Motherboard: Gigabyte GA-AB350-Gaming 3, UEFI F9d (currently, working)
CPU: AMD Ryzen 7 1700
RAM: G.Skill Flare X For AMD 2400MHz (QVL'ed), 2x8GB
GFX: Asus DUAL-GTX1060-3GB O.C
Operating System installed: Irrelevant for UEFI issue, but currently Windows 10 Pro 64bit
Drivers installed: Irrelevant, cannot boot to Windows most of the time with erroneous setup
Power Supply: Corsair 700GS, operating just quite nicely
As this issue seems to be with AGESA, I don't really know which forum would be the correct one... As it has a lot to do with the CPU, I decided to put it up here. MOD's please move this to elsewhere, if I'm posting in the wrong forum!
Simply: when I upgrade my motherboard's UEFI to anything newer, I start to get "No VGA"-error on start. It goes away exactly when I reflash UEFI with older version.
Sometimes I am able to boot with newer UEFI, but mostly not. Clearing CMOS usually helps, but if I change practically anything in the UEFI I get "NO VGA" light on the motherboard.
It might be enough to make the system completely inoperable by simply changing the boot order in UEFI, so the issue has to lie much deeper than just the UEFI itself.
Affected version of UEFI: Gigabyte's F10, F20, F21, F22b
Difference between the many versions of the UEFI:
As far as I know, AGESA. F9d apparently uses the older AGESA 18.104.22.168a, when F10 comes with 22.214.171.124a.
F20 and F21 then comes with 126.96.36.199a (I was told that this is complete rewrite of the AGESA?).
F22b comes with AGESA 188.8.131.52a + SMU 30.67.0 for Raven Bridge.
If I have any of the affected versions flashed, I *might* get the PC to POST. Most likely not. If I do and if I do enter UEFI, I probably fail changing *any* settings. But if I want to be very sure to get a fail, all I need is to tell UEFI to use my XMP Profile for the RAM. Guaranteed to fail after "Save & reset".
Now, even in the (very rare) case I do get to boot to Windows 10, the system is very unstable. Just like with broken graphics card or bad RAM. However, I know that my RAM is not bad (this is already once sent back to G.Skill and replaced, after which I did run extensive memtest86+ on the sticks... The RMA'd set failed at around 12hours 45minutes, so I ran the test for couple of full days to be as certain as possible, that my new RAM was OK.).
Same goes for the rest of the system: I have as much stability with older AGESA as I have ever had! Prime95? Bah, bring it on!
Running Prime95 with 3D Mark simultaneously? No problems! And so on, and so forth. So the hardware *should* be on full working order.
Also worth noting is, that the system is 100% stable with UEFI F9d, which is actually a pretty old BETA - afaik it's from around September 2017 - couldn't find actual release date.
So, do any of you great people have any ideas at all? This issue is not unheard of - in fact, there is one on this very forum on the Graphics card side (link to follow) with similar issue, but on Radeon graphics. Also few people online have had similar issues with AB350 chipset - so it looks to be inherently somehow connected to this - at least to some civilian like I am.
Gigabyte (motherboard manufacturer) has been informed. I am yet to receive a reply...
Asus (graphics card manufacturer) has been informed. I am yet to receive a reply from them as well...
Links to other similar cases I have come across so far:
Many of these are of the same motherboard (GA-AB350-Gaming 3), but not all. So this is highly curious and I hope perhaps AMD could have *any* ideas - or anyone else here!
Thank you for your help in advance
I love my Ryzen, despite this issue!