83KA+: An overclocker's board? Sure doesn't seem like it.

I have all the same hardware as I did with my Iwill. The Iwill would overclock like crazy. It ran perfectly stable at 160 Mhz FSB, 1.84 Vcore, 2.8Vddr.

The 83KA won't even POST at 150Mhz FSB. Not even with the CPU and RAM voltages at their maxes (2.2Vcore, 3.2Vddr).

There's something very wrong here.

Stupid question here, but do you have the FSB jumper set to 133?

Seems to be an awful OC. Personally, I've tried to OC my FSB, but it only went up to 141 with no Voltage changes. As soon as I get my case, I'm going to try for more.

Cheers,
F

I'm not sure, really.

What should I set it to if not 133? I have auto-detect clock speed turned off in the BIOS, so that shouldn't be affecting it.

Just so that we're writing on the same frequency:

There are two hardware jumpers on the mobo, one is for CMOS reset, and the other is for default FSB speed. I think this is near the AGP slot. The mobo ships with the FSB jumper set to 100. If you haven't changed it yourself (or for some other reason) then it still might be at 100, which could be giving you issues.

If it is set at 100, I would of course suggest that you reduce all your voltages before booting with the jumper at 133.

Good luck.

F

i thought on the 8k3a the default jumper is at 133, and the max is 166....isnt that what that jumper does, kt266 boards changed from 100 to 133, and i think 333 boards change from 133 to 166

That's incorrect. The jumper is 100/133 on the KT333 (at least on the 8k3A+ anyway). Check that jumper on the mobo. Needs to be on 133. Mine arrived set on 100.

If your memory will allow it (prob not with PC2100, but who knows?), bump it over 166Mhz and see what happens. At that point the dividers kick in and woohoo!

Check these specs out below. I have had it to 1800Mhz (190 Mhz x 9.5), but am now inching it up to find the most stable speed. Could prob get it over 1800, but haven't tried. Not ready to push TOO hard yet.

This board can, indeed, overclock nicely.

It'll POST at 166Mhz, but it'll lock up as soon as Windows starts.

Even with the jumper set to 133, it still won't POST at 150.

DmdStar said:

That's incorrect. The jumper is 100/133 on the KT333 (at least on the 8k3A+ anyway). Check that jumper on the mobo. Needs to be on 133. Mine arrived set on 100.

If your memory will allow it (prob not with PC2100, but who knows?), bump it over 166Mhz and see what happens. At that point the dividers kick in and woohoo!

Check these specs out below. I have had it to 1800Mhz (190 Mhz x 9.5), but am now inching it up to find the most stable speed. Could prob get it over 1800, but haven't tried. Not ready to push TOO hard yet.

This board can, indeed, overclock nicely.

Click to expand...

Much the same as mine. Found a nice home around 180x10. Have gotten the Crucial 2100 to 191 and the samsung to 192. At 180-183 all ram settings are at the fastest possible.

multi down, fsb up. Go over 166 to get the divisors working. Up the cpu voltage for stability. Ram voltage will not need to be very high at 166. Try about 2.7v on the ram. What multi have you been trying to use?

Crazy Jayhawk said:

It'll POST at 166Mhz, but it'll lock up as soon as Windows starts.

Even with the jumper set to 133, it still won't POST at 150.

Click to expand...

Posting at 166 but going no further sounds like your memory can't handle it.... If you can, stick some PC2700 in there, and I bet it'll go. Upping that voltage may help too.

I've found that the 8K3A+ seems to have probs overclocking between 133 and 166... I couldn't get it over 140 without it locking up on me, but once I tried over 166, I could go all over the place...

Crazy Jayhawk said:

I have all the same hardware as I did with my Iwill. The Iwill would overclock like crazy. It ran perfectly stable at 160 Mhz FSB, 1.84 Vcore, 2.8Vddr.

The 83KA won't even POST at 150Mhz FSB. Not even with the CPU and RAM voltages at their maxes (2.2Vcore, 3.2Vddr).

There's something very wrong here.

Click to expand...

Just a reference info for JayHawk,

The difference may be in the RAM timing setting between your Iwill and new 8k3a (in other word, the same 2-2-2-5-1 setting on Iwill stress less the same 2-2-2-5-1 setting on 8k3a). Even though the hardware looks all the same but Iwill may have softer stress on timing compared to 8k3a's

That's why your Iwill ran ok with 160 mhz because the time setting for RAM is softer than that on 8k3a. You probably will be able to check it by running memory benchmark on same FSB setting for both board. I would guess Iwill's benchmark number will be little lower.

I can not tell what brand of your PC2100 RAM. I guess going in 160s FSB with that RAMs already reach it limitation. I have tried few different RAMs on my 8k3a+ (rev 1.2) and bellow is just a referecen for you:

Among many RAMs I have, the Samsung PC2700 (ATP designed PCB) performs the best and did reach 215 mhz with setting at "normal" and 201 Mhz with setting at 2-2-2-5-1 (turbo plus 1T on time command).

Attached is a image of Sandra report indicating memroy running at DDR 402.


Epox 8k3a+ (rev. 1.2)
AMD 2000XP
Corsair XMS3000 -- 199 @ 2-2-2-5-1
Kingmax PC2700 -- 180 @ 2-2-2-5-1
TwinMos PC2700 (Windbond chip stamping A version) -- fail @ 2-2-2-5-1
Samsung PC2700 (ATP designed; still waiting for Samsung original) -- 201 @ 2-2-2-5-1
CoolerMaster ATC210 modified with 8 fans
Thermalright AX-7 with Delta 80CFM
PC mods rheobus 4 channel fan controller (17W per channel)
digiDoc5 for temp hardware monitoring
2 x IBM ATA 100 HD
Toshiba 16x DVD and Pelxter CD-ROM

MV99

Interesting. The ram results at 201 fastest timings are about what i am getting at 180 (2550) fastest timings on the 8k3a. I tried some different bios's but found that the mem scores where lower. This may be why I am not breaking 200fsb. The ram timings may actually be faster. At 191 the crucial breaks 2660.

Silver said:

Interesting. The ram results at 201 fastest timings are about what i am getting at 180 (2550) fastest timings on the 8k3a. I tried some different bios's but found that the mem scores where lower. This may be why I am not breaking 200fsb. The ram timings may actually be faster. At 191 the crucial breaks 2660.

Click to expand...

Most overclockers here using how high FSB a RAM can go @ either 2-2-2-5-1 or setting at "normal" as idication of how fast their overclocked machine.

But my view of it is little different, highest FSB that passes Sandra normally doesnt give me the highest efficency (for example, I will get much better efficiency with setting at 183x10 @ 2-2-2-5-1; I call it "sweet spot").

I only use highest FSB to judge how much OC a ram can take. It is almost for sure that no machine can run stable under the highest FSB that passes sandra memory test.

"sweet spot" on setting may not give me the highest Sandra score. But I know it will give me a indication of stability. So, I normally run my system @183x10 instead of going after with something like 200x9.5.

As for your amazing score on Crucial PC2100 (I origianlly thought there would be impossible to make that RAM go over 180mhz but you proved me wrong), I have to take my hat of your effort on machine setup and also for those engineers at Crucial. They must have done a very good job on designing the PCB to fully take advantage of Micron chips.

Once I receive my Samsung original, I would like to do a comparision on performace with the ATP made Samsung PC2700 to see how much difference I can see between these 2 PCB design.

The "original" should perform better. Nerverthless, I have already expressed to ATP how I like their RAM and thank for them to give me this RAM at thier cost. BTW, they are actually our neighbor. We sub lease part of our space to them.


Epox 8k3a+ (rev. 1.2)
AMD 2000XP
Corsair XMS3000 -- 199 @ 2-2-2-5-1
Kingmax PC2700 -- 180 @ 2-2-2-5-1
TwinMos PC2700 (Windbond chip stamping A version) -- fail @ 2-2-2-5-1
Samsung PC2700 (ATP designed; still waiting for Samsung original) -- 201 @ 2-2-2-5-1
CoolerMaster ATC210 modified with 8 fans
Thermalright AX-7 with Delta 80CFM
PC mods rheobus 4 channel fan controller (17W per channel)
digiDoc5 for temp hardware monitoring
2 x IBM ATA 100 HD
Toshiba 16x DVD and Pelxter CD-ROM

MV99

More BS from my 8K3A - it locked up in the BIOS screen at 140 FSB.

I'm beginning to think this board was a waste of money.

Crazy Jayhawk said:

More BS from my 8K3A - it locked up in the BIOS screen at 140 FSB.

I'm beginning to think this board was a waste of money.

Click to expand...

For all the things I wrote earlier was trying to tell you that you may need to check your PC2100 Ram.

I think it could be your Ram not the mobo prevent you from high FSB. Unless you have got a defective 8k3a. If so, have you checked with the place you bought the mobo and see if it is possible for a mobo exchange?

If you look at Sliver's Sandra report on his Crucial PC2100. To me that is a extreme case (he is quite good at OC and the FSB he got with that RAM was amazing to me). That should not be considered a typical thing you get from any PC2100 class RAM...

If you can get a good name brand's PC2700 or higher, you will definitely see why many overclockers are using 8k3a.

I hope you can resolve this issue soon and enjoy the OC on 8k3a.



Epox 8k3a+ (rev. 1.2)
AMD 2000XP
Corsair XMS3000 -- 199 @ 2-2-2-5-1
Kingmax PC2700 -- 180 @ 2-2-2-5-1
TwinMos PC2700 (Windbond chip stamping A version) -- fail @ 2-2-2-5-1
Samsung PC2700 (ATP designed; still waiting for Samsung original) -- 201 @ 2-2-2-5-1
CoolerMaster ATC210 modified with 8 fans
Thermalright AX-7 with Delta 80CFM
PC mods rheobus 4 channel fan controller (17W per channel)
digiDoc5 for temp hardware monitoring
2 x IBM ATA 100 HD
Toshiba 16x DVD and Pelxter CD-ROM

MV99

JayHawk,

What frequency are you changing? Are you keeping both FSB and Mem frequencies the same? Do you have an unlocked CPU?

I ask this because it sounds like you are simply upping the entire FSB, resulting in (for example) 140/140. It is probable your CPU can't take it from the Epox (if it did on previous boards). If your CPU is still locked, then you probably won't get far on this OC. The only way you'll be able to OC highly is on the memory frequency, which you can set on another page in the BIOS.

Good luck!
F

francisamf said:

JayHawk,

What frequency are you changing? Are you keeping both FSB and Mem frequencies the same? Do you have an unlocked CPU?

I ask this because it sounds like you are simply upping the entire FSB, resulting in (for example) 140/140. It is probable your CPU can't take it from the Epox (if it did on previous boards). If your CPU is still locked, then you probably won't get far on this OC. The only way you'll be able to OC highly is on the memory frequency, which you can set on another page in the BIOS.

Good luck!
F

Click to expand...

The FSB and memory frequencies are the same. The RAM I have was able to handle a 160Mhz FSB on the Iwill with a lower voltage, so I don't understand what's keeping it from running at 140Mhz FSB on the Epox.

If it's the CPU that can't go that high, that doesn't make sense either. It too was able to handle a 160Mhz FSB, also with a lower voltage. It is not unlocked. My last attempt to unlock it resulted in me needing paint thinner to clean up the conductive paint after it leaked under the tape.

Run out of Ideas

I understand how you feel. I have the same feeling. After reading about so many people doing these tremendous locked OCs on FSB alone up to 170x or whatever on aircooling, no additional volts, suddenly, when it is my turn to try, it dries up at 140.

I am using good components, as far as I know. But, it seems that whenever someone says "degree of OC depends highly on luck" it means I will probably get all the least overclockable components in the world.

So, good luck
F

I have found the same to be true. At least as far as I can test on a 10x. 183 (though it may be a little higher) seems to be the "sweet" spot.

Silver said:

I have found the same to be true. At least as far as I can test on a 10x. 183 (though it may be a little higher) seems to be the "sweet" spot.

Click to expand...

I just want to be little bit more on safer side.... I did burn one CPU before; but not casued by OC. It was because I was too sleepy and forgot to re-mount my AX-7 and it fired my CPU in 5 sec...

By the way, I notice that you are using Pelt approach. I am thinking that as my next project because my system is just too loud. One thing concern me is the condensation from Pelt approach.

Can you give me some input on how that condensation issue should be address? I like the look of watercooling cases offered from "overclocker hideout" (http://www.overclockershideout.com/)
and if I do go for it, I would like to know how to deal with condensation.


Many thanks in advance!!

MV99