KT7A-RAID/Duron o/c probs. 133 FSB no go, settings have no effect

[vagabum]

Okay.. system runs fine at 700 (7 x 100), however when I try to bump the FSB up to 133 it refuses to boot or even give a display. Have to wipe the CMOS to change it back to 700. It will boot at 100 FSB but not at 133. Any ideas why?

Changing the settings to 750(not even user define) also doesn't seem to work. This Duron seems to be stuck on 700 permanently argh! Ideas anyone?

EDIT : speed hold IS disabled. Only movement I can get is when I use the FSB plus and move it up by 1mhz notches. Even underclocking it by selecting 650 instead of 700 doesn't work, it just seems to be stuck on 700 regardless of multiplier. Moving to 133 just makes it refuse to boot *sigh*

YES I did pencil in the L1 bridges, however I may not have done it properly. Should I still be able to change the multiplier in Softmenu even if the CPU is still locked?

 

[AMD4ME2]

its probably locked.. you may have to do the pencil trick on the L1 bridges, or get a conductive ink pen, or defogger repair fluid to join the L1 bridges in order to overclock.

Read this!

 

[DarkMajiq]

Even if the CPU is still locked, you can change the multiplier, but it won't have any effect. I would suggest re-penciling your L1 bridges. Also, if you're trying 7x133=933MHz to start out, that's not a good idea, you'd be better off to do 5.5x133=733MHz, because it may not be able to handle the much higher speed. Start low and take it up.

 

[vagabum]

Update : been slowly stepping up the FSB, it will boot at 107 giving me 750, highest it will post at is 112. This is at 1.7v, 1.85 and pushing higher just gets it to boot further into Windows before crashing. Still, better than nothing.

I redid the pencil marks, reapplied grease etc. Getting lower temps now too but the multiplier still isn't going.

Even though the pencil's not working (maybe the graphite just isn't conducting) think I should try again with a conductive pen? this is really annoying

 

[Turin39789]

are you sure you have a kt7a and not just a kt7?

 

[vagabum]

Third time around seems to have kicked in some life. I'm now running at 750 on a 7.5x multiplier which means it's not a locked down CPU in anyway, just crappy penciling! Woohooo!

Only the 7.5 and the 9 multiplier (unstable) seem to be working so far. Everything else isn't registering. I'm using a 5h pencil and it seems this one just isn't cutting it, plus the lead is still a little bit too fat. I'm going to go out and get myself a conductive pen now. Wish me luck!

and yes it is a KT7A-RAID *stares at the big cihp on the board that says KT7A-RAID, looks over at the manual and box, checks the orange highpoint controllers and the other neon stuff, gazes at the fan on the northbridge I think it is..*

 

[vagabum]

Just before I left, I thought I'd give 9.5 a try and what the heck it worked! messing around some more I bumped the FSB up to 102 and I'm now running at 970 pretty stably. 103 just would not boot cleanly (registry error) so I'll burn this in and stress test it for a day or two before taking a conductive pen to it. The 10 multiplier doesn't work properly and sets it to 9 but this could well be as good as it gets.

Still hoping for that 1ghz!

 

[rarebear]

Start lower I did the same thing left mine at 6.5x133/33 and crashed the cmos.. ans now for some reason it is fine at 7.5x133/33 it will got to 7.5x 136/34 but that is it so far..

 

[Gum]

I have the exact problem as Vegabum. Difference is I have a Duron 800.
Vegabum, are you using PC133 SDRAM?

I shade the L1 Bridges but only 7, 9, and 9.5 mult works. Set it to 7.5, 8.5 and system will not boot.

I am also unable to overclock higher via FSB. My Mushkin PC133 SDRAM with Samsung -G6 chips should at least be comfortable at 133Mhz but this is not the case. I tried to set FSB/PCI at 133/33 and mult at 7, so 7x133=931Mhz, but system won't boot. But if setting is 100/33 and I overclock via FSB-Plus to 20 and mult to 8, so 8x120=960, system boots and works normally!!

Now system runs at 100/33 FSB-Plus at 15 and mult at 8, so it's 8x115=920Mhz @ 1.7vcore and 3.4v vio. Got to try overclocking further!!

 

[vagabum]

I would start lower except the 6.5 multiplier doesn't work, only 7, 7.5 and 9, 9.5 works. Still I've had no problems at 970 so far. I'm running on 1.85v but once I make sure this is okay I'll start lowering the voltage

Yes my RAM is PC133, Micron stick CAS 3. However I'm running it at CAS2 now with no probs. So close yet so far. That's quite interesting about the bus, bus speed seems to be my problem. Anything above 105 FSB just will not boot, and 103. I'll post again if anything new develops or after I take a conductive pen to it

 

[johncar]

Vagabum,
The FSB Multiplier is controlled by a "four" bit binary pattern to the multiplier module in the cpu. Each bit "Adds" a particular value to a "base" of 3X when the bits are set HI. The "Adders" are .5X, 1X, 2X, and 4X. The algorithm for the FSB Multiplier is...

TotalX = 3X + Sum of Adders set HI, (except add 8X when TotalX = 3X thru 4.5X).

So you need the following for 6.5X thru 9.5X settings....
**6.5X = 3X base + .5X + 1X + 2X
7X = 3X base + 4X
7.5X = 3X base + .5x + 4X
**8X = 3X base + 1X + 4X
**8.5X = 3X base + .5X + 1X + 4X
9X = 3X base + 2X + 4X
9.5X = 3X base +.5X + 2X + 4X

We ** the 6.5X, 8X, and 8.5X settings which you say you cannot get as they are the only settings which need the 1X Adder...right?? So the most probable problem is that the L1 bridge connecting to the 1X control circuit is "not closed". It is the second L1 bridge from the left, ie, L to R = .5X, 1X, 2X, 4X.

Use something better than a pencil, try the rear window defogger repair kit...conductive solution...details at......
http://members.nbci.com/candjac/index.htm although server seems to be having on and off or FORBIDDEN problems...keep trying.
Hope this helps. (The value of "Understanding&quot.
John C.

 

[Democritus10]

I've been having similar problems with my 1GHz athlon when I try to run it at 133FSB x 7.5.
Usually, I can only get it to run at 133FSB at 6.5x or lower. Higher multipliers don't even allow the display to come on, until the CMOS is reset. And yes, I've used rear window defogger paint to unlock the multipliers (but who knows, maybe not completely).

HOWEVER, on several occasions, I've gotten the chip to run perfectly at 133FSB x 7.5, but this is only possible after I boot to Win98SE under default settings (100FSB x 10), remain in that environment for a while, restart without shutting down, enter BIOS and change settings to 133FSB x 7.5 and reboot. This process does not always work, but this has been the only way that I've gotten it to run at 133FSB x 7.5 (i.e. I can't get it to work when I power on fresh and enter BIOS without booting the OS). And once I shut down, the computer will not startup again until I clear the CMOS.

I've played around with the voltages, to no luck. And my system is properly cooled with 2 case fans, 1 HDD fan and a Taisol fan/heatsink using grease (temp is usually between 90 and 105 F when checked in BIOS). On a tangential note, the Via Hardware Monitor program under Win98SE usually gives me a CPU temp of about 10 degrees F lower than that stated in the BIOS...what gives?

Thanks.

Tbird 1GHz
Abit KT7A-RAID
Quantum Fireball Plus AS 30GB
Soundblaster Live
Riva TNT2 Ultra 32MB
Teac 8x8x32x
Apacer PC133 (2 x 128MB, 1 x 256MB)

 

[UnDave]

Excellent!!! Thanks for the multiplier info Johncar!!! that explains a LOT!!!!! :-D


-UnDave

 

[tjll11a]

what would happen if you just stuck in a duron 800 (retail) win fop 32 fan on a epox 8kta3 and set it to 133 mhz. You think it might post?

 

[hoihtah]

i've tried iwill kk266 kt133a mobo with duron 800.
and it won't post at 133.
i donno what the problem is.
i have 256mb ram cas 2 133mhz.

help?

 

[Rack]

A 5H pencil is too hard (not enough graphite and too much clay).
Try with a #2 pencil, preferably mechanical (just did it to a t-bird
900 last night, worked like a charm). It took me 2 tries with my
own t-bird 800 (was missing the 9 and 9.5 multipliers), and it's
worked for 2 months now.

George

 

[tjll11a]

I used a locked duron 800 in a epox 8kta3 and set vol to 1.85 and just threw it at 133 to see what happens. To my amazmment it booted up but immedialy went into windows and got a registry error. You think i might have a good cpu here? You think if i low vol to maybe 1.825 it might help. Just kinda dreaming i might not have to lower mult. Probably will have to use defogger kit to unlock and try mult of 7. But I was just wondering if lowering vol might help.


Thanks for all your help
Jason

 

[Mikaelb]

Nah . Mine does the same. However the highest really stable setting is 7X140=980 at 1,8 V .
Probably I could get it stable at 7X145 at 1,85 with a better cooler like a FOP32 or similar.
lower voltage only helps if cooling is borderline otherwise higher voltage help up to 1,85 V.

 

[Mikaelb]

I also have KT7A-raid. Seems like memory probs or a picky harddrive is your culprit. If I go higher then 35 Mhz on PCI I have to disable DMA in Bios on one of my drives or strange things happen.
Try this and try also CAS3 to see what the culprit is. The board can take every setting up to 150 Mhz so....

AMD Duron800@7X140=980 and 1,8 V

Cas2 does not do so much in my opinion so do not sweat if it does not work. Try setting the other memory tweaks to best performance instead and see the performance take off!!

 

[chriscraft]

FSB Limits on all Motherboards with the VIA KT133 chipsets are not able to overclock the FSB much
more than 103-105!!

This is a VIA chipset problem.... nothing else you do can change this..... except by a new KT133A chipset board...

The new KT133A chipset boards CAN overclock their FSB over 105...in fact the two recent tests I've seen, both say 150 FSB is stable on both the ABIT KT7A-RAID, and the ASUS A7V133

 

[Losty]

when you guys just pop in a duron 800 and change the fsb to 133...sometimes it doesn't work...well ... does it occur to you that it won't go that high? that's past 1ghz no? why not lower the mulitplier to 8 or 7.5 ...

i had the multiplier problem on my tbird...couldn't get anything but 9.5 and 10 ... did the pencil trick again and i got them all to work, although i recommend you all use the conductive pen / paint or locitite methods for security reasons and so you don't have to look for it when random day when your cmoputer won't boot...