I've had a similar issue with my IBM PC 300GL. It went foward to the year 2037 or something like that (after being unplugged from the wall for a few months).

It happened only one time for me, so I didn't get a chance to debug this issue.

You could check the voltage of the Lithium battery, it could cause issues if it's under ~2.8V

This seems like a similar issue that I had with my board, so in part it may be related to Y2K problem in BIOS.

Some motherboards store CMOS data in static RAM (like in case of my board), so BIOS settings will persist even if you disconnect the battery, or if the battery has no charge. The battery is required, however, for proper run-time clock (RTC) operation.

I haven't seen disassembled BIOS code yet myself so can't be 100% certain, but suppose if your motherboard stores first two digits of the year in static RAM (20xx) and the last two digits (xx20) in RTC clock, after the power is disconnected while CMOS battery is dead, your last two digits will reset to zero (00). Due to Y2K bug in BIOS, where not so good programmers thought that their software will not live longer than 5 years, so they added a year validity check during POST with allowed years between 1994 and 1999 (ignoring first two digits stored in SRAM). Therefore, after cold boot, your BIOS would "correct" an incorrect year of "00" back to "94", then add first two digits from static RAM, so the final date would become 2094.

So to resume, a solution would be replacing CMOS battery, updating BIOS and/or using some Y2K-fix utility like Y2Kure.

Well, I've triple checked the Lithium battery that is on the board and it certainly supplies 3.58V which is basically at full capacity. So I decided against replacing it for no reason. It still retains all CMOS data except the last two digits of date.

Could it be a strange Y2020 bug? Maybe it is bugged but not at 2000 but at 2020, so when the date gets to 2020 it bugs out and tries to go back to 1994, but the first two digits of the date are stored somewhere else in the CMOS? I should try it out with 2019 again and see if it can retain that year.

Can you please tell the markings that you have on RTC chip on your motherboard? What type of BIOS do you have with what version/date it has?

Also, for the cold boot reset, does the date gets reset even if you disconnect the power and re-connect immediately, or only after 12-24 hours or so?

Static RAM will not retain its content when power is gone. Flash memory does.

I was referring to nvSRAM, which seems to be used in some RTC clocks. There is a link to a thread for my own 486 board, which exhibits similar issue and it's BIOS settings and current date are preserved even without CMOS battery. The only difference that CMOS battery makes is for running RTC clock, otherwise it'll stay at the last date.

I have a SiS 486G motherboard with the same issue. It is certainly a BIOS Y2K bug of some type. I just set it to 1994 and pretend that it is 1994! Maybe "2094" is the next coming of the 486!

Hi eightbit,

Yes, this year 2094 bug is indeed a known BIOS Y2K bug from the 1994/1995 era.
For the Award BIOS there is an easy fix that you can do yourself. You can read all about it in my DIY BIOS patching story:
DIY Bios Modding guide Jan Steunebrink k6-2+/3+ 128gb

But there may be an updated BIOS for your board as well. Please tell us the BIOS-ID string as displayed at the bottom of the first BIOS boot screen. When you hit the Pause key during memory count, you can freeze this screen and have ample time to read it.
Or just post a picture of this screen. 😉

Jan

Noticed today my 486 was doing this and wondered if it was some kind of Y2K bug. Board is a DataExpert EXP4045 with the 01/18/95-SYM-SL82C491/49-2C4J6D01-00 BIOS which is the latest on ultimateretro.

For now my solution is network time - set the clock from an NTP server every boot.

Hi davidrg,

That's a neat solution, using an NTP server to sync the clock on this board. 😀

I've checked this 01/18/95-SYM-SL82C491/49-2C4J6D01-00 BIOS and it has indeed the year 2094 bug.
Let me know if you need help patching it.

Cheers, Jan

Chkcpu wrote on 2022-03-26, 18:10:

Hi davidrg, […]
Show full quote

Hi davidrg,

That's a neat solution, using an NTP server to sync the clock on this board. 😀

I've checked this 01/18/95-SYM-SL82C491/49-2C4J6D01-00 BIOS and it has indeed the year 2094 bug.
Let me know if you need help patching it.

Cheers, Jan

Would you help me with the Chicony BIOS from the OP as well? Cheers,

Hi appiah4,

Yes, I can help you with the Chicony 471A ver 3 BIOS.
Can you make a good dump of this CH-471A ver 3 BIOS?
I don't see it on UltimateRetro. 😉

Jan

Chkcpu wrote on 2022-03-26, 22:04:

Hi appiah4, […]
Show full quote

Hi appiah4,

Yes, I can help you with the Chicony 471A ver 3 BIOS.
Can you make a good dump of this CH-471A ver 3 BIOS?
I don't see it on UltimateRetro. 😉

Jan

Of course, I will do that next weekend when I next have access to the system it is in. I will be downgrading it from a U5S-33 to an i486-33 so I'll just dump with UNIFLASH.