Arch Linux

incinerator

Member

From: Edinburgh, Scotland

Registered: 2005-02-15

Posts: 80

Racing clock problem with nforce2 board [SOLVED]

Hi,

I was having a problem with the system clock on my Abit NF7-S v2.0, which is an Nforce2 motherboard. By accident, I found an easy fix. I'll first describe the problem and then how I fixed it. Hope this helps all those nforce2 users with racing clocks out there.

Problem: The system clock was (literally) racing. When the computer was switched on, the system clock was gaining several seconds per hour, regardless of what operating system was running. This was so bad it even running time synchronisation software like ntp only provided marginal help. I first suspected the "C1 CPU disconnect" feature to be the cuplrit but was proven wrong.

Solution: Reboot your computer. When the machine is resetting, call up the motherboards BIOS, this is usually done by pressing the "Delete" key once the machine starts initialising itself shortly after reset or powerup. Once in the BIOS go to the "Advanced Chipset Features" menu. Set the options "FSB Spread Spectrum" and "AGP Spread Spectrum" to DISABLED. Then "Exit & Save" which will reset the computer again.
That solved the problem for me. The clock is behaving normally, clock drift is within normal parameters now.
I suspect the first of both options may already be enough to get it right, but I can't be bothered to reboot my comp to find out *ggg*

I guess this behaviour may be reproducable with many motherboards having the nforce2 chipset, possibly other nforce chipsets may also be affected. On the other hand, it may just be a bug in that specific motherboard's BIOS.

Cheers,
Dominik

smith

Member

From: Crescent City, CA

Registered: 2005-02-19

Posts: 77

Re: Racing clock problem with nforce2 board [SOLVED]

Finally!  I've got an MSI Nforce2 board and have been seeking a solution for the racing clock problem for some time.  Thanks.

---

I have nothing to say, and I am saying it.

RedShift

Member

From: Belgium

Registered: 2004-07-16

Posts: 230

Re: Racing clock problem with nforce2 board [SOLVED]

For who's interested: spreading clock spectrum means each clocktick will have a slightly different modulation than the last one. This option was built in to prevent clockgenerators from interference.

---

:?

Convergence

Member

Registered: 2005-07-02

Posts: 377

Re: Racing clock problem with nforce2 board [SOLVED]

Was there a time when your clock worked fine with spread spectrum enabled, or did it cause this issue from the getgo? I'm asking because my computer which had no trouble for over a year, suddenly started racing recently.  It IS nforce2 based.

---

It's a very deadly weapon to know what you're doing
---  William Murderface

byte

Member

From: Düsseldorf (DE)

Registered: 2006-05-01

Posts: 2,046

Re: Racing clock problem with nforce2 board [SOLVED]

I was having timing issues (albeit with a 7 year old notebook) myself today. Either the clock went twice as fast or five times slower, depending on which workaround I was testing.

acpi_skip_timer_override [HW,ACPI]
                        Recognize and ignore IRQ0/pin2 Interrupt Override.
                        For broken nForce2 BIOS resulting in XT-PIC timer.

The kernel docs are an endless source of wisdom, methinks...

The solution for my problem was "clocksource=pit", btw.

---

1000

elasticdog

Member

From: Washington, USA

Registered: 2005-05-02

Posts: 995

Website

Re: Racing clock problem with nforce2 board [SOLVED]

It is a known issue with nforce2 boards, which I wrote about a while back .  It's a little strange though, Convergence, that your's just started to race, but it very well could be related.  When I was tackling the problem, I noticed it right after I built the system, so I can't say for sure if it could occur after being normal for so long...