Locking PCI bus @ 33mhz

[subhuman]

I read somewhere (or maybe I dreamed it? ) that some new motherboard (which?) will let you run the CPU overclocked by adjusting the FSB, but that the PCI bus would remain locked at 33mhz. I thought it was the i845-DDR but I can't find where I read this again. Anyone know? I was hoping it was the ASUS P4B266...

 

[subhuman]

Okay I think I found where I read it, in the i845-DDR Anandtech review:

http://www.anandtech.com/mb/showdoc.html?i=1567&p=4

Now, still trying to figure out if all i845-DDR boards do this, or just the Abit. Will keep looking, but anyone know?

 

[tapir]

This is a feature known as PCI dividers and is implemented by board makers specifically. Companies like Abit put them on most of their boards. They don't "lock" the PCI slots, persay, but they do prevent them from getting severely overclocked. You can look around for an explanation, but just check reviews of boards for "PCI dividers" and you'll find what you want.

 

[subhuman]

Of course, it's obvious that it's a PCI divisor - just curious if the i845-DDR chipset specifically had a large selection of divisors?

Basically - are divisors part of a chipset, or an implementation of a chipset...?

I know, for example, that there was never a 440BX board with an AGP divisor, so running 133mhz FSB results in 89mhz AGP, etc. Also curious if there is a list of available divisors for the Abit board, and when they kick in. Will go download the Abit i845-DDR manual now...

 

[subhuman]

Seems to be a feature of the Abit BD7, pretty cool. I hope other Intel motherboard makers follow suit, as P4 chips are multiplier locked this could really help with overclocking while keeping the rest of the system in spec!

 

[Peter]

It's twofold ... the clock synthesizer PLL chip must be capable of producing output clocks at those ratios, and then the
system chipset must be able to cope.

Not all chipsets do it the same way. Traditionally, SiS chipsets can run completely asynchronously, so you can use a
PLL that outputs a constant 33 MHz clock for PCI regardless of the CPU and RAM clocks. VIA and Intel chipsets
do nothing but integral multiples (2/1, 3/1, 4/1, maybe 5/1 too), while some of ALi's also support running on
fractioned multiples (5/2 and such) but not real async.

Then there's AGP to add to the game ... the AGP clock is coming out of the chipset, and this is where the chipset has
to get it right for every thinkable CPU/RAM/PCI clock combination. (E.g. this is why Intel BX can't do 133 MHz CPU bus
right, because it can't do AGP=CPU/2.)

regards, Peter

 

[subhuman]

Thank you Peter, you're just excellent!

 

[kursplat]

<< ABIT even offers the ability to fix the PCI operating frequency to 33MHz regardless of what the FSB frequency is. >>


sounds pretty much asyncronously to me...

 

[wouldbeen]

It seems like Asus A7N266-E is capable to do the same thing.
Here what Kyle Bennett says in his A7N266-E review:

When the 158MHz FSB did not prove to be enough for us, Asus shipped us one over that would do 172MHz. One thing that you really do need to pay attention to here is that no matter what we set the FSB to, the PCI bus stays at a spec 33MHz. Schweet.

 

[subhuman]

Anyone confirm or deny this feature is on the ASUS P4B266?

 

[Rebels7]

I just recently (two days ago), installed the Asus P4B266-C and I can say that I don't think that it supports this feature (may be an option with an upcoming BIOS upgrade). I also had to RMA a Asus P4S333 that did not have this feature. Maybe the P4B266 or the P4B266-E has this feature, I am sure that someone has one of these boards and can respond.

 

[subhuman]

I don't think P4B266 has it, couldnt find it anywhere in the BIOS and no mention of it in the manual... guess its the Abit BD7 if you need that then, hmph.

 

[Duvie]

read anandtech's recent i845 roundup...

It pretty much laughs at abits 33mhz pci lock cause it feels you can't even get the board to fsb oc enough to need to use it...quite laughable????


Look at using the pci divider....

Many boards can get well over the 133fsb stable, such as

Msi ultra ARU

GA-8IRXP

Asus P4B266

I haven't even heard of the abit mentioned in samebreath as these other when it comes to good ocing i845d mobos...

The abit seems to be feature rich and that is definitely a great assess, but I wouldn't recommend it based on my research and talking with fellow anandtechers in the private messages...