X670E Carbon Wifi Stuck on Debug Code AB

[ftbscreame15b402e7]

Hi

Something very strange is happening.

On cold boot my PC will stay on Code A6 with both CPU and DRAM light on. This will happen with EXPO on or off. If I then shut down the PC by holding in the power button and power it back on again it will boot without issue.

Restarts and shut downs during the rest of the day will be fine, it is only when the PC has been off during the night.

System specs:
7950X3D
2x 16GB Gskill F5-6000J3038F16GX2-TZ5NR
Samsung 960 Pro 500 GB MZ-V6P512BW (Windows 11 installed)
Mushkin Pilot-E 2TB MKNSSDPE2TB-D8
Crucial P1 1TB CT1000P1SSD8And
3x SATA drives

 

[lostking20015bf02e6]

What happened if you just wait, for example 10 minutes. will it boot itself?

 

[Encryptme]

Generally It means it's searching for a boot device on the SCSI (or SATA) bus. If you haven't installed the OS yet, and are able to boot to the BIOS then nothing is abnormal. Once the OS is installed, it will find the bootloader in that step and then move on. Maybe adjust the boot order of your devices?

 

[ryegrass]

There was a very similar post about a month ago with a Code A6 (SCSI Detect) intermittent error at boot which turned out to be the early stages of a drive failure. It would probably be a good idea to check your existing drives with software similar to CrystalDiskInfo and of course the builtin Windows drive checking utilities to eliminate that as a possibility.

 

[Alan J T]

Are you turning off the PSU overnight
And have you disabled all boot options except for your Windows UEFI Install drive

 

[Alan J T]

There was a very similar post about a month ago with a Code A6 (SCSI Detect) intermittent error at boot which turned out to be the early stages of a drive failure. It would probably be a good idea to check your existing drives with software similar to CrystalDiskInfo and of course the builtin Windows drive checking utilities to eliminate that as a possibility.

Yep one was a brand new Samsung Drive that set itself to read-only due to the firmware problem
I think the other was an older Spinner taking to long to power up or spin up

Interestingly one of my USB external SSD System Backup drives is detected as SCSI
And it has a M.2 Drive in it a Patriot P300 M.2 PCIe Gen 3 x4 1TB SSD- P300P1TBM28

 

[Encryptme]

Yep one was a brand new Samsung Drive that set itself to read-only due to the firmware problem
I think the other was an older Spinner taking to long to power up or spin up

Interestingly one of my USB external SSD System Backup drives is detected as SCSI
And it has a M.2 Drive in it a Patriot P300 M.2 PCIe Gen 3 x4 1TB SSD- P300P1TBM28

View attachment 172853

This may have been a Samsung 990 Pro, because mine have both needed firmware updates a while back.

 

[ftbscreame15b402e7]

What happened if you just wait, for example 10 minutes. will it boot itself?

It will stay there indefinitely unfortunatly.

Generally It means it's searching for a boot device on the SCSI (or SATA) bus. If you haven't installed the OS yet, and are able to boot to the BIOS then nothing is abnormal. Once the OS is installed, it will find the bootloader in that step and then move on. Maybe adjust the boot order of your devices?

The Samsung 960 Pro is the first boot device. I'll try and disable the rest like AlanJT sugggested.

There was a very similar post about a month ago with a Code A6 (SCSI Detect) intermittent error at boot which turned out to be the early stages of a drive failure. It would probably be a good idea to check your existing drives with software similar to CrystalDiskInfo and of course the builtin Windows drive checking utilities to eliminate that as a possibility.

CrystalDiskInfo says the drive is healthy and in good order.

Are you turning off the PSU overnight
And have you disabled all boot options except for your Windows UEFI Install drive
View attachment 172852

Thank you I'll try that. Yes the power is off by the wall socket over night.

 

[Alan J T]

Thank you I'll try that. Yes the power is off by the wall socket over night.

Try not turning it off for a couple of nights and see what happens
As I have had mine lock up at boot when PSU has been unplugged

The Drive is OK then it was the 980 or 990 Samsung drives that had the prob with the firmware

BUt it good idea DL Samsung Magician and check for firmware updates

 

[ftbscreame15b402e7]

So to give feedback so far, the disable other boot options did not help. I have updated the Samsung 960Pro to latest firmware today, and will see how that goes.

Please bear with me, I am changing one thing at a time and since it only happens on the first boot of the day, it is taking a while to troubleshoot

 

[Encryptme]

I just reieved two more firmware updates for my 990 Pro M.2 drives - kinda weird since the other ones were only a few months back.

 

[Ash4324]

Pretty sure you're chasing down the wrong rabbit hole. The error code more likely is Ab which is "Setup input wait". Getting the same error when using Ryzen master pbo settings apply. Works fine when setting the same PBO values in BIOS.
Also getting the same error with the newest beta bios (1.74) as soon as I enable EXPO.

 

[Alan J T]

Just to make sure

 

[ftbscreame15b402e7]

OK so far the following did not work:
1) Updating the Samsung 960 Pro nvme firmware to the latest did not help.
2) Since it was the DRAM light and CPU light staying on I thought could be a training issue. Memory context restore enabled or disabled did not help. Was a long shot anyway.

Pretty sure you're chasing down the wrong rabbit hole. The error code more likely is Ab which is "Setup input wait". Getting the same error when using Ryzen master pbo settings apply. Works fine when setting the same PBO values in BIOS.
Also getting the same error with the newest beta bios (1.74) as soon as I enable EXPO.

You might be correct, it is "Ab" then. I confirmed just now, the LED readout goes 15 -> C5 -> Ab. And then stuck there. If I then press reset it will boot fine.

I do not have expo enabled, using Buildzoid's timings with 1.25V SOC etc. I upgraded to BIOS 174 as soon as the PC booted so unfortunately don't have any experience with the non beta BIOS v16. The BIOS on it was from 2022 and had to upgrade for the X3D chip anyway.

@Alan J T could you please change the thread title if you think its worth it?

I'll try stock ram settings next just to see if it goes away.

 

[lostking20015bf02e6]

OK so far the following did not work:
1) Updating the Samsung 960 Pro nvme firmware to the latest did not help.
2) Since it was the DRAM light and CPU light staying on I thought could be a training issue. Memory context restore enabled or disabled did not help. Was a long shot anyway.

You might be correct, it is "Ab" then. I confirmed just now, the LED readout goes 15 -> C5 -> Ab. And then stuck there. If I then press reset it will boot fine.

I do not have expo enabled, using Buildzoid's timings with 1.25V SOC etc. I upgraded to BIOS 174 as soon as the PC booted so unfortunately don't have any experience with the non beta BIOS v16. The BIOS on it was from 2022 and had to upgrade for the X3D chip anyway.

@Alan J T could you please change the thread title if you think its worth it?

I'll try stock ram settings next just to see if it goes away.

"15 -> C5 -> Ab" is the iconic memory training error.
AMD AGESA is so bugged that almost all errors end up Ab. It should stay at "C5", which differs with different error.

You mentioned you don't enable EXPO. Then try to enable it.
EXPO does couple of things. 1. Frequency; 2. Timings; 3. Voltages (VDD, VDDQ, VDDIO, VSOC, VDDP).
Misconfigure any of them above may lead to error. You'd better leave them in auto and then enable EXPO.

With v174, I recommend to leave VSOC in auto. It is limited out of box. You don't need to worry about that.
There is a bug that, if you need to override VSOC, you have to use "Override" option, instead of "AMD Overclocking". Otherwise, system will not boot (at least for me). The "Override" option is hidden by default. You have to use "Expert" mode under OC menu to unlock this option. @Alan J T , could you report it as a bug.

I own MSI x670e Carbon since last October. I can say BIOS v172 is the best.
v174 is less stable on memory overclock standpoint even with the same VSOC configuration on v172.

 

[Alan J T]

"15 -> C5 -> Ab" is the iconic memory training error.
AMD AGESA is so bugged that almost all errors end up Ab. It should stay at "C5", which differs with different error.

You mentioned you don't enable EXPO. Then try to enable it.
EXPO does couple of things. 1. Frequency; 2. Timings; 3. Voltages (VDD, VDDQ, VDDIO, VSOC, VDDP).
Misconfigure any of them above may lead to error. You'd better leave them in auto and then enable EXPO.

With v174, I recommend to leave VSOC in auto. It is limited out of box. You don't need to worry about that.
There is a bug that, if you need to override VSOC, you have to use "Override" option, instead of "AMD Overclocking". Otherwise, system will not boot (at least for me). The "Override" option is hidden by default. You have to use "Expert" mode under OC menu to unlock this option. @Alan J T , could you report it as a bug.

I own MSI x670e Carbon since last October. I can say BIOS v172 is the best.
v174 is less stable on memory overclock standpoint even with the same VSOC configuration on v172.

Think It may be you Bug will check my BIOS can you drop in a couple of screenshots of the problem.

 

[lostking20015bf02e6]

Think It may be you Bug will check my BIOS can you drop in a couple of screenshots of the problem.

Thank you.
I am running 128GB@5600. I would try to run just 2 sticks and report back regrading manual VSOC.

 

[Alan J T]

I get teh AB code when I'm in the bios try turning the monitor on and off just to check you did not boot to the BIOS
I will sometimes get a hang on B4 from a BIOS exit and have to turn off the PSU power to fix that one

AMD Override is working OK for SOC

Also Added complete build over on PC builds

https://forum-en.msi.com/index.php?threads/show-us-your-machine.162770/post-2182093

 

[Alan J T]

Thank you.
I am running 128GB@5600. I would try to run just 2 sticks and report back regarding manual VSOC.

Yep with Four 32GB sticks, you are likely to need the 1.300v for SOC as it is a tad harder to get stable

 

[lostking20015bf02e6]

Yep with Four 32GB sticks, you are likely to need the 1.300v for SOC as it is a tad harder to get stable

It tunes out there is other issue that cause my system locking up. It is not related to VSOC.
https://forum-en.msi.com/index.php?threads/msi-x670e-carbon-dram-power-down-bug-with-expo.386078/