SE6: overclock problems (nothing extreme, just FSB 66->100)

[GinTonix]

Hi ya!

I got a "dead" (no picture said the guy, I was buying a 128MB SDRAM from him and I got the mobo for free) SE-6 (v.1.0) some time ago and finally, on Thursday I put together a websurf box. A piece of cake - just bios reset and it was ready to serve. It has a Celeron II 566, /w Golden Orb, Club 3D ATi Radeon 7500, and two pieces of 128MB 133MHz SDRAM's, in the two slots farthest from the processor socket. The first SDRAM is an old two-sided one, and the other one is a bit newer, single sided type of SDRAM. The latest bios (fresh from Abit web pages) has been downloaded and installed.

Now I'm having some peculiar problems:

If I run the system in auto settings (66MHz FSB, 100MHz for the SDRAM) everything goes fine.

If I rise the FSB to 100MHz to squeeze the last drops out of the Celeron, I have to also rise the voltage from the stock setting of 1,5V to 1,7V's before the system gets stabilized. Now, everything SEEMS to be OK, but for some obscure reason I don't seem to be able to run the 3DMark99MAX. It freezes every time. On the other hand, 3DMark 2000 is running perfectly and so do everything else. Except that one and only piece of software, which will crash every single time.

The power should be good enough, it's a 300W Seventeam (a coincidence - Zalman is selling the same type of PSU's under their own label in other countries, I mean its got the same product code, specifications, looks and everything).

Suggestions?? I have already superficially checked the caps and they SEEM fine (yes, I know it doesn't necessarily mean anything if they just look fine). Is there something I should know about Radeon on this mobo (or Radeon on ANY mobo, never had any before this one)? How about bios settings (AGP Aperture, AGP 2/4, delayed transactions and whatever the ones having an effect here are), I have kept them as close as possible to the default ones? Driver issues? Latest set has been downloaded from ATi pages and installed. Should the (mobo) chipset heat sink be equipped with a fan if I rise the FSB, it currently only has the ugly green thing on it? Could it be so that this particular Celeron doesn't like to be overclocked and she (processors are always female, think about their temper and strange ways of behaviour just gets confused if I try to make her work too hard? Well... if someone has something constructive to say, please do so. While waiting for that, I'll remove the single-sided DIMM and try running the test with the double-sided only. For some reason I feel it could be the memory (those two probably won't run together, not even if they were both in perfect condition).

 

[GinTonix]

"GinTonix" kirjoitti viestiss�

Hi ya!

...problems cut off...

And, it really is that only the 3DMark99MAX doesn't work. Am I missing some general knowledge here?

gt

 

[Colon Terminus]

Nothing extreme ... just a 50% overclock!

Does your 3DMark program run with the FSB set to 66Mhz.?

Try Prime95 ... Google for it. Does it run error free at 100Mhz.? If not, then back off on the FSB till Prime95 runs error free.

"GinTonix" wrote:

Hi ya!

I got a "dead" (no picture said the guy, I was buying a 128MB SDRAM from him and I got the mobo for free) SE-6 (v.1.0) some time ago and

finally, on

Thursday I put together a websurf box. A piece of cake - just bios reset and it was ready to serve. It has a Celeron II 566, /w Golden Orb,

Club 3D ATi

Radeon 7500, and two pieces of 128MB 133MHz SDRAM's, in the two slots farthest from the processor socket. The first SDRAM is an old two-sided one, and the other one is a bit newer, single sided type of SDRAM. The latest bios (fresh from Abit web pages) has been downloaded and

installed.

Now I'm having some peculiar problems:

If I run the system in auto settings (66MHz FSB, 100MHz for the

SDRAM)

everything goes fine.

If I rise the FSB to 100MHz to squeeze the last drops out of the Celeron, I have to also rise the voltage from the stock setting of 1,5V to 1,7V's before the system gets stabilized. Now, everything SEEMS to be OK, but for some obscure reason I don't seem to be able to run the 3DMark99MAX. It freezes every time. On the other hand, 3DMark 2000 is running perfectly and so do everything else. Except that one and only piece of software, which will crash every single time.

The power should be good enough, it's a 300W Seventeam (a

coincidence -

Zalman is selling the same type of PSU's under their own label in other countries, I mean its got the same product code, specifications, looks and everything).

Suggestions?? I have already superficially checked the caps and they SEEM fine (yes, I know it doesn't necessarily mean anything if they just look fine). Is there something I should know about Radeon on this mobo (or Radeon on ANY mobo, never had any before this one)? How about bios

settings (AGP

Aperture, AGP 2/4, delayed transactions and whatever the ones having an effect here are), I have kept them as close as possible to the

default ones?

Driver issues? Latest set has been downloaded from ATi pages

and installed.

Should the (mobo) chipset heat sink be equipped with a fan if I

rise the

FSB, it currently only has the ugly green thing on it? Could it be so that this particular Celeron doesn't like to be overclocked and she (processors are always female, think about their temper and strange ways of

behaviour

just gets confused if I try to make her work too hard? Well... if someone has something constructive to say, please do so. While waiting for that, I'll remove the single-sided DIMM and try running the test with the double-sided only. For some reason I feel it could be the memory (those two probably won't run together, not even if they were both in perfect condition).

--
gt

 

[Kent_Diego]

I have an SE6 on an old computer and it is very stable. I have PIII-733 at FSB=140MHz. Could go higher but the onbord sound will quit. Since everything works at rated speeds and you only have problem at overclock you can assume the overclock is problem. The OpenGL tests are most sensitive to memory/CPU errors. To isolate, try Memtest86. Also be sure not to overclock AGP/PCI bus too much. Experiment with memory timings. Be sure to Assign IRQ to VGA, disable Passive Release and VGA Pallet Snoop, and not cache anything. I replaced a bulging/leaking capacitor on mine.

-Kent

 

[Kent_Diego]

Are you using the DIP switch to set frequencies or Soft Menu? The best is to set switch to allow BIOS to set frequency as you will have a lot more control.

-Kent

 

[Colon Terminus]

Yeah, but you're only overclocking about 5%. The OP is looking at a 50% OC. There's an enormous difference.

"Kent_Diego" wrote:

I have an SE6 on an old computer and it is very stable. I have

PIII-733 at

FSB=140MHz. Could go higher but the onbord sound will quit. Since everything works at rated speeds and you only have problem at overclock you can assume the overclock is problem. The OpenGL tests are most sensitive to memory/CPU errors. To isolate, try Memtest86. Also be sure not to overclock AGP/PCI bus too much. Experiment with memory timings. Be sure to Assign IRQ to VGA, disable Passive Release and VGA Pallet Snoop, and not cache anything. I replaced a bulging/leaking capacitor on mine.

-Kent

 

[GinTonix]

"Kent_Diego" kirjoitti viestiss�

I have an SE6 on an old computer and it is very stable. I have PIII-733 at FSB=140MHz. Could go higher but the onbord sound will quit. Since everything works at rated speeds and you only have problem at overclock you can assume the overclock is problem. The OpenGL tests are most sensitive to

memory/CPU

Yes, everything goes fine without overclock. Probably it is the CPU - it seems that it is not a good sample of those early FC-PGA Celerons (a 566 one). Not all of them can take the 100FSB and it seems I have one of them. Maybe more voltage would do the trick but I'm not going to fry it. I'll rather have my kids (5,5 and 2 yrs have a live 566 Celeron in their games machine than a dead one in the closet and a 350 P2 in the machine instead.

errors. To isolate, try Memtest86. Also be sure not to overclock AGP/PCI

bus

Remind you, FSB 100 is one of "the safe ones", that's why I'm either going to get 100 or not going at all. At 100MHz FSB everything is in specifications, i.e. AGP, PCI and memory.

too much. Experiment with memory timings.

Timings are as slow as I can get them on SE6.

Be sure to Assign IRQ to VGA,

Sure...

disable Passive Release and VGA Pallet Snoop, and not cache anything. I

Hmm.. passive relase disabled..? Gotta try that one. Pallette snoop has always been disabled in my PC's.

replaced a bulging/leaking capacitor on mine.

All of them seem to be doing fine, and MBM5 (installed it AFTER getting the problems) shows that all the voltages seem to be OK.

-Kent

 

[GinTonix]

"Colon Terminus" kirjoitti viestiss�

Nothing extreme ... just a 50% overclock!

Yup, for the early FC-PGA Celerons it is nothing extreme. Just like the good old Celeron 300A - if you had one which wouldn't run at 100FSB you'd have been really unlucky.

Does your 3DMark program run with the FSB set to 66Mhz.?

Yes, it does. I know - just don't overclock and that's it. Maybe the SE6 just cannot force the CPU to run on 100MHZ FSB and that's the end of the story.

Try Prime95 ... Google for it. Does it run error free at 100Mhz.?

Gotta check that one out. Thanks for the tip.

If not, then back off on the FSB till Prime95 runs error free.

I'll go back to 66MHz FSB if it won't go flawless at 100. Or then maybe I'll try the P3 650E at 133 FSB.. the Celeron would then go to a BF6 which gives a lot more oppoturnities than SE6 v1.0.

 

[GinTonix]

"Kent_Diego" kirjoitti viestiss�

Are you using the DIP switch to set frequencies or Soft Menu? The best is to set switch to allow BIOS to set frequency as you will have a lot more control.

I've gotten used to do the trick in bios, just for the reason you described. Besides, it is a lot easier.

-Kent