overclocking kills onboard audio?

[bprewitt]

Hey folks,

I have a soyo 7iwm/l w/ a 650mhz Pentium III. I have it overclocked at 806mhz right now (124fsb/41pci bus) at a multiplier of 6.5.

The motherboard boots up, Windows loads good, but the onboard audio (that ****ty AC97 codec) doesn't load up when I have it overclocked. It just simply seems to be disabled. Knock it back down to 650mhz, the onboard loads just fine.

Anything above 124fsb/41pci bus and the system will completely refuse to boot. No post, no nothing. And I'd like to have this overclocked, seeing as how 156mhz is a nice gain. But I'd rather not lose my only audio source in the process.

Any ideas as to why this is happening? Any help is graciously appreciated.

Other specs:
soyo 7iwm/l v1.0 (never flashed this mb)
Windows 2000 Adv Server
AC97 audio codec
etc

Thanks in advance!

[BipolarBill]

Because the onboard audio is sensitive to OCing - that's why. Try disabling it in BIOS and putting in another audio card.

Welcome the the world of overclocking - where some PCI components fail before others because pumping up the FSB pumps up the PCI bus frequency. Not all of the components can take this.

Realistically, I wouldn't go over 112-118MHz on the FSB. You're asking for trouble if you do. Things will break. If you think you can't afford a faster CPU, can you afford a new motherboard? How about replacing fried parts?

I recommend spending $50 on a faster CPU. Overclocking is for people with more money than sense, IMO.

[SPEEDO]

Overclocking is for people with more money than sense, IMO.

Great line...

SPEEDO

[Rugor]

I don't know if O/Cing is always for people with more money than sense, but there are good Overclocks and bad ones. For example, we have in the house both a Celeron 300A, and an Athlon 1800+. The Athlon is running at default, it's more than powerful enough and why bother O/C it. The Celeron 300A is on a BX motherboard and has been running at 450MHz (PCI bus in spec) for the last two years and more. Overclocking a Celeron 300A is a good choice, other O/C's may not be so good.

However, if you run your PCI bus as high as 41MHz you are asking for trouble, that's a good 20% over spec. I limit myself to 37.5 if O/Cing the PCI bus.

[bprewitt]

Originally posted by Rugor
However, if you run your PCI bus as high as 41MHz you are asking for trouble, that's a good 20% over spec. I limit myself to 37.5 if O/Cing the PCI bus.

Well, I didn't like the idea of o/cing the PCI bus that high, but like I said above, the system won't even boot when I clock higher than fsb of 124mhz. I'm not quite sure why, but the system won't even load the bios or post, so I have to manually reset it.

I'd love to be able to o/c at a lower pci bus, but considering Soyo has some really ****ty fsb/pci bus options in the bios, it's a little hard when the system won't even boot. That and the voltage controls aren't on-board in the bios, so I'd have to play with the jumpers (I believe...).

Would playing with the voltage level allow me to select a higher fsb with a lower pci bus? I'd like to do 133/33, but like I said, that speed won't even boot.

[BipolarBill]

I never could push the FSB past 124 on any Soyo board.

The PCI bus frequency is tied directly to the FSB in older boards unfortunately. Worse than that is the limited AGP multplier choices. 2/3 of 128 is 85.6. That's too high. It should be selectable to 66Mhz.

All of this soured me on OCing a long time ago. I will occasionally try a 12% boost, but the first time anything weird happens, I shut it down. I need this thing.

[Rugor]

I know the last Soyo I had started working the PCI bus from a 1/4 divisor before 133 MHz, but I can't remember where the transition was.

You MAY be able to go in on the 100 FSB, set it to something like 127 or higher where the PCI bus is under spec. Some Soyo boards use jumpers to set the FSB range, so that may do it-- If you have the FSB jumpered to 133 base and then go to something in the 120's your PCI should be underclocked and thus more stable.

[bprewitt]

Thanks for all the help, folks. I really appreciate it.

I think I will give up on overclocking for now, as I, too, need this thing. And the idea of frying it for a few extra mhz's doesn't sound like a just cause for me (especially with the limited range I have to overclock). So, again, thanks!

[Rugor]

Good idea,

A really good O/C chip will normally be one that's made on the same process as the higher speed CPU's available but on a lower FSB. Then you get one that has a multiplier low enough that if you bump it to the higher FSB you still remain inside the normal range for that core.

As an example, I predict that if Intel keeps the Celeron on the 400MHz bus and P4 on the 533MHz bus, Northwood core (.13micron process) Celerons rated at 2.0GHz (20 x 100MHz base FSB) will quite often O/C to 2.66GHz just as the current 1.6A GHz Northwoods can often do 2.1GHz if the motherboard supports it.