Rog Zenith II Extreme Alpha stuck on "Memory Test"

[David Petrofsky]

I just turned on the new PC I'm building for the 1st time and the OLED is indicating "Memory Test" forever, no beep or GPU output. all ram sticks have RGB, and all fans are spinning.  I saw a video where Linus had that happen and said the BIOS needed to be updated, so I used Bios Flashback to update the BIOS of the board.  It didn't fix the issue.  Also tried clearing the CMOS, with no effect.

 

Specs of the build:

ROG Zenith II Extreme Alpha board

AMD 3970x CPU

8 sticks of this memory: https://www.newegg.com/g-skill-128gb-288-pin-ddr4-sdram/p/N82E16820374002?Item=N82E16820374002

rtx 2080 ti founder's edition GPU

EVGA SuperNova 1200 P2 PSU

Noctua U14s TR3 cooler w/ 2 fans

Phanteks Eclipse P600S case

 

What can I do to diagnose the issue further?

[svmlegacy]

3 minutes ago, David Petrofsky said:

I just turned on the new PC I'm building for the 1st time and the OLED is indicating "Memory Test" forever, no beep or GPU output. all ram sticks have RGB, and all fans are spinning.  I saw a video where Linus had that happen and said the BIOS needed to be updated, so I used Bios Flashback to update the BIOS of the board.  It didn't fix the issue.  Also tried clearing the CMOS, with no effect.

 

Specs of the build:

ROG Zenith II Extreme Alpha board

AMD 3970x CPU

8 sticks of this memory: https://www.newegg.com/g-skill-128gb-288-pin-ddr4-sdram/p/N82E16820374002?Item=N82E16820374002

rtx 2080 ti founder's edition GPU

EVGA SuperNova 1200 P2 PSU

Noctua U14s TR3 cooler w/ 2 fans

Phanteks Eclipse P600S case

 

What can I do to diagnose the issue further?

Have you gone through these instructions yet?

 

[David Petrofsky]

It looks like this exact memory is not in Asus's compatibility list while the 3200 MHz is.  How likely is it that there is an actual incompatibility due to using 3600 instead of 3200?  

[svmlegacy]

2 minutes ago, David Petrofsky said:

It looks like this exact memory is not in Asus's compatibility list while the 3200 MHz is.  How likely is it that there is an actual incompatibility due to using 3600 instead of 3200?  

Unlikely, G.SKILL has QVL'd the memory for the motherboard. see: https://www.gskill.com/qvl/165/326/1582266217/F4-3600C18Q-128GTZN-Qvl

Regardles, it should boot up initially with JEDEC speeds, not the faster XMP anyways. JEDEC is always lower than 3200 MT/s.

 

[David Petrofsky]

I did all the steps that I can from that guide.  I don't have any other computers that I can test the RAM in or any other RAM I can test in here.

[svmlegacy]

Quote

12. Try booting your computer with only a motherboard, CPU and one stick of RAM attached (if you don't have an iGPU, plug in your GPU as well ).

This is the most crucial step here. Make sure to remove all USB and front panel devices, along with all drives (Including M.2)

Single DIMM should be placed in slot A1, per ASUS ROG Zenith II Extreme Alpha user manual.

 

Verify the torque on the CPU bolts, as well.

 

If it does not boot, cycle DIMM's, it's very very unlikely to get 8 bad DIMM's.

 

At this point, we have to assume hardware failure.

[Guest]

Have you tried posting with maybe less memory like one or two sticks only?

[svmlegacy]

Verify position of the following switches, as well:

RSVD: OFF

SLOW_MODE: OFF

M.2_3: ON

FS_MODE: OFF

LN2_MODE (Jumper): PINS 1-2 JUMPED

[David Petrofsky]

I've got 1 stick in the A1 slot (post above says A2 but manual says A1) and nothing else connected.  All front panel, etc. connectors unplugged.  All M.2 drives removed. I'm trying the sticks 1 by 1 now.

[David Petrofsky]

Same result all 8 sticks. All switches and jumpers are in the described configuration.

 

[svmlegacy]

14 minutes ago, David Petrofsky said:

Same result all 8 sticks. All switches and jumpers are in the described configuration.

One last ditch effort, try the second BIOS.

 

If that doesn't work, it's bad motherboard, bad CPU, bad CPU installation, or unlikely, bad RAM/compatibility.

Inspect the CPU socket for bent or missing pins.

[David Petrofsky]

I'm trying to pull the cooler off so I can check the CPU torque.  The screws on the cooler seem completely loose after I loosened them and I can twist the cooler a little bit back and forth but it does not come off.  Is that the seal made by the thermal paste?  How do I break it?

[svmlegacy]

2 minutes ago, David Petrofsky said:

I'm trying to pull the cooler off so I can check the CPU torque.  The screws on the cooler seem completely loose after I loosened them and I can twist the cooler a little bit back and forth but it does not come off.  Is that the seal made by the thermal paste?  How do I break it?

Yes. The thermal paste can hold onto a CPU tightly. Twist axially back and forth while gently pulling until it releases. This comes from the massive surface area of sTRX4

[David Petrofsky]

Second BIOS didn't fix it.  Still trying to pull off the cooler.

[David Petrofsky]

I finally got it off.  Attached some pics. Do you see any alignment issues as it is?  Did I put too much thermal paste on? I used X and 4 dots like I see in a lot of videos.  Could the paste at the edges by shorting out some pins or something?

 

I should probably clean the paste off before I open up the cpu assembly.  I've heard I should use alcohol - what kind of cloth should I use?  Regular towels will leave fibers and microfiber will have static.

[David Petrofsky]

It's the paste that comes with the noctua cooler if that makes a difference.

[svmlegacy]

1 minute ago, David Petrofsky said:

I finally got it off.  Attached some pics. Do you see any alignment issues as it is?  Did I put too much thermal paste on? I used X and 4 dots like I see in a lot of videos.  Could the paste at the edges by shorting out some pins or something?

 

I should probably clean the paste off before I open up the cpu assembly.  I've heard I should use alcohol - what kind of cloth should I use?  Regular towels will leave fibers and microfiber will have static.

The thermal paste application looks just fine. Most pastes are non-conductive.

I pretty much always clean with paper towel and 99% isopropyl alcohol. Just don't use Acetone/Nail Polish remover.

[David Petrofsky]

Here are some pics of the cleaned off CPU (in case anyone spots alignment issues or whatever) followed by the pins on the board and the sockets on the CPU.  Witih my feeble human eyes I do not see any out of place pins or sockets.  With the camera, I see one area that looks a little strange but then with a pic from a slightly different angle it's fine, so probably lighting and reflections.

[David Petrofsky]

So clearly something I have is bad.  How in the world can I figure out what it is?  I hope it's not the motherboard - this is the 3rd copy of that motherboard I tried.  The first one came in a box that was totally beat up and torn all the way open on one side so I just rejected the package outright because I couldn't take the chance of it being abused and exposed like that, and then ordered a 2nd one.  Then that one had a rattling screw or something else metallic inside, and now this one may or may not be bad.  Maybe I shouldn't have rejected the 1st package - that might have been the best one!

 

My motherboard is from Amazon and my CPU is from B&H Photo.  My RAM is from newegg.  So the RMA situation will be interesting.

 

Is there some way I can tell what is bad using the multimeter probe ports?

[HM-2]

Did you build the second one? Has any of the three boards you had booted?

I doubt it's a RAM issue given you've tried individual sticks, but it could easily be the CPU or motherboard. 

[svmlegacy]

11 hours ago, David Petrofsky said:

Is there some way I can tell what is bad using the multimeter probe ports?

You could probe the probe points if you have a DVOM (multimeter). If something is way off, you can determine that it's the motherboard.

 

In your position, I'd contact ASUS support for guidance on determining if your motherboard is eligible for an RMA. Either the motherboard or CPU must be changed to tell which is the fault here.

 

I don't see anything obviously wrong with the socket, but it is very difficult to tell through pictures.

[David Petrofsky]

ok I tested the voltages at the probe points and these were my results.  I don't know what the engineering tolerances are so I give the full readout of my meter here and only truncate if trailing zeros.

 

VCORE: 0V (sometimes 1 mV)(should it be higher at this point in the boot process?)

MB_SOC: 0.89

DRAM_AB: 1.19

DRAM_CD: 1.193

1.8_PLL: 1.757 (5% below what its name suggests, is that normal)

SB: 1.2

GND: 0 (because I used it as the reference terminal)

 

All were measured when the system is in that same state I'm so used to seeing now: just sitting there with "Memory Test" on the OLED with all fans and lights doing their thing but no beeps or output.

 

I measured before hitting the "start" button and all were effectively zero.

 

I also did some continuity checks in various places on the case to see if there was some sort of short from mounting the motherboard.  I didn't find anything.  I guess I could check if VCORE is shorted to GND as long as I have the meter out.

[David Petrofsky]

Regarding testing of the components with different boards, etc.:

 

Board #1 was rejected on delivery due to the state the shipping box was in.  That may have been a mistake, but my instinct said the contents of that box were over $3000 (CPU + motherboard + other stuff) and it was treated very badly.  I've never seen a shipping box in that shape before.  I had to make a snap decision while the UPS guy was there so I chose the safe route.

 

Board #2 had the CPU mounted but I had the wrong cooler so I never got to try booting it up. I submitted the return request to B&H based on the fact that I heard a loose metal screw while I was trying to find the mount points for the cooler, so it's lucky I had the wrong cooler or I wouldn't have noticed that sound.

 

Board #3 is the one I have now.  I don't hear any bad noises when I move it, and everything SEEMS to be fine except for whatever is going on right now with the Memory Test stuck issue.

 

I haven't sent board #2 back yet though.  I was going to today, but I guess as long as I have it, I might as well put the CPU, RAM, and PSU to it and see what happens.  If it works, I guess I know board #3 is bad.  If it doesn't work, I still don't know anything at all.

[David Petrofsky]

Since I know I'm not keeping board #2, when I test the CPU on board #2, is it ok to just leave the current thermal paste on the cpu and cooler?  I know if it were a long-term seating that would be bad, but for this test it can't make that huge a difference, right?

[svmlegacy]

17 minutes ago, David Petrofsky said:

VCORE: 0V (sometimes 1 mV)(should it be higher at this point in the boot process?)

MB_SOC: 8.89

DRAM_AB: 1.19

DRAM_CD: 1.193

1.8_PLL: 1.757 (5% below what its name suggests, is that normal)

SB: 1.2

GND: 0 (because I used it as the reference terminal)

I believe VCore should be ~ 1.4 iiiiish volts at this point. Did you end up verifying if it was shorted to ground? The DRAM is getting proper voltage, so it's difficult to say with certainty.

5% tolerance should be OK for bootup. I tend to get worried around a 10% deviation.

that being said, is the MB_SOC really almost 9V? That's somewhat concerning, but I'm not sure what it's actually doing.

 

9 minutes ago, David Petrofsky said:

Since I know I'm not keeping board #2, when I test the CPU on board #2, is it ok to just leave the current thermal paste on the cpu and cooler?  I know if it were a long-term seating that would be bad, but for this test it can't make that huge a difference, right?

Definitely test the CPU in board #2. Check the VCore there while you're at it. Generally you can leave the paste on for temporary tests.