p5w-dh Cpu or board reached its limit?

Specs in sig,

Here are the pics,







If I go any faster with the fsb than 420 it just fails superpi and wont even begin to orthos. Around 415 it wont boot but 420 it boots fine. I'm really stuck as to whats holding me back now, what voltages should the north bridge be at to handle this speed.

This is on air? Its probably the CPU, but this mb is not famous for FSB speeds over 400. Im sure you already knew that though. I would think at least 1.65 MCH will be needed for those speeds...Not sure about the north bridge. It doesnt seem to matter quite as much as the MCH from what I am seeing. What are your voltages now? Vcore? Temps? I would maybe drop the multi to 8 and see how much FSB you can push to rule out the CPU. At any rate, Im impressed you can get it to even post at those speeds. I have a crap chip, so im already jealous.

It took the clocks so well, at first i bearly had to up the voltage at all.. then i slowly had to up it every 10mhz fsb i added or so.. its around 1.53 volts now and 100% stable, orthos 24-7.

Idle around 50C core temp and 38die temp. I know its not my ram... so basicly the board is topped out then eh?

Wow 50C idle? Thats a typo right? Also, is that vcore whats set in the BIOS, or what it measures after droop? Regardless, I would say the MB is maxing out for sure. The question is whether the CPU is too. If 50C idle is not a typo, I would really do something about that heat problem. What are loaded temps?

Another question: Is your Orthos stable OC at 3.7, the 3.3 in your sig, or what?

52C for core temps is not hot.. the die idles at 38 which is cool.

I've got the Ultra 120 with a delta strapped to it.. and im rock stable all the time. Im pretty confident with the mounting, it whent togther good and I spread the as5 paper thing all way way acrost it... I dont like the dot method.

Here are pics from coretemp and speedfan.

The speedfan cpu temp is accurate for the die temp.




Orthos is stable at 3.7 24+ hours.

He says the load temps are around 60C on 1.53V [actual]. I'm not sure why the idle is 50C. I think he needs to try remounting because that seems pretty excessive to me, though the load temp is still reasonable.

I think you need to get a 965 baord to test this chip on. You could sell that board, get a P5B Deluxe and still break even.

I also think 50C is hot reported through Core Temp for idle. Its not hot loaded, but for idle its pretty up there.

Any one else have this overclock on an E6600 ?

yes

mine does 3.6 on 1.35v

my mobo starts to crap out at 420+ so I cant nail down 3.8 ... till next week that is...

go for 4ghz dont wuss at 3.8

Hello
I have my P5W DH @ 432 FSB x 8 Multi - 3.45ghz with 1.46vcore set in bios, 1.65v MCH and 2.3v for memory. Any less voltage on the MCH and the Memory it wasn't stable. Also disabled hyperpath. I have watercooling on the CPU and 40mm fan on the Northbridge heatsink.
The P5W DH doesn't overclock as high as the P5B Deluxe is because the P5B changes to the 1333 strap with looser timings @ 401 FSB. The P5W DH doesn't, it stays on the 1067 strap with tighter timings. So 420fsb on the P5W would have better performance than 420fsb on the P5B.

Are you sure you have the latest BIOS? 1901??

Mine changes FSB to 1333 at 333 x 9 on the P5W DH

Yeah, thats right.
I'm not talking about FSB though. The northbridge has internal timings like memory timings. As your FSB increases your northbridge's internal timing will change, loosen, at certain points to maintain stability. Different motherboards will change straps at different FSB points.
What it means is when you overclock and strap changes, your performance will actually get worse for a bit until the FSB increases enough to make up for the looser internal timings of the northbridge.

I found on the P5W DH that one of the strap changes occur is at 220-221 FSB. It goes from the 667 strap to the 1067 strap according to Systool which can read northbridge straps.

220 FSB on the 667 strap 2M Super PI = 00.59.406

221 FSB on the 1067 strap 2M Super PI = 01.00.468

As you can see the super PI actually got worse at the higher FSB. You can trick the board into not changing a strap by using clockgen and not rebooting. This is so we can see what would have happened if the strap didn't change.

221 FSB on the 667 strap 2M Super PI = 00.59.140 (clockgened up to 221 in windows to keep 667 strap)

The super Pi got better as you would expect, but in reality, at 221 the strap would have changed and performance would have decreased.