This talk of PCI Divisor of 5 has me thinking......KT7AR BIOS Tweak

AMDGuy · Jul 11, 2001

If the PCI divisor is controlled by the BIOS, why couldn't we use a program like WPCREDIT to adjust the default PCI divisor? If we can find what register in the bios controls this setting it may be possible to change it to a 5.

Anyone out there good at this type of thing? Unless the chips on the board aren't capable of this, I don't see why they wouldn't be, then it should be possible.

I'd like to find out about this on an ABIT KT7AR and see how it would work.

Any thoughts?

Nagorak · Jul 11, 2001

I don't think VIA boards even have firmware support for a /5 PCI divider. But something like this may be possible on AMD boards.

Bender · Jul 11, 2001

What chip controls the Divisor? Its the north bridge correct? You could use wpcredit and wpcrset. You don't even need the pcr files if the chip isn't supported. I would do an environment save with wpcredit with the divisor at 3 and at 4, then compare register settings. All you would have to do on your board would be to change the register to the 5 divisor register setting. I think my logic is sound as long as the device that controls the divisor can be edited by wpcrset.

AMDGuy · Jul 11, 2001

Bender (Jul 11, 2001 08:25 p.m.):
What chip controls the Divisor? Its the north bridge correct? You could use wpcredit and wpcrset. You don't even need the pcr files if the chip isn't supported. I would do an environment save with wpcredit with the divisor at 4 and at 5, then compare register settings. All you would have to do on your board would be to change the register to the 5 divisor register setting. I think my logic is sound as long as the device that controls the divisor can be edited by wpcrset.

Actually you'd have to do an environment save with a divisor of 3 and a divisor of 4. This would show what register is changing. Since a divisor of 5 is desired and not an option in the BIOS, you can't set it to 5 for comparison. I assume this is a typo and you meant 3 and 4, this is a good theory.

Ferg · Jul 11, 2001

Well to do that then you would have to get your fsb bus to 166 to make the pci slots 33 Mhz. I dont think that a VIA KT133A chipset can get even close to that. So unless you found a way to add more voltage to your northbride, add massive cooling to your northbride; ie: water cooling, and get some very high performance ram then your not going to be able to do this without hurting your system performance.

Me · Jul 11, 2001

The iwillkk266 can achieve 166mhz FSB with decent ram

Ferg · Jul 11, 2001

It would take PC-2400 probably. The Abit KT7A however probably cannot.

SickBoy · Jul 12, 2001

Ferg (Jul 11, 2001 09:36 p.m.):
Well to do that then you would have to get your fsb bus to 166 to make the pci slots 33 Mhz. I dont think that a VIA KT133A chipset can get even close to that. So unless you found a way to add more voltage to your northbride, add massive cooling to your northbride; ie: water cooling, and get some very high performance ram then your not going to be able to do this without hurting your system performance.

I've talked to/heard from several guys running KK266's at darn near or above 166 MHz FSB.It's definitely possible... although I think trying to implement it on an Abit KT7A would be kinda pointless, as it is not one of the boards that can acieve ultra-high FSB.

SickBoy

asmodean · Jul 12, 2001

Maybe it's the lack of divisor 5 that KT7A can't take that high FSBs?

Bender · Jul 12, 2001

So are we in agreement that it is the Northbridge that controls the divisor? If so it wont be incredibly difficult to do this little hack.

AMDGuy · Jul 12, 2001

Bender (Jul 12, 2001 10:03 a.m.):
So are we in agreement that it is the Northbridge that controls the divisor? If so it wont be incredibly difficult to do this little hack.

As far as I know the Northbridge does control the divisor.

Bender · Jul 12, 2001

I am pretty sure the kt133a chipset will do a divisor of 5 so I will play around with it when my A7V133 arrives later today. I'll let you know if I discover anything interesting.

Len · Jul 12, 2001

A 1/5 divisor seems to be possible with AMD 760/VIA 686B since a new BIOS for the Epox 8K7A just was released that supports 1/5. People over on the amdmb.com forums are going nuts trying out 180+ FSB speeds with plain old Crucial PC2100 overvolted +0.4v to 2.9v (another great feature of the Epox 8K7A).

I personally question the benefit of these very high FSB speeds since you typically must use slower memory timings to achieve them. This would seem to negate much of the performance gain from using the higher FSB in the first place. However, I think more user testing & benchmarks are necessary to see what the real performance advantage of these 180+ Mhz FSB speeds are.

wild_andy_c · Jul 12, 2001

I have Nanya PC2700 184Mhz CAS 2. It's like the Mushkin Rev 3 of the DDR world.

This is possible using the 1/5 divisor which is automatically invoked in BIOIS after the PCI bus exceeds 41.5 Mhz at 1/4 divider. At 208Mhz, the 1/6 divider kicks in, only problem is that I can't get that high at this moment - I've only managed 204 at CAS 2.5 but can't maintain modem stability there!!!!

Eat chalk dudez!! wild_andy_c in da area, getting da fast stuff da quick style

asmodean · Jul 12, 2001

Wild_Andy_C, what mobo are you using?

el · Jul 12, 2001

the 8k7a! At least I think that is what we are talking about. andy mem scores at 204ddr are they around p4 levels? where did you get your ram and how much $.

asmodean · Jul 12, 2001

Just read a thread started by andy...

Okay, has anyone tried taking a snapshot of the Northbridge PCI registers of KT133A with a bus of 100 and 133? I know I will do that now.

The settings were exactly the same, except the other had a bus of 133+1MHz and the other had 100Mhz.
Here's the output of WinHex' compare:

Search for differences

1. D:\Documents and Settings\Administrator\Desktop\100 bus.rg: 256 bytes
2. D:\Documents and Settings\Administrator\Desktop\134 bus.rg: 256 bytes

50: 17 18
51: A4 A3
52: EB EC
64: 56 52
65: D2 D6
68: 44 45 - System Frequency Divider?
6A: 65 43
B7: 0D 05 - Reserved settings... Unknown then.

8 difference(s) found.

Will do research now.

Done, added suspicious registers. Will make System Frequency Divider "reserved" now

I just hope it doesn't die.

Done, it wouldn't stay at 47, always switched back to 44... Any other ideas, guys?

Bender · Jul 12, 2001

Here is the northbridge of my A7V133 at 100 and 155 fsb. I haven't drawn a conclusion yet I am still trying to figure it out. It looks like a few things I wouldn't have expected to change did change.
[img="[URL]http://home.talkcity.com/Arcade/spock84/100-0.jpg[/URL]"]
[img="[URL]http://home.talkcity.com/Arcade/spock84/155-0.jpg[/URL]"]

AMDGuy · Jul 12, 2001

asmodean (Jul 12, 2001 04:45 p.m.):
Just read a thread started by andy...

Okay, has anyone tried taking a snapshot of the Northbridge PCI registers of KT133A with a bus of 100 and 133? I know I will do that now.

The settings were exactly the same, except the other had a bus of 133+1MHz and the other had 100Mhz.
Here's the output of WinHex' compare:

Search for differences

1. D:\Documents and Settings\Administrator\Desktop\100 bus.rg: 256 bytes
2. D:\Documents and Settings\Administrator\Desktop\134 bus.rg: 256 bytes

50: 17 18
51: A4 A3
52: EB EC
64: 56 52
65: D2 D6
68: 44 45 - System Frequency Divider?
6A: 65 43
B7: 0D 05 - Reserved settings... Unknown then.

8 difference(s) found.

Will do research now.

Done, added suspicious registers. Will make System Frequency Divider "reserved" now I just hope it doesn't die.

Done, it wouldn't stay at 47, always switched back to 44... Any other ideas, guys?

I guess the question is how to determine what a 3, 4, or 5 is equal to in Hex and how it reacts with the rest of the settings. I'd think these settings would be somewhere in the 60 - 64 register range as those are the ones that deal with memory and timings.

wild_andy_c · Jul 13, 2001

RAM - trade sample direct outta Taiwan, cost $0 but it ain't quick enough. It's this exciting thing where you just want for more and more. Now the whole divisor issue is gone, we can really go ballsout.

Wait for NForce - that can do memory at a different speed to the Northbridge !!!!

This talk of PCI Divisor of 5 has me thinking......KT7AR BIOS Tweak

Member

Senior Member

Mysteriously Changing Senior

Member

Member

Me

Guest

Member

Inactive Moderator

Senior Member

Mysteriously Changing Senior

Member

Mysteriously Changing Senior

Member

NAKED Overclockers Unite! Senior

Senior Member

el

Guest

Senior Member

Mysteriously Changing Senior

Member

NAKED Overclockers Unite! Senior

Similar threads