View Full Version : Real time clock, inconsistent?

Son Goku
11-18-05, 07:52 PM
OK, this is the first time I've seen this. The clocks on the computers are not entirely accurate (and tbh, the timing crystal that the bus clocks/CPU clock is generated from isn't 100% accurate either, as the timings tend to fluctuate a bit). Anyhow, I have seen systems that either are a tad too slow, or a tad too fast. AKA, an old Pentium 100 I used to have lost about 1 minute every 3 months or so, and the Pentium II 400 I had gained a minute every 2 months or so. I've found this to be normal, and expected.

However, this beats the heck out of me. Especially because it's an odd fluke, and not consistent. My computer's clock seemed to be keeping fairly accurate time, and as winXP Pro updates itself periodically off Internet time, or one of the time services, all seemed well for about a month or more since building my Athlon 64 here.

Then about 2 days ago, I got up, and my computer's clock was 15 minutes ahead of where it should be. Mind you, the time was correct only 8 hours earlier. So I looked, and figured (eh, perhaps it just got set wrong), and at the time that time server was setting it a bit ahead. So I switched to the other listing (not the time.microsoft.com, but the second entry) and all listed OK. So yesterday comming home, it was ahead of where it should be, and I switched it back to time.microsoft.com and got the right time.

However, in several hours hours it was about 15-20 minutes ahead of where it should be. I checked the clock, and the seconds were zooming by about 2-4 every second, and the clock was falling out of sync fast. So I rebooted, checked it in BIOS, and just sas fast.

I made it synchronize again, but here's the thing. Today it's keeping the right time. WTF? And has someone seen this before? As I said, I've seen the real time clock on different computer systems be a bit off (by a minute every so many months for instance), but not by this amount, and not inconsistently. In those other 2 systems, they were always off by that amount, and didn't fluctuate as I noticed here. Huh?

11-18-05, 11:12 PM
the little battery on your mobo.. yah.. replace it..

11-18-05, 11:36 PM
It shouldn't matter if the mobo has power, but worth a shot.

The only time I've seen this is with CnQ enabled or with dual cores when apps try to read from the RTC on both cores back and forth.
The AMD driver and dual core MS hotfix fixes both of those though. In theory if you add /usepmtimer to your boot.ini line it should fix windows for you atleast. There was a thread (in the windows forum I think) where a couple people had this problem after installing an HP printer (strange but true).

What I would do is monitor voltages and your cpu speed and see what's happening.