KT600 the 1/6 divider wont kick in at 200mhz -why?

[papasmurf]

I have a ga-7vt600L based on kt600 and I'm trying my darndest to make the best out of it. I'm running a barton 2500 in it trying to do a fsb overclock as high as possible, for some reason my 1/6 divider will not kick in at 200mhz like it should be. As this is a kt600 I can not lock the pci and agp bus, you see how irritating this is. I am currently running my cpu at 194mhz which puts my agp bus at 77mhz and my pci at 38 which is stable. The cpu's multiplier is 11 so 11x194= 2133 which is completely stable, I can boot into windows at 199 but that raises the pci bus to 39mhz where there is lots of instability, at a fsb of 200mhz the pci bus raises to 40mhz and the agp to 80 which is just too much for my componants. It SHOULD assign the 1/6 divider at that freq and bring the bus speeds down to 33x66 yet it does not. I am wondering if perhaps there is a wire trick I can do to make the motherboard believe that the cpu is default at 200mhz fsb if indeed that is the problem.

Join the TomsHardware IRC channel <A HREF="http://skulls.sytes.net/tom/ " target="_new">http://skulls.sytes.net/tom/ </A>

 

[Codinerx]

Hey man I had the same problem with the Epox8kra2+ mobo.. unfortunately there is no way around it, unless you can somehow mod the cpu to default at 200 fsb. Supposedly it is a limitation of the kt600 chipset, so there will not be any fixes coming.

 

[phsstpok]

I don't think wire tricks will change anything but try ocinside.de's XP/Barton pin-mod guide (see link). If you start with a 133 Mhz CPU you can mod to 166 mhz or 200 mhz. If you start with a 166 mhz CPU you can only mod to 200 mhz unless you cut the L12 bridges.

<A HREF="http://www.ocinside.de/go_e.html?/html/workshop/pinmod/amd_pinmod.html" target="_new">http://www.ocinside.de/go_e.html?/html/workshop/pinmod/amd_pinmod.html</A>

Wait for the overlay and if you change views click on the picture. (Sometimes the view doesn't auto-update).

I don't think this will fix your divider issue (but I guess it's worth a try). Divider selection is usually a function of the motherboard, the oscillator chip, and the BIOS. If the divider isn't working using BIOS selections then it probably won't work with wire mods.

What makes you sure the 1/6 divider isn't working? How are you measuring PCI and AGP speeds?

<b>56K, slow and steady does not win the race on internet!</b>

 

[papasmurf]

In my bios the agp and pci speed is reported on the same page that allows me to increase my fsb. I also have a handy program from gigabyte that tells me the same story. I believe this will fix it because the kt600 CAN run cpus with a 200mhz fsb and that means there must be some way it distinguishes between them and sets the 1/6. On the motherboard there is a single dipswitch for setting the fsb speeds you can use in the bios, the options are 100mhz(auto) and 166mhz in the manual it says to use the 100mhz(auto) switch if you have a 200mhz cpu. If you try to use the 100mhz setting without the 200mhz chip I believe the highest fsb you can set in the bios is 166, which leads me to believe that it can determine what speed the cpu is default at and give me the 1/6 divider with a 200mhz fsb if it believes that's what the cpu is supposed to be running at. Thanks for the link I'll be doing this today

Join the TomsHardware IRC channel <A HREF="http://skulls.sytes.net/tom/ " target="_new">http://skulls.sytes.net/tom/ </A>

 

[phsstpok]

I see.

Well, good luck!

<b>56K, slow and steady does not win the race on internet!</b>

 

[papasmurf]

I tried raising the fsb to 200 and for some reason it would not post, I then tried to unlock the multipliers but with this board I can not change them apparently. I am going to be buying a new motherboard soon hopefully!

Join the TomsHardware IRC channel <A HREF="http://skulls.sytes.net/tom/ " target="_new">http://skulls.sytes.net/tom/ </A>

 

[phsstpok]

I tried raising the fsb to 200 and for some reason it would not post

I'm not surprised. Simply raising the FSB to 200 mhz without changing the default multiplier and/or Vcore might cause POST problems.

If you can't change your multipliers then try increasing VCore. Use a pin mod for this if you have to.

Check your mods. Take note of the orientation of the chip with the "notched" corners, the two corners that are missing pins.

I then tried to unlock the multipliers but with this board I can not change them apparently.

Pin mods/wire mods aren't a function of the motherboard. They work on ALL Socket A motherboards (at least in theory). In practice, there are issues on some boards, old and new, but there are generally solutions as well.

All you are doing with pin mods is you are strapping the pins to either Vcore or to ground. Since the pinouts are the same on all socket A motherboards then in theory the mods work all motherboards.

Obviously if a chipset does not support a high FSB then you have limitations with high FSB speeds.

The voltage mods should always work.

If you can't change multipliers with pinmods then your processor is LOCKED. (Note that AMD began locking processors on Week 39 of 2003). The only mod that will still work on a locked processor is changing the L3 bridges on the CPU.

Changing multipliers via pin and wire mods can be tricky because the mods combine with the CPU's own defaults. For example, a Barton XP2500+ has a default multiplier of 11x. The L3 bridges are all closed, meaning all the multiplier bits are pulled LOW. Don't confuse the LOW setting with logic 0(zero). For this processor you need only connect the appropriate multiplier pins to Vcore, which resets the respective bits to HIGH.

Many people confuse manipulating the 8X multiplier bit with unlocking. Actually this only enables the high range or low range of multipliers (depending on which way you strap the pin) and it only works if the processor in already unlocked.

With Thunderbirds and Palominos we unlocked the processors by connecting the L1 bridges. Tbreds and early Bartons had these connected by default. (The connections were under the surface). The new processors are once again locked but this time the old L1 trick no longer exists. The processors are PERMANENTLY LOCKED.

We've been hearing about locked XP2500+ for months now but recently I started hearing about locked Tbreds, starting week 0343. (I didn't even know Tbreds were still being produced!) I've also heard claims of locked Applebreds and other Barton models.

This is bad news for someone hanging on to old motherboards (like me) who relies on the multiplier remappings in order to use new processors in old motherboards.

I believe we can still mod the L3 bridges (though I'm not positive). However, the wire and pin mods for multipliers don't work with LOCKED processors.

<b>56K, slow and steady does not win the race on internet!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 12/11/03 02:20 PM.</EM></FONT></P>

 

[papasmurf]

ah I was under the impression that wire mods worked with locked procs, unfortunately I have a week 45 barton I'm just going to buy a new mobo and fsb overclock it all the way. This board has no option in the bios to adjust the multiplier so even if I could unlock it it would do nothing I would have to choose specific multipliers with the wire tricks. My fsb mod was flawless and I know that the cpu posted earlier at 215mhz at the voltage I was using so I very much doubt that's why it did not work. I blame the motherboard, it just hates me lots I think I had my proc running at 2.1ghz but then after a few days stability appeared to be less than I had thought, my files started becoming corrupt (probably a memory issue) and the high frequency that my agp bus was at was causing horrible ailiasing with my overclocked r9700. I have just lowered it back to default to prevent damage to my componants and keep my windows install working till I can replace the motherboard. It's a pain in the ass to reinstall windows on a sata hard drive via controller, some idiot decided to make it only possible to install the controllers drivers via floppy <- the most worthless device ever. My floppy drive can never read my disks and my floppy disks keep dying. Anyways if my windows install gets corrupt then I have to go through hell trying to fix it.

Join the TomsHardware IRC channel <A HREF="http://skulls.sytes.net/tom/ " target="_new">http://skulls.sytes.net/tom/ </A>

 

[pIII_Man]

does your motherboard have a BSEL jumper that you can use to select the bus speed. For example some boards have a jumper that selects 100-133-166-200 i beleive that will give you the 1/6 and 1/3 dividers.


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[phsstpok]

Could be you need better northbridge cooling.

Regarding locked vs unlocked processors, yeah an unlocked one is need for wire and pin mods. The old circuit which the L1 bridges are part of is the circuit that feeds multiplier information back from the motherboard to the CPU. Whether you are resetting multipliers by built-in jumper, BIOS multiplier selection, wire mods, pin mods, or motherboard mod that circuit has to be available.

AFAIK this is not true of the L3 bridge mods. Changing these changes the true default multiplier of the CPU. Of course, modifying and re-modifying CPU bridges would be very difficult.

I've heard that these new "locked" processors have no contact points for the L1 bridges, thus can't be unlocked.

Would you tell me if the L3 bridge contact points still exist? Thanks.

Sorry about your troubles with that motherboard. I swore I wouldn't get another motherboard with a VIA chipset. You haven't changed my opinion.



<b>56K, slow and steady does not win the race on internet!</b>

 

[papasmurf]

my northbridge is well cooled with an old geforce2 gpu cooler, the l3 contacts are on my chip and it is not the chipset that's causing the problem the motherboard is not giving me my freaking divider.

Join the TomsHardware IRC channel <A HREF="http://skulls.sytes.net/tom/ " target="_new">http://skulls.sytes.net/tom/ </A>

 

[phsstpok]

Sorry I wasn't much help. Perhaps a future BIOS update will give you the 1/6 divider. I'm pretty sure it's the BIOS that tells the oscillator chip to set the dividers.

I'm really puzzled why the FSB mod doesn't work for you.

Thanks for the info regarding the L3 bridges.

<b>56K, slow and steady does not win the race on internet!</b>

 

[pIII_Man]

sometimes it is the chipest...in this case i could see how problems may arise.


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[phsstpok]

I wonder if it possible to force dividers with mods?

Is there something that can be tweaked on an oscillator chip?

I assumed the BIOS uses an algorithm, ie If FSB is 133 Mhz you get a 1/4 divider, if 166 Mhz you get a 1/5 divider, if 200 Mhz a 1/6 divider. Maybe I'm wrong, but something programs the oscillator chip based on bus speed.

<b>56K, slow and steady does not win the race on internet!</b>

 

[phsstpok]

One last thing. You might look into a program called wpcredit (or something like that). It's a very useful utility for looking at and maniniputing low-level chipset options. It might give you an idea of what is going on.

I don't know if there is a plug-in for the Kt600 chipset, though.


<b>56K, slow and steady does not win the race on internet!</b>

 

[pIII_Man]

Like i said it depends...i have only looked into the agp/pci clocking of the 440bx and the i815 chipsets. I can tell you that the 440bx's agp clock is regulated by the chipset (divided from fsb input) and the pci bus was regulated by a PLL. The 815 was all done by the PLL. My guess in the case of the KT600 the reason most if not all (because i have heard this problem before on many mobos) mobos cannot manipulate the agp/pci divider probably because the chipset manually senses the BSEL pins (or whatever the bus speed selection pins are called on amd cpus) and sets the divider from there (therefore the bios cannot asign the dividers as the kt600 impliments them on its own). With the 440bx chipset (i know its old but good for an example) the bios was able to modify registers so on its own it could change the agp divider to either 1x or 2/3x. One easy way to find out if the PLL is controling the agp and/or pci buses is to find the PLL's model number, go to the manufacturers website and see what outputs it has. My guess is the kt600 sets the agp bus on its own and the PLL sets the pci bus. Now if this is the case you can modify the pLL (through logic pins that are either set high or low) to give you the bus speed and dividers that you want. If the AGP or pci (i doubt the pci bus is controled by the chipset) bus are controled by the chipset it is extremely hard to do anything, really. Probably the best solution is to modify the processor so that it is autodetected at 200mhz fsb.


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[pIII_Man]

yup looks like my theory is correct...

We took our concerns to Gigabyte and were told unfortunately that they were unable to provide either aj AGP/PCI bus lock or the ability to manipulate the divider values due to the limitations of the KT600 chipset. The divider value is calculated based on the default value detected from the installed CPU upon boot up. Shame on you VIA!

<A HREF="http://www.hardcoreware.net/reviews/review-171-6.htm" target="_new">http://www.hardcoreware.net/reviews/review-171-6.htm</A>


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[phsstpok]

Yes and PapaSmurf already tried the only known mod, a pin mod. There are no corresponding bridge mods for FSB.

<b>56K, slow and steady does not win the race on internet!</b>

 

[pIII_Man]

maybe we need to preform the bus speed selection mod in conjunction with a vid mod so that the cpu posts at a high enough voltage.


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[phsstpok]

That's what I suggested. I think PapaSmurf tried this.

The problem is the mods on the backside of the CPU just finagle sense pins. The mods are akin to putting jumpers on the motherboard. They don't change the defaults of an XP per se. Only bridge mods can do this. Unfortunately there aren't any bridges for FSB on the CPUs; multipliers, voltage, SMP enable, cache size (on Applebreds) yes but no FSB.


<b>56K, slow and steady does not win the race on internet!</b>

 

[pIII_Man]

I have no experience with athlon xp cpus, however i thought that bridge mods were just cutting/closing little interconnects between pins on the cpu's substrate, therefore what would be the diffrence between connecting 2 pins with a wire or closing a bridge with some conductive paint?


If it isn't a P6 then it isn't a procesor
110% BX fanboy

 

[phsstpok]

I'm no engineer so I can't really explain it but there is a difference.

If you mod the bridges for multipliers or voltage you can still select any multiplier or voltage via BIOS.

If you do a mod via pins or socket you are hard setting the affected signals.

For example if you pull the 8X and 2X multiplier bits LOW they are now permanently pulled to ground. This means the bits cannot be reset to HIGH. In other words, you can no longer change those bits via BIOS. Now you can only manipulate the remaining bits, in this example 0.5X, 1X, and 4X multiplier bits.

<b>56K, slow and steady does not win the race on internet!</b>

 

[pIII_Man]

i doubt this is the case with voltage and bus speed selection. Example, the cpu sets the multiplyer, later you can modify settings in the mobo but ultimatly bridges pulled high and low tell the cpu what factor the multiply the fsb by only by modifying these bridges (wheather manually or through the bios) can the multi be manipulated. However things like voltage and bus speed selection are a little more passive, meaning that the motherboard doesnt really have to follow the bus speed selection pins, it can set any bus speed it wants no matter what cpu is in there withought having to trick the cpu by pulling pins high or low. Same with voltage the cpu really has no say other than <i>recomending</i> what voltage to run.


If it isn't a P6 then it isn't a procesor
110% BX fanboy