Fatal1ty X470 Gaming-ITX/ac: Won't enter BIOS

[Zarathustra[H]]

Anyone ever have this issue before?

So I am doing a end of life refresh of my better halfs PC.

It was previously a Ryzen 5 2400G with this motherboard, and it never quite worked right. I had put a DDR4-3466 kit in it, and it never quite worked right from the beginning. Random freezes would happen every once in a while.

I had chucked it up to the original Zen being really picky about RAM and was going to fix it some day, but then we needed to move, and the system went into a box for what was supposed to be a couple of months, but turned out to be like 4 years.

I decided to do a refresh on the system. I'm building it into a new Fractal Terra case, and dropping a 5800x3d in it to last her a few more years. Since she doesn't really play games, I decided my old 6GB Kepler Titan would be OK for a GPU.

(In process pic, cable management to be cleaned up later)

The BIOS flash process required me to install an old Ryzen 5 1600x I had kicking around, and using that CPU flash it to the latest BIOS that supports the 5000 series x3d chips. It was a little inconvenient, but finished without incident.

I was hoping that With the original DDR4-3466 RAM issue would be resolved by using a 5000 series CPU that was more RAM tolerant, but no such luck.

At XMP settings, running a large FFT Prime95 memory test, one of the threads invariably crashed within an hour or so.

I figured I might have to boost the voltage on the RAM controller a little but decided that I didn't really want to fight this RAM, that it was probably defective from the very beginning and to replace it instead.

I ordered a kit of DDR4-3600 C18 RAM, installed it (with great difficulty due to the ITX form factor), went into the BIOS and loaded up XMP settings and then went to do my Prime95 large FFT test.

(Had to unclip and move the fan over to get enough clearance to put the RAM in, after which it was a bloody nightmare to clip the fan back on)

Well, this time it failed even harder than with the previous RAM. In less than 10 seconds Prime 95 calls it quits and crashes.

So I'm thinking it's probably a matter of boosting the voltage to the memory controller then, and go to load up the BIOS, but I can't get the damn BIOS to open.

Hitting DEL during post just makes the screen black, and then it sits there doing nothing. After trying multiple times, I clueing pulling all power and giving it a time out and trying again, I decided to clear the CMOS.

After clearing the CMOS I still can't get into the BIOS.

The funny thing is, the system will still boot. Now that the CMOS is cleared the RAM is at default speeds of 2166 which is obviously undesirable.


Has anyone had an issue like this before? In my 30 years of doing this, I have never encountered a system that refused to load the BIOS.

I'm wondering if maybe the decade old GPU could be the problem, but I have no idea.

I have a few old 630's and 720's, and maybe a 1030 kicking around I could test with to see if that is the case, but other than this I am a little stuck, other than just replacing the motherboard, which would completely defeat the purpose of the end of life upgrade...

(I started this as a quick drop in CPU upgrade, and now I have already replaced the RAM and am talking about replacing the motherboard. If I was going to do all of that I would have moved her to an AM5 system.... I wonder if Amazon will take back an opened CPU )

Anyway, appreciate any suggestions here, as I am kind of stuck with issues unlike anything I have ever seen before.

 

[Darunion]

Wonder if the bios chip is actually trashed. ESD spike could have hit it in handling. Just my off the wall random stab in the dark. Not being able to get into the bios makes me think its that chip or a really bad power supply rail on the motherboard. Might be able to find a pre-flashed chip out there or even a donor board that someone is getting rid of for another failure reason.

 

[LigTasm]

ASRock has a restart to UEFI program floating around, I would try that. The fast boot option is usually on by default and if CSM is off it goes straight for the boot drive and doesn't even give you a chance to hit the button.

This is on your boards support page:

 

[Dopamin3]

ASRock has a restart to UEFI program floating around, I would try that. The fast boot option is usually on by default and if CSM is off it goes straight for the boot drive and doesn't even give you a chance to hit the button.

Yeah this should work, alternatively it's a native function of Windows. From either the Sign on screen or the Start menu, select Power (

) > hold Shift while selecting Restart. Select Troubleshoot > Advanced options > UEFI Firmware settings. It works on Windows 10 and 11 but not sure on older versions.

I feel like most AM4 CPUs can stabilize 3600MHz, but some just can't. What is the board defaulting to for vSOC voltage? You can use something like HWinfo64 to check. Maybe try around 1.1V - 1.15V to see if that helps stabilize it.

 

[Zarathustra[H]]

Wonder if the bios chip is actually trashed. ESD spike could have hit it in handling. Just my off the wall random stab in the dark. Not being able to get into the bios makes me think its that chip or a really bad power supply rail on the motherboard. Might be able to find a pre-flashed chip out there or even a donor board that someone is getting rid of for another failure reason.

Thanks for the suggestion. It wasn't really touched between first boot with the new RAM (when it worked, and I could get in there and adjust the settings) and later boots though.

ASRock has a restart to UEFI program floating around, I would try that. The fast boot option is usually on by default and if CSM is off it goes straight for the boot drive and doesn't even give you a chance to hit the button.

This is on your boards support page:

View attachment 607074

Yeah this should work, alternatively it's a native function of Windows. From either the Sign on screen or the Start menu, select Power (View attachment 607080) > hold Shift while selecting Restart. Select Troubleshoot > Advanced options > UEFI Firmware settings. It works on Windows 10 and 11 but not sure on older versions.

Unfortunately this is not going to do the job, since there is no Windows on this machine.

I don't think it is the triggering of the BIOS that is the issue. The POST screen disappears when you hit DEL (or F2) and it looks like it is trying to load the BIOS, but it just doesn't display anything on screen.

I'm wondering if maybe the decade old GPU could be the problem, but I have no idea.

I have a few old 630's and 720's, and maybe a 1030 kicking around I could test with to see if that is the case

It looks like my hunch here may have been correct.

I happened to have an old Radeon R5 230 kicking around, so I popped that in, and now I can get into the bios no problem.

I have heard of some older GPU's having issues with BIOS when "Above 4G Decoding" is enabled. I wonder if that is what is going on with the old Titan. I've just never seen it before.

I might just have to pick her up some lower end GPU to go along with it. Maybe an RX 7600...

I feel like most AM4 CPUs can stabilize 3600MHz, but some just can't. What is the board defaulting to for vSOC voltage? You can use something like HWinfo64 to check. Maybe try around 1.1V - 1.15V to see if that helps stabilize it.

Unfortunately I am a little rusty when it comes to BIOS settings labels (and they seem to vary from manufacturer to manufacturer).

In the BIOS menu of the Asr4ock I have the following choices:

Do any of them look like the vSOC you are talking about? Is SoC/Uncore the one?

The description also calls it VDD_SOC.

Appreciate the assist!

 

[Dopamin3]

Thanks for the suggestion. It wasn't really touched between first boot with the new RAM (when it worked, and I could get in there and adjust the settings) and later boots though.



Unfortunately this is not going to do the job, since there is no Windows on this machine.

I don't think it is the triggering of the BIOS that is the issue. The POST screen disappears when you hit DEL (or F2) and it looks like it is trying to load the BIOS, but it just doesn't display anything on screen.




It looks like my hunch here may have been correct.

I happened to have an old Radeon R5 230 kicking around, so I popped that in, and now I can get into the bios no problem.

I have heard of some older GPU's having issues with BIOS when "Above 4G Decoding" is enabled. I wonder if that is what is going on with the old Titan. I've just never seen it before.

I might just have to pick her up some lower end GPU to go along with it. Maybe an RX 7600...




Unfortunately I am a little rusty when it comes to BIOS settings labels (and they seem to vary from manufacturer to manufacturer).

In the BIOS menu of the Asr4ock I have the following choices:

View attachment 607139

Do any of them look like the vSOC you are talking about? Is SoC/Uncore the one?

The description also calls it VDD_SOC.

Appreciate the assist!

WIthout Windows, you can boot to a live Linux distro (e.g. Ubuntu) and issue "systemctl reboot --firmware-setup" to see if that works. That's definitely a strange issue.

And yea when I mentioned vSOC that would correspond to SoC/Uncore OC Voltage in that BIOS.

 

[LigTasm]

Do any of them look like the vSOC you are talking about? Is SoC/Uncore the one?

The description also calls it VDD_SOC.

Appreciate the assist!

My bad, I thought there was Windows on it. Yes, SOC uncore is what you want. I'm a little confused as to what CPU you're actually using, but safe limit for most AM4 chips if you're not going for a hero run is 1.200V.

 

[Zarathustra[H]]

My bad, I thought there was Windows on it. Yes, SOC uncore is what you want. I'm a little confused as to what CPU you're actually using, but safe limit for most AM4 chips if you're not going for a hero run is 1.200V.

Thank you.

It's a Ryzen 7 5800x3d, if that makes a difference.

 

[LigTasm]

Thank you.

It's a Ryzen 7 5800x3d, if that makes a difference.

I'm shocked you're even having an issue. My 79$ B450 board has run 4000mhz with no problem on every chip except my 1800X which simply cannot handle memory speeds that high.

 

[Zarathustra[H]]

I'm shocked you're even having an issue. My 79$ B450 board has run 4000mhz with no problem on every chip except my 1800X which simply cannot handle memory speeds that high.

I blame this Asrock board. Its my first Asrock board and might be my last

I'm going to see what pumping this voltage up a little will do.

 

[Dopamin3]

I'm shocked you're even having an issue. My 79$ B450 board has run 4000mhz with no problem on every chip except my 1800X which simply cannot handle memory speeds that high.

Motherboard plays an insignificant role in stabilizing high frequency memory. That's 99% your CPU's integrated memory controller. I've used AM4 boards from Asus, MSI, Gigabyte, and Asrock with the same results when using the same CPU across them. I guess AGESA plays a small role too because historically some updates have improved memory compatability.

On that 4000MHz are you running 1:1: infinity fabric:uclk:memclk? You often negate performance benefits if you unlink it.

I blame this Asrock board. Its my first Asrock board and might be my last

I'm going to see what pumping this voltage up a little will do.

I'd blame the CPU for the RAM frequency in both scenarios. AMD spec is 3200MHz the 5800X3D. 3600MHz usually works on Zen 3 but it's not a guarantee.

The 2400G, which with a leading "2" in the model you might think it's based on Zen+ architecture (e.g. 2700X), nope you're looking at an original Zen chip. Good luck stabilizing 3466MHz on that, it was very uncommon and AMD spec was 2933MHz.

I guess the BIOS issue you could blame on Asrock? That's just weird. Or like LigTasm suggested the issue could be caused by fastboot on / CSM off. I've never ran that configuration so don't know how it affects getting to the BIOS.

 

[Zarathustra[H]]

Motherboard plays an insignificant role in stabilizing high frequency memory. That's 99% your CPU's integrated memory controller. I've used AM4 boards from Asus, MSI, Gigabyte, and Asrock with the same results when using the same CPU across them. I guess AGESA plays a small role too because historically some updates have improved memory compatability.

On that 4000MHz are you running 1:1: infinity fabric:uclk:memclk? You often negate performance benefits if you unlink it.


I'd blame the CPU for the RAM frequency in both scenarios. AMD spec is 3200MHz the 5800X3D. 3600MHz usually works on Zen 3 but it's not a guarantee.

The 2400G, which with a leading "2" in the model you might think it's based on Zen+ architecture (e.g. 2700X), nope you're looking at an original Zen chip. Good luck stabilizing 3466MHz on that, it was very uncommon and AMD spec was 2933MHz.

I guess the BIOS issue you could blame on Asrock? That's just weird. Or like LigTasm suggested the issue could be caused by fastboot on / CSM off. I've never ran that configuration so don't know how it affects getting to the BIOS.

Well, manually setting the vSOC to 1.15v seems to have possibly solved the issue.

I wonder if the BIOS just has some sort of bug and sets too low of a voltage in auto mode.

Prime95 (or mprime as it is called under Linux) has been running for several minutes now without a crash, which is a huge improvement compared to where it was before.

I'll need to run it for a couple of hours without a problem before I have confidence in it, but this is a huge improvement.

Fingers crossed.

Thanks for all the help guys.

Now I just need to figure out which low cost, reasonably modern (but not totally ineffectual) GPU to buy since the old 6GB Titan proved to be a problem on this motherboard.

I'm leaning towards an RX 7600, but I wonder if I can get a better bang for the buck. Any suggestions in the mid to low end space?

 

[motqalden]

This is sadly how the quick boot for asrock works and why they have that utility. FYI grub boot menu also has an option to restart into uefi

 

[Zarathustra[H]]

This is sadly how the quick boot for asrock works and why they have that utility. FYI grub boot menu also has an option to restart into uefi

I'm used to QuickBoot features with Windows, but usually they don't even give you the opportunity to hit del to enter BIOS.

This one does (at least with Linux) but when you press the del key, it just gives you a black blank screen.

 

[Nobu]

Fastboot skips initialization of devices like the gpu, and lets the OS handle it, iirc. If the efi fails to init the gpu when you interrupt boot, it would appear to hang.

 

[Nobu]

I usually disable fastboot etc if I don't have windows on the machine. It's buggy, and linux is fast anyway.

 

[learners permit]

I've had some early Ryzen boards that were actually more stable with the bios switch for SOC/UNCORE OC mode set to disabled even though the board was actually ocing the SOC.

 

[Zarathustra[H]]

I'm going to call 4 hours of large FFT's in Prime95 a succes, at least for now.

Thanks for all the suggestions!

 

[Zarathustra[H]]

I've had some early Ryzen boards that were actually more stable with the bios switch for SOC/UNCORE OC mode set to disabled even though the board was actually ocing the SOC.

Interesting. This one seems to default to Enabled even after resetting the BIOS to defaults, which is kind of weird.

It seems stable now at 1.15v, but I may try setting it to Disabled at some point and see how it does.

 

[Dopamin3]

I wonder if the BIOS just has some sort of bug and sets too low of a voltage in auto mode.

I wouldn't consider it a bug. 1.1v vSOC is enough for most people to run 3200MHz XMP, and sometimes even 3600MHz on Zen 3. Every CPU is going to have different capabilities in terms of memory controller quality.

Now I just need to figure out which low cost, reasonably modern (but not totally ineffectual) GPU to buy since the old 6GB Titan proved to be a problem on this motherboard.

I'm leaning towards an RX 7600, but I wonder if I can get a better bang for the buck. Any suggestions in the mid to low end space?

RX 7600 would be the way to go IMHO. Retailing around $269, you're not going to find many new cards that can outperform in rasterization in that price class. It's better than the RTX 3060 and Arc A770 which would be the closest competitors.

It seems stable now at 1.15v, but I may try setting it to Disabled at some point and see how it does.

Don't bother. You know it's not stable at 1.1V and it works at 1.15V. Disabling that could make it run unnecessarily higher, or still lower to the point where it isn't stable. Either way, it's a waste of time. I wouldn't bother fine tuning it since 1.2V is generally the accepted safe vSOC on Zen 3 but if you want to really fine tune it you can try 1.125V or whatever.

If anything, next you want to do the curve optimizer and set a negative offset to the CPU voltage. It isn't officially supported on the 5800X3D, there is a workaround but it appears to be Windows only so I'm not sure how that would work if you're using Linux. This will typically improve thermals and allow for a little more performance. But even a 5800X3D at the "stock" settings with 3600MHz RAM and a mid-tier GPU, you're realistically always going to be GPU bottlenecked in games and it won't really give much of an uplift in typical tasks.