• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Overclocked Dullie FSB issues.

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Karifan

Member
Joined
May 23, 2003
Overclocked Dullie FSB issues. (Now running, Check the benchmarks)

I recently did the L3 and L5 mod to two Athlon XP 2100 (tbred B).
These processors have been tested in single cpu systems(on a A7V8X motherboard) and ran stable at 2300 mhz each at the following configs.
144 x 16 = 2304 mhz.
178x13 = 2314 mhz.


I have also ran each cpu in the dual cpu board(A7N266D) at 140 x 16 = 2240 mhz. I have tested each cpu in each of the available sockets on the board and it ran stable including all the benchmarks, Sisoft Sandra, 3Dmark2001 and PcPitstop.

My problem is when I put the cpu/s together in the dual, it will not boot at anything higher than 131x16 = 2096 mhz per cpu.
If I try to start it at 133x16 it gives me a windows error that the register file HIVE could not be loaded and may be corrupt.

Initially the board would not boot up at anything higher than 125x16 = 2000 mhz but I narrowed it down to a voltage issue.
I had the voltage maxed out at 1.87 and 1.8 for cpu 0 and cpu1 respectivley.
Now the voltage for the cpu/s is set at 1.7 and 1.65 for cpu 0 and cpu1 respectivley and is running stable at 131x16.
What could be the problem? its only 2 mhz on FSB but its causing a lot of problems.
I was running dual Athlon MP 2000 before this and I have ran stable at 12.5x144. Should I back the multi down to 15.5 or 15 and run a higher fsb or should I just be happy with this?

Appreciate any help in advance.
 
Last edited:
I guess that the CPU's aren't completely stable in SMP operation any higher than 2100 MHz. This could be because of the CPU's or the motherboard/power supply. It's possible that the motherboard/power supply can't supply enough juice (consistently) for operation at 2300 MHz.

I would suggest just running with it for a while. Maybe after it has some burn in time, you may be able to up the FSB a bit more.

What Power Supply are you using?

What heatsinks are you using?

What are your full load temps?

Have you run two instances of Prime95 to test for stability?

Did you run Prime95 on the chips before, when they were running 2300 MHz on a single board?
 
This forum is the BEST, and you gotta be the quickest memeber cmcquistion. lol :)

Ok one by one.

1)What Power Supply are you using?
I do not rememeber the name of the power supply but its an Intel authorised 430 watt psu.
I did a lot of power monitoring and with 4 fans, 4 HDs in Raid 0 a Ti500 and Ti4200 as well as 3 pci slots filled including two Thermal Take Volcano 7 fans the voltage was always normal in SiSoft sandra, I only have the Hds and two fans running right now.
I do not think its a voltage issue, atleast it does not seem to be.
Reason being that I had to back the voltage down to 1.7 and 1.65 gradually to get it to boot at 131x16 vs any other higher voltage at 125x16.

2)What heatsinks are you using?
TT 7 Volcano 7+

3)What are your full load temps?
Currently after using AS2 on the cpu/s its 39/38 c on idle and about 42/41 to 44/42 celcius at full load for cpu0/cpu1.

4)Have you run two instances of Prime95 to test for stability?
I do not have that program but do intend to get it, any advise as to where I can get it cheap?

5)Did you run Prime95 on the chips before, when they were running 2300 MHz on a single board?
See #4

Thank you very much for your prompt response cmcquistion.
 
You can get Prime95 from www.mersenne.org

430 watts isn't necessarily enough juice. A lot of dual AMD owners have had problems with 400-450 watt Power Supplies. The 430 watt Enermax is a notoriously awful power supply, for example.

The PSU may not be the problem, but it is definitely suspect.

Sometimes too much Vcore is a bad thing, too. That's what you figured out. I suspect that the CPU's were probably scorching hot from all the voltage and that's why they weren't stable.

You should ALWAYS run Prime95's torture test to test for overall system stability. It's easy to overclock and run a system with CPU's that are making erroneous calculations. That's what Prime95 is good for. It stresses the CPU's, but it also checks their data. If they come back with bad calculations, then the program stops and tells you. This is a better method of testing for stability than other programs, because other programs won't stop and tell you if the data is screwy.

For a dual, you will have to run two instances of Prime95. I wrote a guide a while back on how to set this up. It has been copied and pasted into many threads, so it shouldn't be hard to find. Do a search for "Prime95-2.exe" and you will find it.
 
I ran into the same kind of problem with my big arse Antec PSU and MSI dually mobo. My tbred b1700's would run way faster at any voltage in my old KR7A than they would when both were together in the dually. I'm rock stable up to 2100mhz with both in the dually but both would go way higher in the abit mobo. I'm thinking it has to be onboard power or chipset limitation.
 
I have bios revision 1009, could a bios update fix the problem?
I have noticed that there was an issue with bios revisions when going to higher clock speeds on faster AMD cpu/s(As in with faster cpu support there is always a bios revision update).

Its either that or the PSU is what I can narrow it down.
I just thought that maybe someone had the same issue like I do and knew what the problem was b4 hand.

Thanks again, I will see if this update takes care of it and will keep you posted.
 
spldart what are your system specs and what fsb x multi are you running on those 1700, vcore and tempratures?
Any other info would be very helpfull, like how many Hd/s and other components.

Ty
 
Generally, newer BIOSes add support for faster CPU's that require higher multipliers. If your board recognizes the multiplier, then that may not be the issue.

It couldn't hurt to update your BIOS, of course, but I'm not sure that would fix the problem.

Keep in mind, when you're running two CPU's that your overclock can only be as strong as the weakest link. It makes perfect sense that both CPU's can overclock slightly higher, when run in single CPU mode. The two, together, are never going to be able to overclock together, as high as they could individually.

The question is, how much of a margin is acceptable to you?

Right now, you're seeing a 200 MHz margin. If that is unacceptable, then I would suggest trying another PSU, first. From what you have said, so far, that seems like the weakest point in your system.
 
OK good news.

I updated bios to 1010A7MD revision 1010 and also changed the voltage on the memory.
Not sure which did the fix but now I am running at 137x16 = 2192mhz.
The memory voltage was set to 2.6 and all I did was back it down to 2.5.
Now I will try to increase the voltage on the cpu/s from 1.7 and 1.65 to 1.75 and 1.7 and see if I can get it upto to 145x16.

I might also need to replace my psu just for stability.

Will keep ya posted and ty for your help.
 
I am now running stable at 140x16=2240 mhz per cpu with the vcore increment on the cpu/s.
The vcore is now set to 1.75/1.725 for cpu0/cpu1 respectivley.
I ran a lot of benchmarks but not Prime95. As soon as I figure out how to use it(I have not even tried it yet to be honest) I will run it.
I know that there is a trick to running it on a dual cpu, do I fire up two instances and just set the processor affinity for each one on its own cpu? or is there another trick?

Will keep ya posted.

I will post the benchmarks for sisoft sandra and pcmark2002 as well as 3dmark2001 and 2003 here so stop by when you have a chance.

Benchmark Results
 
Last edited:
Yes set each procesor affinity and also make sure each version i not trying to run using 75% of the memory.
i might have an out of date version but click: options / cpu. and then set the memmory.
 
Karifan said:
I am now running stable at 140x16=2240 mhz per cpu with the vcore increment on the cpu/s.
The vcore is now set to 1.75/1.725 for cpu0/cpu1 respectivley.
I ran a lot of benchmarks but not Prime95. As soon as I figure out how to use it(I have not even tried it yet to be honest) I will run it.
I know that there is a trick to running it on a dual cpu, do I fire up two instances and just set the processor affinity for each one on its own cpu? or is there another trick?

Will keep ya posted.

I will post the benchmarks for sisoft sandra and pcmark2002 as well as 3dmark2001 and 2003 here so stop by when you have a chance.

I wrote a guide to doing this, here in the SMP forum. It has been copied and pasted into several threads. Do a search for "Prime95-2.exe" and you'll find it.
 
Hey there ppl, finally I get some images of the benchmark results.
It seems like the cpu/s are broken in now, because they are scoring much higher than I had seen at the initial phase of the overclock.
All benchmarks are ran at 138x16=2208 mhz and all results exceed that of a Dual Athlon MP 3000. Oh yea I can not believe what I see. The dual Athlon XP 2100s are recognised as dual Athlon MP 2600 bcs of the MP mod and the 16 x multiplier mod.


Benchmark Results

The system has a higher potential I believe with a psu upgrade as well as a vcore increase. I have ran @ 16x143=2288 mhz stable with the current vcore settings, on my single cpu board I have tested each cpu at 16*149=2384 mhz stable with a temp of 45-50 idle vs full load.

Thanx for you help and all suggestions are welcome
 
Last edited:
Back