Slightly OT: CR2032

[Mike Clarke]

A flat "CMOS" battery is hardly likely to cause total failure of the PC. You'll just have the inconvenience of going into the BIOS at boot time and resetting the clock. In the worst case you might need to alter some settings from the default. After that just leave the PC plugged into the mains for the 2 or 3 days it takes for the new CR2032 to arrive.

[Roger Hayter]

That's it! Though I seem to have imagined the speed advantage, it seems to have been purely a backup arrangement.

[tabbypurr]

It shouldn't be any great problem. But worst case is needing to tweak ram timings to get it to work at all - unlikely though.

NT

[Mike Tomlinson]

En el artículo , Roger Hayter escribió:

A bit of a marketing gimmick, IMO. The days of a bad BIOS flash bricking the board are long gone.

Some premium-quality boards even included a spare pre-flashed BIOS chip in the box.

[The Natural Philosopher]

Not seen it for years. Largely for several reasons I suspect

1/. Bios is no longer in slow RROM but in reasonably fast flash RAM

2/. No one uses the BIOS any more except to boot the machine up. Its speed is simply irrelevant. Once memory got large enough the bios was simply an embarrassment. Once protected mode 386s arrived no one wanted to make BIOS calls anyway. ghastly non re-entrant slow code. I dont think anyone has used te BIOS post boot since windows 3 have they?

3/. Such config info as was set by the BIOS is now set by software as well. Or is not needed by the software. I.e. programming of subsections of the hardware that are invisible to the OS on the 'once at boot time' basis doesn't need to be fast.

In short apart from a very very few things the BIOS is simply not used by modern machines at all, beyond the boot process itself.

Once the OS and drivers load, they take over all the hardware. And configure it according to what is in the config files.

I cant remember when I last fiddled with a BIOS beyond setting the boot devices and order. And using it to set the real time clock.

I know that gamer MBS have lots of overclock options.

I've never used em.

"The BIOS in modern PCs initializes and test the system hardware components, and loads a boot loader or an operating system from a mass memory device. In the era of MS DOS, the BIOS provided a hardware abstraction layer for the keyboard, display, and other input/output (I/O) devices that standardized an interface to application programs and the operating system. More recent operating systems do not use the BIOS after loading, instead accessing the hardware components directly."

(wiki/BIOS)

I.e. there's no point in making the BIOS run faster in RAM as no one is using it anymore.

I wonder if it even exists - the old int 21 stuff..

[T i m]

Whilst it may not be very likely (and not a 'total (permanent) failure' as such), I've experienced it myself at least a couple of times now. Switch PC on and it appears to be dead. No POST beep, no video display, nowt. Remove reset / RAM, turn back on, nowt. Replace CMOS battery, turn on, boots straight away. The last instance of the above was a Netbook that had a built-in (main) battery FWIW.

That is the most typical outcome, yes. I've recently dug out some old laptops and PC's and had to do that on most of them. ;-(

In the old says it might have been the hard drive geometry and now days the AHCI / Legacy IDE settings, the boot order or Secure Boot etc.

I have also given the call a quick flash on my bench PSU and that often seems to lift the voltage / charge long enough till the replacement can be obtained (don't try that at home kids). ;-)

Cheers, T i m

[The Natural Philosopher]

How little you know...

[Johnny B Good]

Ok then. Whatever it was you were thinking of, it had nothing to do with "The CMOS RAM". It looks like Mike has figured it out. Dual BIOS would appear to be what you're now describing.

ISTR it being a feature of Gigabyte MoBos but since I never used any Gigabyte boards for any of my builds, my recollection is limited to the marketing hype alone.

Gigabyte MoBos just seemed to be yet another run of the mill product - nothing to write home about yet still infinitely better than Foxconn and PC Chips boards (but then everything else was way better than those two brands anyway).

Even though the use of flash ram to replace the roms and eproms used to store the bios code could have also been used to store the "CMOS settings" without need for a battery, it never was other than for user and admin passwords which had initially been stored in the RTC chip's registers which could all be reset by using the cmos clear jumper.

I think the use of flash ram to store passwords was implemented in laptops first before the idea caught on with desktop MoBo makers. Even today, bios settings still use the battery backed RTC registers since the need for a 'battery' to maintain the system date and time during loss of AC power events (power outages, relocating the PC elsewhere or hardware updates and so on) has never gone away meaning there was never any advantage to be gained in moving away from a well and truly tried and tested system to using flash ram to store the settings (If it ain't broke, don't fix it!).

[Johnny B Good]

There were *some* MoBos (thankfully, a very few models, typically PC Chips afair) that didn't take kindly to a totally discharged CR2032 coin cell. AFAICR, the quick 'n' dirty solution was simply to remove the dead coin cell and run without.

Unless it was a pre-ATX machine, you'd only have to load the defaults or configure the cmos the one time when rebooting it afterwards (unless you were habitually shutting the AC power off after the OS had switched the machine off at the end of each session). If, as was the usual case, the machine was left connected to the AC supply in between sessions, the 5VSB rail would maintain the RTC and the cmos settings stored in its 70 odd spare registers so you could run without a coin cell installed with little to no hassle when starting it back up each time afterward.

Although a pound shop purchased CR2032 might prove unsuited to car key fob use, it would certainly be entirely suited for use as a replacement RTC battery in any desktop PC so you needn't have to deprive yourself of a replacement coin cell for very long unless you're located on a remote Scottish island and your local general store just happens to have run out of stock.

[Scott]

So I can change the battery preventively with no risk to the BIOS? I was thinking about clearing any dust from the fan and the vents so, as I have two spare cells sitting here, I might as well change the battery at the same time. Where can I find the battery in the computer?

[Johnny B Good]

I very much doubt the MoBo manufacturers have moved from a proven and reliable system of storing configuration data in favour of using NVRAM to store anything other than passwords[1] so I wouldn't bank on that being the case.

The RTC chip (usually integrated into one of the VLSI MoBo chipset chips) still needs battery power to maintain date and time during disconnects from the mains supply and its 70 odd spare registers are still available so there's little to be gained from using NVRAM, especially as corruption (or misconfiguration) of this data isn't so easily correctable with a simple CLR CMOS jumper which could potentially leave the board locked into a unbootable state.

[1] I may be wrong[2] but if NVRAM is used instead of the RTC registers, extra complication in the form of a more sophisticated POST detection and handling of any problems with corrupted or misconfigured cmos settings will be needed to reduce the risk of the MoBo getting into such an unbootable state. [2] I suppose it's quite possible that the cmos config settings data may have gotten bloated beyond the RTC's ability to store them all by now. That's the only reason I can think of that would cause the MoBo manufacturers to switch to a higher risk system requiring more complex coding to guard against show stopping lock ups.

After googling to see whether the settings on modern PCs are still being stored in the RTC's spare registers, the best I could find was that a small dedicated NVRAM chip *might* be used as an alternative rather than put the main NVRAM chip used to store the BIOS or UEFI code at risk.

Giving the BIOS code exclusive use of the Flash ram makes protecting it against malware attack a lot simpler (utilising a jumper link to disconnect the erase voltage pin or the write enable pin to make such software based attacks totally impossible for example).

IOW, if NVRAM is being used to store cmos settings, it's unlikely to be spare memory in the BIOS flash ram chip itself; more likely to be a small

8 pin I2 NVRAM chip (possibly socketed). The fact that Flash ram chips have been used for nearly two decades in place of eprom chips to store the BIOS code is no reason for such a change in the way the cmos settings might now be being stored without reliance on the RTC battery. Indeed, the use of such an extra chip may be simply to handle the 'bloat', allowing the show stopping critical settings to still be kept in the RTC registers whilst the extra NVRAM chip is used simply to store additional but uncritical settings.

The chance of needing to reprogram the cmos settings after a leisurely cmos battery change is likely to still exist even when an additional NVRAM chip has been used to store the "CMOS Bloat" on a modern MoBo. In any case, the settings in use are likely to be the default settings to begin with for most users who might only discover a long dead cmos battery after unplugging the machine from the wall by seeing the "Press F1 to load default settings and continue booting" message when they next power the machine up, leaving them with just the date and time to reset from the booted OS (windows typically).

If, as is typically the case, the machine is never normally unplugged between use, a dead CR2032 is unlikely to cause much, if any, inconvenience to the end user. The RTC and its registers will be maintained by the 5VSB rail all the time the PC is left plugged into the mains supply whether it's running or not.

[PeterC]

Last time, on a PC built in '07, I copied the BIOS first as I wanted to clean out the CPU's cooler and renew the TIM under it. The backup helped but some things need doing manually.

[Dave Liquorice]

Good point, it has spent most of that time not isolated from the mains. It must be watching this thread though, it's just thrown a couple of CMOS CRC check errors on boot up. Must find a battery and change it...