Gigbyte BIOS cycling on Boot

ctrob

Distinguished
Mar 11, 2003
35
0
18,530
My Gigabyte GA-Z68XP-UD3 motherboard has recently developed a problem booting after 5 years. This is after 5 years of use and I would like to make it clear that once the system boots (eventually) then it runs perfectly. It may restart ok or not but generally gets stuck again once it's been off power for a while.

I was still on an early BIOS version when this started I've since re-flashed to the latest BIOS and then to the UEFI BIOS but this has had no effect.

I have disabled legacy USB devices and even removed all peripherals except the keyboard/mouse to no effect and I've tested the battery, which is also still on full charge. I have reset the settings to 'optimised defaults' (there is no other default setting).

The cycling behaviour tends to be a quick restart followed by a longer one. Since I've had the UEFI BIOS, on the longer restart, it will get to a screen with Gigabyte UEFI across the top, pause and then reboot - sometimes I can see BIOS options across the bottom. The difference on a good boot is that it will boot straight to an "American Megatrends" splash screen and then immediately into windows.

I can break the cycle, at least for now but not 100% reliably, by resetting or powering off the system during a boot. With luck, it will then do a short boot cycle and then a good boot.

I am really puzzled by this. There don't appear to be any faulty components, at least the system is fine once it's up, but the system is 5 years old. Makes me wonder if there is a problem with the power regulation on start up or perhaps some component that is only access by the BIOS that's getting flaky.

Any ideas?

As a part 2: I looked into replacing the motherboard and had a bit of a shock. There are virtually no LGA1155 motherboards still on sale and most of those are rather inferior. It looks as though I will have to replace the entire system (CPU/RAM) if I want replace the motherboard.

This is annoying as my system is otherwise ok. I bought 16GB DDR3 only about 6 months ago and I will have to pay more to replace the Intel Core i7 2600K with an equivalent processor now than I did in the first place. Where's Moore's law been in the last 5 years?
 
Solution
Might actually not be a bad idea to REPLACE the CMOS battery as well. Five years is a rather long time for a CMOS battery, especially since there's no telling how old that battery was when it was picked from the bin for use on your motherboard in the first place. I've seen brand new motherboards come with dead CMOS batteries that had apparently already been sitting in the warehouse for years before they were installed on the boards, so yours could even be older than five years.

It might not be totally dead yet, might just have a very low charge/partial life remaining and hence the system works ok when it's been on for a while most of the time, but doesn't want to POST from a cold start.
Try booting with only one or the other of your memory modules installed. Sounds like it could be an issue with a bad module. Could also be a graphics card that's slowly failing. You might try resetting the bios entirely by unplugging it, removing the cmos battery for about five minutes and then reinstalling it and powering back up. That should completely reset the UEFI hardware tables.

Honestly, considering the age of the platform I wouldn't be terribly surprised if it's a faulty bios rom or another issue on the MB, but I wouldn't assume that it is without troubleshooting other possibilities first. A failing capacitor or power IC on the board could easily explain an intermittent failure to POST though.

 

ctrob

Distinguished
Mar 11, 2003
35
0
18,530
Thanks!

I can try the removing the graphics card and switching the memory, but both of those components are rather more recent than the mobo and there are no stability issues post the boot.

I can rule out issues with CMOS settings though - I have had the battery for far longer to measure its voltage plus the problem arose before I re-flashed the BIOS to UEFI.

Back before I switched to UEFI BIOS I did get some messages from the legacy BIOS about the CMOS or FLASH being corrupt and it would re-flash the BIOS (double FLASH image) but I think this was just due to me interrupting the boot process in an attempt to get it to boot.

I agree though I think it's more likely to be something on the mobo. I am mostly putting this out in case I have missed something: The Gigabyte guide has a remarkable number of things that cause boot cycling in the BIOS that are fixable.
 
Might actually not be a bad idea to REPLACE the CMOS battery as well. Five years is a rather long time for a CMOS battery, especially since there's no telling how old that battery was when it was picked from the bin for use on your motherboard in the first place. I've seen brand new motherboards come with dead CMOS batteries that had apparently already been sitting in the warehouse for years before they were installed on the boards, so yours could even be older than five years.

It might not be totally dead yet, might just have a very low charge/partial life remaining and hence the system works ok when it's been on for a while most of the time, but doesn't want to POST from a cold start.
 
Solution

ctrob

Distinguished
Mar 11, 2003
35
0
18,530


I just got an email saying this has been selected as the best answer but in the mean time things have progressed a little, though I still don't have a definitive answer.

I am reasonably confident that it is not the CMOS battery.

Currently there are still three potential culprits: the motherboard, the graphics card and the power supply.

Up until the end of October the PC continued to behave in its unsatisfactory but work-around-able fashion: the system would boot up in two different ways: either it would display the AMI logo and then boot through to windows and then work perfectly or it would get stuck in a boot loop with the Gigabye logo showing (which is what I get if I choose to enter the BIOS). This loop could be broken by pumping the reset button at the appropriate point in its reboot cycle but would otherwise persist.

Towards the end of October, the video crashed while running a (3d) game and since this time I have been unable to run the card with athe AMD drivers (HD 7850) for more than a few seconds without losing the display. It does work fine, as far as it goes, with the stock MS VGA driver. I managed to get the on-board graphics going but it didn't seem to be able to boot with this without the 7850 running as well. In this combination I managed to catch an error message along the lines of "stuck in driver thread" from the on board display after the 7850 driver had crashed.

With both video devices active it behaved much as before, with the same boot behaviour, although it was somewhat random as it which display was active and sometimes I would have to disable/enable the on-board adaptor before windows could use it. I tried various AMD drivers for the 7850 but nothing made any difference.

Eventually I got fed up with this and replaced the 7850 with an nVidia GTX 1060, disabling the on-board graphics again.

Since this change the PC is now working flawlessly again. So the most obvious answer is that this has been a slowly failing graphics card - and I think there is reason to believe this but I don't have another machine to test it in and it's been recycled now anyway.

The nVidia cards actually draw far less power than the AMD cards. So while it could be the graphics card, it could be just that the new card is putting less stress on the motherboard or PSU power management.

Whatever, it is working well enough now that I can hold off upgrading the rest of the system for the time being.