[SOLVED] B350M and Ryzen 7 2700x won't post even after BIOS update

May 17, 2021
35
1
45
Replaced my old B350m Bazooka with a Mortar Arctic due to a broken PCI slot and only having dedicated GPU.
Obviously when I first put the CPU in it didn't post, needed to update the BIOS. So I bought an A6-9500 and updated to the latest stable bios (AKM), checked out the speed (abysmally slow), and then shut down to swap CPUs. Didn't post. Swapped again and updated to latest beta (AI), and it still didn't work. So I then took my old Ryzen 3 1800x and tried that with both BIOS versions, didn't post. I went to the MSI website and checked the compatibility section for my mobo and cpu, I needed BIOS AC, flashed that and tried again with both chips, still won't post.

To update the bios I have been:
  1. Flash usb to FAT32
  2. extract the .aio file to the usb
  3. post, bios, m-flash, reboot, select aio file
  4. wait for 100% and reboot
I have taken out the CMOS battery for about an hour and no difference.
I have let the computer sit mid post for an hour and no difference.
I have swapped ram, both speed and slot, and no difference (2x 8GB 2666, 2x 16 GB 3200)
I have tried 1 stick, 2 stick, 4 stick in all combinations
I have swapped PSUs

not too sure what it could be at this point, I can see the different bios versions displayed in bios.
Any help would be appreciated,
Thank you,
Mark
 
Last edited:
Solution
I think I misunderstood the file naming and extension syntax with these files. The name is always 7A37AMS, but the file extension shows the file version, eg
7A37AMS.ACO
7A37AMS.AIO
7A37AMS.AKS

Currently BIOS is correctly showing version ACO, which is the version the MSI website lists as compatible with the Ryzen 7 2700x. I can still try to update to AIO or AKS, but I doubt it would make a difference at this point...
I think every version should work with an 1800x, all the way back to the very first (except vK0). It's a B350 board, after all. That's what makes no sense: why it works with the A6 but not a Zen1 CPU. And the thing is, you should have been able to make all the updates with the 1800X meaning the A6 wasn't necessary...
May 17, 2021
35
1
45
Since you can get into the bios, how far along in the POST does it get? What code you have displayed?
The MSI board (Mortar Arctic) has debug lights, not a code display.
Normal operation has each light (CPU,RAM,VGA,BOOT) come on for about half a second then turn off, then the next down does the same. If something is wrong the malfucntioning component's respective light stays on.
In my case the CPU light comes on for about 2 seconds, then turns off, but no other lights come on, almost like POST gets frozen while checking the CPU.
 
Last edited:
May 17, 2021
35
1
45
Where are you getting your BIOS files from?
Latest stable is 1J and the latest beta is 1LQ
https://www.msi.com/Motherboard/sup...www.msi.com/Motherboard/support/B350M-BAZOOKA

Perhaps you didn't get a 350m bazooka as the replacement?

Yes... I dissapoint myself daily. I left out the small detail that I replaced the Bazooka with a Mortar Arctic, not another Bazooka. Whoops...
Also I switched around the latest stable and beta release names, beta is AKM and stable is AI.

I did use the MSI website: https://www.msi.com/Motherboard/support/B350M-Mortar-Arctic#support-cpu
 
Yes... I dissapoint myself daily. I left out the small detail that I replaced the Bazooka with a Mortar Arctic, not another Bazooka. Whoops...
Also I switched around the latest stable and beta release names, beta is AKM and stable is AI.

I did use the MSI website: https://www.msi.com/Motherboard/support/B350M-Mortar-Arctic#support-cpu
The beta (AKm) is really intended for a Ryzen 3000 CPU. It's probably not going to gain you anything for a 2000 CPU and it may even make reverting to an old BIOS difficult so you should avoid it. I had a lot of difficulty reverting my B450 Mortar from it's Ryzen 3000 BIOS.

Version AI seems the best, but I'm confused when you say it 'still didn't work'. What happens exactly when you try to start the system? How do you know you got a successful update of BIOS? Are you able to get into BIOS screens at least?

There's a note that: " MSI strongly recommend to update AMD chipset driver 18.10xxxxxxx or latest version before update BIOS." Did you do that at the very first? If you can still get into BIOS but it just won't boot into Windows that may be the problem.
 
May 17, 2021
35
1
45
Version AI seems the best, but I'm confused when you say it 'still didn't work'. What happens exactly when you try to start the system? How do you know you got a successful update of BIOS? Are you able to get into BIOS screens at least?
When I try to start they computer it stops mid post. Normally if something was wrong the debug light for that component would stay lit. For instance, a normal post has each light(CPU, RAM, VGA, BOOT) flash as it tests the component, then the next one flashes as it gets tested. However, if I forgot to insert RAM, the CPU light would flash, then RAM light would stay on, no other lights.
What is happening is the CPU debug light stays on for about 2 seconds (normally half a second), and then it goes out as if it completed the check, but it doesn't move to the next component as it should. So I end up with no debug lights lit, but my computer is still in POST (no display, all fans at full speed). I do not have a buzzer on the mobo to hear if a beep code is being played.

This is the same for both Ryzen CPUs, however with the A6 the computer is able to complete POST and boot normally. I have not updated the chipset driver, and will be doing that immediately. I must have overlooked that notice. I will post another reply with the results, I hope this is it and it's user error...
 
May 17, 2021
35
1
45
There's a note that: " MSI strongly recommend to update AMD chipset driver 18.10xxxxxxx or latest version before update BIOS." Did you do that at the very first? If you can still get into BIOS but it just won't boot into Windows that may be the problem.

Unfortunately updating the chipset drivers seemed to not have any effect. Attatched is a video of the debug light sequences with the Ryzen CPUs (both do the same), and the A6, which gets into BIOS.

https://drive.google.com/drive/folders/17ZtBv7oapmS0xDJVxyYYRdeQz_vl3tfK?usp=sharing

The dark picture with both CPU and RAM lights on is after quickly pressing the restart button 10x with the Ryzen CPU in. I read somewhere this can sometimes get you past BIOS errors, but this is what happens instead.
 
Unfortunately updating the chipset drivers seemed to not have any effect. Attatched is a video of the debug light sequences with the Ryzen CPUs (both do the same), and the A6, which gets into BIOS.

https://drive.google.com/drive/folders/17ZtBv7oapmS0xDJVxyYYRdeQz_vl3tfK?usp=sharing

The dark picture with both CPU and RAM lights on is after quickly pressing the restart button 10x with the Ryzen CPU in. I read somewhere this can sometimes get you past BIOS errors, but this is what happens instead.
I didn't think it would after reading the post previous to this.

When you POST into BIOS with the A6 processor, can you see what version BIOS you're on?
 
May 17, 2021
35
1
45
When you POST into BIOS with the A6 processor, can you see what version BIOS you're on?

I uploaded the picture of my BIOS onto the drive link. Weird thing is, looking at it, I see version AMS, instead of AC. Frankly I don't know if this is normal, the date seems to be right. I also left my CMOS battery out, but I don't think that would roll back the BIOS version, right?

I'll try to update the bios while I wait for your reply. I'll try version AI (stable), and see if the version updates. I remember seeing the right version when I did that one...
 
May 17, 2021
35
1
45
When you POST into BIOS with the A6 processor, can you see what version BIOS you're on?
I think I misunderstood the file naming and extension syntax with these files. The name is always 7A37AMS, but the file extension shows the file version, eg
7A37AMS.ACO
7A37AMS.AIO
7A37AMS.AKS

Currently BIOS is correctly showing version ACO, which is the version the MSI website lists as compatible with the Ryzen 7 2700x. I can still try to update to AIO or AKS, but I doubt it would make a difference at this point...
 
I think I misunderstood the file naming and extension syntax with these files. The name is always 7A37AMS, but the file extension shows the file version, eg
7A37AMS.ACO
7A37AMS.AIO
7A37AMS.AKS

Currently BIOS is correctly showing version ACO, which is the version the MSI website lists as compatible with the Ryzen 7 2700x. I can still try to update to AIO or AKS, but I doubt it would make a difference at this point...
I think every version should work with an 1800x, all the way back to the very first (except vK0). It's a B350 board, after all. That's what makes no sense: why it works with the A6 but not a Zen1 CPU. And the thing is, you should have been able to make all the updates with the 1800X meaning the A6 wasn't necessary.

I'm wondering if there's a defect in the motherboard that is only triggered with full CPU's, possibly related to enabling all 16 PCIe lanes since APU's only use 8 lanes to the the first PCIe slot. This starts to make sense considering you had a defect in the PCIe slot of the board you're replacing so it might have transferred to one of the GPU contacts; it happens like that sometimes.

In fact, it might be the GPU that's causing this: do you have another GPU to try? anything at all would do. You might also carefully examine the card-edge connector of both the mother board and the GPU. It's tiny and hard to see so use a bright light and magnification. Look for anything that's not right, or doesn't look like the others, on both sides of the board.

I'd definitely consider an update to 7A37vAI....which is in a file named E7A37AMS.AI0. That's the latest BIOS built for 2000 series. Don't go to AK0 as it's a Ryzen 3000 BIOS and might only make things worse. But I agree, it won't help since the 1800x should work and it isn't. The problem has to be elsewhere.
 
Last edited:
Solution

Karadjgne

Titan
Ambassador
Bios updates should be done in order. ALWAYS read the bios notes first. Those older mobo's had smaller, cheaper bios chips that barely have enough room in them for a single update data plus instructions. By skipping updates and just installing the latest, you also skip most of those instructions too.
For example, the first update might change line XYZ to 123. The following update might change 123 to ABC. But because of size restraints, there is no included additional instructions to also change XYZ to ABC.

Then comes the data for all the 3000 cpus, which is more than those smaller bios chips can handle with also containing 1000 and 2000 values. So when installing a 3000 capable bios, it can and will delete a lot of the 1000/2000 irrelevant data,

Meaning the 1800x will no longer work.

You'll need to start over, choose a bios way back on the list, like the original bios version, and pray it takes it, or you've basically bricked the bios by skipping updates.
 
...
Meaning the 1800x will no longer work.
...
Except... I have a B350 Mortar (not the Arctic, but otherwise identical to OP's) and I jumped all the way from a very early BIOS to the rev just before the Ryzen 3000 rev and my 1700 worked perfect. In testing BIOS and trying to find one with a certain AGESA I also jumped back several revs...then back forward several revs again. It always worked perfect, never a failed POST. MSI has no notes about updating in progression, only notes about losing support for Bristol Ridge and to update chip set drivers. And then this doesn't explain why his A6 continues to work well, nor why he's getting such odd debug LED response on failed POST.

Your concerns may be valid for Asrock, who use a highly non-standard BIOS structure.
 
Last edited:

Karadjgne

Titan
Ambassador
jumped all the way from a very early BIOS to the rev just before the Ryzen 3000 rev
Exactly. If there's nothing that actually says that you need a certain update first, then you are clear to move onto the next, or next, or next etc. But gotta read the notes first to be certain. And you stopped prior to the 3000 series.

- MSI strongly recommend to update AMD chipset driver 18.10xxxxxxx or latest version before update BIOS.

It must install nvidia graphic driver ver 391.01 or latest version when using AMD Raven Ridge CPU and nVidia graphic card.

The 2 notes listed that might apply, but I suspect that those only apply to APU's. There's a third stating not to update to a higher version if using Windows 7 as it's not supported.

I'd still drop back to a prior bios, like 7A37vAC which sets Agesa and compatibility with the Bristol/Raven series, then update to newer.
 
May 17, 2021
35
1
45
I think every version should work with an 1800x, all the way back to the very first (except vK0). It's a B350 board, after all.

In fact, it might be the GPU that's causing this: do you have another GPU to try?
Look for anything that's not right, or doesn't look like the others, on both sides of the board.

I'd definitely consider an update to 7A37vAI....which is in a file named E7A37AMS.AI0. That's the latest BIOS built for 2000 series. .

Welp I'm mainly trying to get my R7 2700x to work, but yes, the R3 1800x hasn't worked either.

I don't have another GPU to try, but this GPU has been in two separate test benches (not mine) and run flawlessly, so if it is still related to the GPU it's another MOBO with a bad PCI slot. I have noticed my PCI running in x8 even though I deactivated the A6's integrated GPU, I'll try switching that to x16 and see if I run into problems. Also, if it were a problem with the GPU I would imagine the debug light would catch there, instead of the weird half CPU fault its giving.

I've run AI already with no success, In fact I should be running it now. Still no dice.

Well it seems i either have the option of using x8 or X4+X4. I read on another TH thread it's not uncommon with the 350M line, and it could be the A6 throttling as well. I also don't know if the gpu is using all lanes under load, just that bios has PCI slot 1 listed as x8.
 
Last edited:
May 17, 2021
35
1
45
You'll need to start over, choose a bios way back on the list, like the original bios version, and pray it takes it, or you've basically bricked the bios by skipping updates.

It sounds like this is a pretty controversial take, but I'll try it. At this point I'm grasping at straws just trying to get my computer back.

It doesn't guarantee anything, but I've already skipped from the stock BIOS to AKS, the latest beta, downgraded to AI, then down again to AC, so I suspect I will be able to again downgrade to the beginning and go up the list.
 
May 17, 2021
35
1
45
Well I've taken everything out of the case and reassembled on the carpet, no difference. I was thinking maybe the case was somehow shorting the x16 PCI lanes when a non APU was used, but it doesn't seem to be the case.

Post without GPU produces no differences
Post without RAM illuminates both CPU and RAM lights, both stay on until power is cut.
Post with no CPU illuminates only the CPU light, stays on until power is cut.

Could this point to it being more of a problem with RAM than the GPU or CPU? I doubt it, but figured I'd ask. Either way it seems post isn't getting to the GPU to test it / the GPU has nothing to do with my problems so far...
 
May 17, 2021
35
1
45
If anything it really seems to be a mobo issue at this point... There's a small PC repair shop around the corner from my house, the owner has let me use his test bench and spare parts before. I'll have to see if I can swap everything to a different motherboard to see if it will post.
 
...
Well it seems i either have the option of using x8 or X4+X4. I read on another TH thread it's not uncommon with the 350M line, and it could be the A6 throttling as well. I also don't know if the gpu is using all lanes under load, just that bios has PCI slot 1 listed as x8.
Keep in mind I'm just throwing out an idea about it being the GPU. I'm speculating on a reason the A6 still works while two full-up CPU's fail POST in a very peculiar fashion. Other differences include that the CPU's use PCIe gen 3 while the APU is limited to gen 2 but how that could contribute I haven't an idea.

The X4+X4 mode PCIe bi-furcation feature is for running certain types of PCIe add-in cards such as an NVME adapter. So you could use the iGPU in the A6 for your display and have three PCIe gen 2 NVME's (one on the motherboard, two in an adapter in the PCIe slot normaly used for a dGPU) for extremely fast data transfers, which would be very useful in some applications.

Otherwise, APU's equip only 8 of the PCIe lanes to the GPU PCIe slot, the other 8 are reserved for the iGPU even if you don't use it. CPU's equip all 16 lanes.

That board also has a 2nd PCIe x16/4 slot, if you haven't yet you might try the GPU in that slot with one of the CPU's. It's not a desireable operating mode (as it's only x4 / gen 2) but it might be informative if it boots to BIOS at least. I believe the only restriction is you can't be using certain of the SATA ports or it steals it's 4 PCIe lanes.

One question I have is why you got the A6 to do the updates in the first place since you already had an 1800x which should have done the job. Did it not work when you first got the board? Updating to the vAKS beta BIOS was a mistake (since you don't have a Ryzen 3000) but both the 1800x and the 2700 should still have worked. I know this because it did with my 1700 when I also tried that version on my B350 mortar (I disliked how it changed BIOS interface so immediately went back). So I was hopping around with BIOS rev's too but I always used a 1700 to do it with.
 
Last edited:

Karadjgne

Titan
Ambassador
Bios updates are just that, an update. New data to replace old or buggy, additional data, instructions to install both. Drea, when you updated your bios, you stopped before adding the additional 3000 series, so your bios contains the full version of zen/zen+ needs. But the Zen2 addition, the cpu data, ram tables, installation instructions etc is too large for those smaller bios chips. Your 1700 still worked because your bios had the full Zen bios. Op's is shot because half or more of the Zen/Zen+ bios is missing, replaced by the Zen2 bios and data. For all intents and purposes the bios won't recognize a zen/zen+ cpu, but the A6 works because the APU's were a halfway deal, pretty much seperate bios files from a regular 1000/2000 series CPU.

If you look at ASR bios notes, they'll tell you, Stop! Do Not install this if you have Raven/Bristol. There's No going back. You were entirely correct in saying it was a mistake to install the latest, hopefully Op can either revert to prior bios versions or install a zen/zen+ only bios and replace all that the Zen2 files deleted.

Like you I dislike ASR's bios update procedure, but with the bios chip size restrictions on those older/lesser chipset boards, having the split/definitive updates and warnings stands a chance of preventing the mistake Op made of including the Zen2 bios files and messing up the original architecture needs.

My belief is that this is less Op's fault, and More MSI lackluster support on such a difficult and potentially hazardous procedure.
 
Bios updates are just that, an update. New data to replace old or buggy, additional data, instructions to install both. Drea, when you updated your bios, you stopped before adding the additional 3000 series, so your bios contains the full version of zen/zen+ needs. But the Zen2 addition, the cpu data, ram tables, installation instructions etc is too large for those smaller bios chips. Your 1700 still worked because your bios had the full Zen bios. Op's is shot because half or more of the Zen/Zen+ bios is missing, replaced by the Zen2 bios and data. For all intents and purposes the bios won't recognize a zen/zen+ cpu, but the A6 works because the APU's were a halfway deal, pretty much seperate bios files from a regular 1000/2000 series CPU.

If you look at ASR bios notes, they'll tell you, Stop! Do Not install this if you have Raven/Bristol. There's No going back. You were entirely correct in saying it was a mistake to install the latest, hopefully Op can either revert to prior bios versions or install a zen/zen+ only bios and replace all that the Zen2 files deleted.

Like you I dislike ASR's bios update procedure, but with the bios chip size restrictions on those older/lesser chipset boards, having the split/definitive updates and warnings stands a chance of preventing the mistake Op made of including the Zen2 bios files and messing up the original architecture needs.

My belief is that this is less Op's fault, and More MSI lackluster support on such a difficult and potentially hazardous procedure.
A. Don't compare ASR (Asrock??) BIOS to anyone else's; Asrock completely ignores AMD's recommendations for how to construct the BIOS. So much so that TheStilt couldn't figure it out in order to create some modded BIOS's for early Zen2 support back in the day (he did, however, for gigabyte, MSI and Asus at least...I used one for a while!). Not that it has any bearing on this, it just illustrates there's a lot more going on and that makes an unfair comparison for much of anything.

B. OP DID update the chipset drivers...go back a few posts and you'll see. But I'm not all that certain it really matters at this point because it should really only affect booting into the OS in UEFI mode...he can't even POST to BIOS screens.

If anyone's to blame IMO it would be AMD. They should have gone full-Intel and not supported so many generations of CPU and APU in the exact same socket as that leads to BIOS confusion. I am, of course, being sarcastic...but that's still what's so unique about this.

But even so, I'm still curious what compelled OP to get an A6 CPU way back in the beginning when he had an 1800x in hand that would have been adequate for the job of updating BIOS all the way through.
 
Last edited:

Karadjgne

Titan
Ambassador
If anyone's to blame IMO it would be AMD. They should have gone full-Intel and not supported so many generations of CPU and APU in the exact same socket as that leads to BIOS confusion.
Yes! Lol. Totally agree. New architecture, New platform. And even though 6-9thgen Intel is basically all skylake architecture, there's enough changes to warrant a new socket for 8/9.

Amd tried to please everyone, ended up only doing so half-way. It's still a mess. Should even call the Zen3 cpus something other than Ryzen, since that architecture is somewhat different again for Zen.

Agreed. I'd have left the 1800x in there, updated to AC, swapped cpus to the 2700x.

But there was an issue to start with. First flashed AKM, was abysmally slow with the A6, then flashed AC on top, and no joy with any cpu. Broke an 'Armored' pcie slot? So far, the A6 is the only thing that works, none of the non-APU are recognised.

Might need to flash by dos prompt from bootable usb instead of mflash.
 
....
Might need to flash by dos prompt from bootable usb instead of mflash.
MSI on the MSI forums has the UEFI utility for doing just that. It's extremely easy to use: just unpack onto a fresh formatted (FAT32) USB stick and set up BIOS to boot to a flash drive. Then rename the BIOS you want to burn to one the script will recognize, put it in the root and re-boot to that flash drive.

It will completely erase every block in the BIOS and repopulate every block with the new BIOS, so it's a complete 100% update. I had to use it to burn my modded BIOS's, and I also used it to revert from a BIOS MSI didn't want us to revert from.

OP: here's the link:


In the first post Svet (the author) cautions won't work for B350 boards but if you read through the thread you'll see (towards the end) one or two people have used it on their Mortar Arctic's. You might need to create user account and login to download it. You'll also need something to un-RAR (7-zip works great).
 
May 17, 2021
35
1
45
Keep in mind I'm just throwing out an idea about it being the GPU. I'm speculating on a reason the A6 still works while two full-up CPU's fail POST in a very peculiar fashion.
That board also has a 2nd PCIe x16/4 slot, if you haven't yet you might try the GPU in that slot with one of the CPU's. It's not a desireable operating mode (as it's only x4 / gen 2) but it might be informative if it boots to BIOS at least. I believe the only restriction is you can't be using certain of the SATA ports or it steals it's 4 PCIe lanes.

One question I have is why you got the A6 to do the updates in the first place since you already had an 1800x which should have done the job. Did it not work when you first got the board? Updating to the vAKS beta BIOS was a mistake (since you don't have a Ryzen 3000) but both the 1800x and the 2700 should still have worked. I know this because it did with my 1700 when I also tried that version on my B350 mortar (I disliked how it changed BIOS interface so immediately went back). So I was hopping around with BIOS rev's too but I always used a 1700 to do it with.

I did try swapping pci slots last night with the computer breadboarded. No difference.

Two things here: the 1800x is listed as needing BIOS version A5, while I think the board came with A0 or at least below A5, so it does make sense that the mobo might not have worked with it from the get go. Also, I just realized my Ryzen 3 has been missing for about a week, as I was just cleaning the thermal paste off of the "Ryzen 3 1800x" I had and it turns out its a Ryzen 5 1400. It must have gotten mixed up while I was picking everything up from the computer repair shop the first time. (Asked if he could do the update, said yes, called a day later and said I could wait a week for the cpu to come in or I could do it myself. Did it myself.). So that might explain some of the issues...
Directly to your question of why, I read from multiple forums (here and elsewhere) and a couple websites the Mortar Arctic had guaranteed support for any of the A series. Also it had an integrated GPU, which I wanted anyways since this dedicated board is getting old. Also I tried the 1800x in it as it was the only CPU I had at the time and it didn't work (this was the day or two before the CPU mix up), so I figured I should grab a cheap A series and try it. At least my computer runs for now?
NVM, the R5 1400 should work from A1 and up, so that probably still isn't an issue.
 
Last edited: