Vcore dropping under load...

[FIFO]

I don;t know if this is better posted here or in the mobo forum, but here it goes...

I just got my hands on a t-bird 900 and started overclocking it. The default Vcore is 1.75v and I hastily set the Vcore up to the max that the 8KHA+ goes, which is 1.85v. Now I know that the Vcore will fluctuate under load, but is it common to see it drop by as much as 0.09 from idle state. When Idle the Vcore is reported by MBM5 as 1.87 and when under heavy load (3DMARK2000 say) it dropped to as low as 1.78v.

NOTE: I currently have the ratio at 6 x 166 for 996 MHz, but cannot run at 10.5 x 100 for 1050 MHz. I am assuming my volt problem is the source of this limitation. I am running an Enermax model 365 (350 w) PS.

Now I figure that doing the v-mod on the 8KHA+ will probably make up for the drop, but I was wondering if anyone has any insight into this phenomenon.

Thanks in advance.

Jason.

 

[CraigRT]

That is just plain strange. I am not sure I've ever seen that or not... have you tried any other hardware components to rule anything out?

 

[FIFO]

I know. It certainly seems wierd, but I have seen it with my duron cpu as well. I have also tested this at standard FSB speeds.

It could easily explain why EPoX put Vcore available up to 2.2v on the 8K3A

 

[MrThompson]

This is a classic symptom of 5 volt line problems. In my experience, anything below 4.8 volts on the 5 volt line will cause this.

 

[FIFO]

Does anyone know if there is a trim pot on the Enermax model: EG365P-VE for the +5v line?

The +5 actually does not go below 4.8 but if there is a trim for that line I could try to turn it up a bit.

Thanks,
Jason.

 

[zephyrprime]

I sounds as though you don't have adequate wattage available. But you already have a 350watt so go figure.

 

[FIFO]

Tell me about it.

I just can't wait to get the sub-micro test clips to use to do the v-mod on the mobo. I am hoping that it will be enough to get the mojo going. Otherwise, 8K3A, here I come.

 

[ElusivePeon]

I think all P4 vcore drops when it is under load.

 

[Richardito]

You either need a higher quality power supply or more watts... Just get a 400+watt PS and you should be fine.

 

[Nate420]

There is another Vcore thread going around here, and it seemed that everybody (amd and intel alike) saw a small Vcore drop under load, but not near the .09V that you are seeing. My Vcore drops by about .02~.03V during prime95. It doesn't seem to matter what size PSU you have either, I had an Antec 300W and last week replaced it with an Antec True550 (550W) and still see the same Vcore drops.

 

[FIFO]

Thanks to everyone that has replied...

I did some further testing keeping my Vcore @ 1.85 in the BIOS and dropping my MHz to well below 1 GHz and the drop in voltage is MUCH less as I decrease the speed that I run the CPU.

I am assuming that this is one of the reasons that Aluminum interconnects are so much worse than Copper. I have had opportunity to play with an AXIA T-bird on my system and have had it running as high as 1.6 Ghz from a stock of 1.2 and there were no instabilities. Believe it or not, this is with the same Coolermaster HS + Delta 36 CFM that I am using now.

I guess that I will either have to find someone willing to part with their AXIA or get those damed micro clips and to the Vcore mod.

Thanks,
Jason.

 

[MrThompson]

Sounds like you definitely have a PSU issue here, especially if the problem disappears when you lower CPU speed. A volt mod is not going to remedy the situation, it will make things worse.

Last June I was having problems with an Athlon running at over 1.7 GHz and 2.1 Vcore. I spent some time on the phone with engineers from Intersil and PC Power & Cooling. The solution was a PSU with more current and tighter voltage regulation. Much of the discussion was about the power requirements for AMD chips once they pass the 2 GHz barrier. It?s going to see what happens with the Thoroughbred. It requires 45 amps! Running a pair of them on a dually board is going to be a chore for both the PSU and the motherboard voltage regulators.