 GHz with a max V-CORE of 1.35V; so it looks like the AUTO settings and the old Chipset may have been the problem. Everything seems stable currently and hopefully will remain that way (finger's crossed).

[Mark Kaine]

18 hours ago, W1nt3r said:

This was really interesting, and I took a look through it all and the various tests. Idle voltage bounces a fair bit, but if I had to guess in CPU-z it would average between 0.3-0.5V, and in Ryzen Master it was about 0.4-0.5V, so it seems to be okay.

 

I suspect you may be right that the Chipset driver update has perhaps solved the problem.

 

EDIT ~ I got the CPU back up to 3.6 GHz with a max V-CORE of 1.35V; so it looks like the AUTO settings and the old Chipset may have been the problem. Everything seems stable currently and hopefully will remain that way (finger's crossed).

I see, nice you got it fixed, so it was the chipset drivers afterall.