Lapping A64 IHS and MSI K8N Neo2 Platinum Chipset (DONE, upto 16c Lower) *56k beware*

[javascripterror]

[QUOTE=phextwin] I'll beleive you
But what was load? And why were you idling at 51°C ?[/QUOTE
The IHS was concave that why it was idling at 51 degrees. Would you put a load on a CPU that idles at 51?
Here is how much that had to be removed to smooth the IHS to level.

Click to view full-sized image!
Hosted by UGBox Image Store

[phextwin]

Whoa.

That's truly terrible.

I just did a dodgy flatness test with my a64 IHS with a metal ruler, seems okay to me. *shrugs*

[chisholm]

Im not accusing you of being dishonest, but 16 degrees seems WAY too much of an improvement for not even removing the IHS

I know the neo2's temp probe is unreliable...hell, a couple of hours ago it reported 18 degrees idle for a few minutes lol.

Or perhaps you just had a particularly un-flat IHS or due to a combination of other factors..either way, good on you for going to the trouble of sharing your results and info

[javascripterror]

phextwin,
Now try the same test with some engineers blue on a glass plate.

[petfoodonly]

Originally Posted by javascripterror

petfoodonly,
I do not want to revisit the previous comments about Intel and AMD Cpu's apart from showing that Intel actually have a raised portion of the heat sink base supplied with their CPU's stock coolers that makes contact with inside of the IHS at its centre.

*snip*

but AMD doesnt...

[javascripterror]

petfoodonly
That is correct....... but it does not mean that the IHS is flat as FearTec has illustrated with his own CPU. As you have already indicated both companies have problems with CPU heat. Non flatness in IHS may affect only a small number but nevertheless it does occur.

[petfoodonly]

lol, I don't think you're taking my point very well.

I'm making the following assumptions:

1) People like fertec and yourself are correct. That is to say when you hold an CPU with an attached IHS in your hand, it it concave. 

2) Point 1 is a rule that may have exceptions, not the exception to a rule.

3) AMD and INTEL are both sufficently advanced that they could make a bit of metal flat if they chose to. 

4) Both AMD and INTEL realise that 'cool' procesors are 'better' than 'hot' ones.

5) AMD and INTEL would not have to reason to purposly subotage their own CPUs in such a way that they do run hotter than they could.

It follows that:

 Despite the concave nature of the IHS when held in your hand, when under typical conditions (ie heat and pressure) the IHS does not form a significant barrier to heat.

As a Relation of Ideas I think that is quite sound. Of course results say that that relation of ideas is not a matter of fact. What I want to know is where in the Relation of Ideas above have I gone wrong.

I hope that is clear.

[MR CHILLED]

10C reduction after lapping??? And moreso it is only lapping of an IHS. I'm *very* skeptical of the results.

[FearTec]

I will over clock more and update the results.

I do not care if people do not believe me, I did not risk killing my $430 processor to gain a few friends on a forum.

I stand by my results.

[javascripterror]

petfoodonly,
I have no idea what you are getting at with your assumptions so I will leave you to resolve your logic. You may want to follow the suggestion given to phextwin and test your own CPU with some engineers blue and a glass plate then come back with the visual results.
MR CHILLED,
You have an option like anyone else to review FearTec's data from WebTemp to see where the results come from. The connection to those results is after all in his sig.

[MR CHILLED]

Originally Posted by javascripterror

MR CHILLED,
You have an option like anyone else to review FearTec's data from WebTemp to see where the results come from. The connection to those results is after all in his sig.

It makes no difference whether it is in his sig or in a website, or on the license plate of his car.

I don't disbeleive his results and I don't disbeleive him, I just question the results after seeing lapping results over the years, and I'm talking direct cpu lapping, not IHS lapping which would presumably yeild less of a result than cpu lapping.

Again, I don't disbeleive his individual results.

[petfoodonly]

Originally Posted by javascripterror

petfoodonly,
I have no idea what you are getting at with your assumptions so I will leave you to resolve your logic. You may want to follow the suggestion given to phextwin and test your own CPU with some engineers blue and a glass plate then come back with the visual results.

It's not that complicated, I'm just looking for an answer "WHY would this be the case" not "IS this really the case".

as per my sig I don't have an IHS on my overclocking CPU.

Originally Posted by MR CHILLED

I don't disbeleive his results and I don't disbeleive him, I just question the results after seeing lapping results over the years, and I'm talking direct cpu lapping, not IHS lapping which would presumably yeild less of a result than cpu lapping.

Again, I don't disbeleive his individual results.

Well it does stand to reason that lapping an allready flat surface (be it a CPU die or the base of an HSF) would only yield a small result. You are merely improving contact.

It appears in this case that the cooling block was making no direct contact with the CPU die. Heat would have to travel from the die, into the center of the IHS, then to the edge of the IHS, then into the cooling block. In this case the lapping is making contact, not just improving it.

[SteakTheMooCow]

Depends on locality? Secondly I've seen some PRETTY DAMN concave CPUs in my time... =\ I'm thinking about lapping an old 450mhz P3, but the board isn't OC capable, so there'd be no point.

[javascripterror]

petfoodonly,
I really had no intention of sounding dismissive in my last comment to you. The answer to your question of "why" is contained in the philosophy that;

These parts (IHS) are mass produced just as the CPU cores are.
That there are going to be differences within the mass production.
That when a CPU is manufactured it must sit within a tolerance range or it is dicarded.
Providing the IHS acts to keep the CPU within a coolant "range" when tested it really does not matter if the IHS is perfectly flat or not. In Intels case perhaps when it became obvious that there were heat issues with the Prescott family that Intel changed the HSF to accomodate the problem and added a raised portion on the centre face of the HSF. This raised disc acting to apply localised pressure which forced down the IHS onto the core and increased the coolant affect. This would remove the need for a flat IHS altogether would it not?

In AMD's case the issues of IHS integrity and heat did not present as many complaints as Intel because of the different design technologies within AMD CPU's. Maybe as a result they felt there was no need adjust IHS flatness or to adopt more aggressive cooling measures. Add to this the development of thermal throttling "cool and quiet" technologies and reality that Mr Joe Average out there would be oblivious to his CPU or for that matter any internal computer temps. The bottom line is that only a small, comparativley speaking, group of performance geeks would be affected.

Please do not take this as conspiracy theory it is not. It is simply an attempt to explain.

[petfoodonly]

Thanks JSE, that sounds pretty reasonable. Does mass production really produce error like that? I don't know.

I suppose this goes to my first and seccond assumptions. Is a concave IHS the exception, or is it the general norm?

I guess the only way to find out is to lap more CPUs.