Noctua NH-D14 - Assumption is a waste of time to lap heatsink...

[crashtech]

Well yes, though the actual TIM itself (contact problems aside) is no problem if you're using IX or CLU or something like that.

The interface with TIM is the most insulating part of the setup in any case aside from solder*, so eliminating an interface may not be so trivial in extreme cases.

*Actually, copper has about 5 times the thermal conductivity of indium, so even WITH solder, the interface is the most insulating part.

 

[DrMrLordX]

Didn't know AMD was pulling this too, geeze.

Yup! Every APU from Llano to Kaveri has had TIM under the IHS.

The interface with TIM is the most insulating part of the setup in any case aside from solder*, so eliminating an interface may not be so trivial in extreme cases.

*Actually, copper has about 5 times the thermal conductivity of indium, so even WITH solder, the interface is the most insulating part.

Right, but remember that k value can be mitigated by bond thickness, which in and of itself is governed by the quality of the contact between the bonded surfaces. If you have an excess of epoxy holding the IHS onto the PCB (or worse, uneven application of the epoxy), then delid/relid is going to reduce bondline thickness big time, or at least promote uniformity of such.

Don't get me wrong, I'm not saying that having the IHS there, lapped or otherwise, is just as good as direct-die cooling. It's just that it isn't going to be as big of a deal once you use the right TIM and correct for irregularities caused by the factory epoxy application.

Ideally, Intel and AMD would sell us chips with the IHS loose in the package for those of us who want to DIY. Then HSF and board manufacturers would team up to give us improved retention mechanisms that would allow us to adjust mounting height.

 

[crashtech]

Right, but remember that k value can be mitigated by bond thickness, which in and of itself is governed by the quality of the contact between the bonded surfaces. If you have an excess of epoxy holding the IHS onto the PCB (or worse, uneven application of the epoxy), then delid/relid is going to reduce bondline thickness big time, or at least promote uniformity of such.

Don't get me wrong, I'm not saying that having the IHS there, lapped or otherwise, is just as good as direct-die cooling. It's just that it isn't going to be as big of a deal once you use the right TIM and correct for irregularities caused by the factory epoxy application.

Ideally, Intel and AMD would sell us chips with the IHS loose in the package for those of us who want to DIY. Then HSF and board manufacturers would team up to give us improved retention mechanisms that would allow us to adjust mounting height.

No argument. But I'm sure you know the IHS is not really there to spread heat but to eliminate die breakage upon mounting the HSF. The only line I could ever see Intel leaving the IHS off of would be HEDT, and since those are soldered with the same process as Xeons, it's moot.

 

[Yuriman]

In my case, even a lapped and remounted IHS is still worth 5-10c of increased temperature at the extremes (e.g. 4.8ghz, 1.40v), when compared with bare die mounting.

 

[BonzaiDuck]

It's not the nickel plate that's messing things up it's:

1) Top of IHS is not flat (most that I've measured have been concave although a few were convex/wavy)
2) Bottom of the IHS isn't making good even contact with the cpu slug. Cores 1/4 should always be coolest (they are after I rebuild them, and within about 5C of each other) because they're surrounded by darker silicon. Before working on them I've seen all kinds of different temperature deltas/huge spreads (up to 15C) between loaded core temps.

Oh, sure. No doubt about those factors. The TIM and "gap" problem would've been the greatest thermal issue overall for those processors, but less so for the Haswell Refresh. I'm only saying: if you're going to lap the processor for flatness, eliminating the nickel-plate also offers a marginal improvement. Any such improvements will be diminished in the face of the change in the Intel production process and design.

Those difficulties were at least supposed to be diminished in the later Haswells. If I had a problem cooling my 4790K, I might consider de-lidding. If not, the lapping is worth a couple C, the right TIM (between HSF and IHS) is worth a few C.

 

[DrMrLordX]

No argument. But I'm sure you know the IHS is not really there to spread heat but to eliminate die breakage upon mounting the HSF. The only line I could ever see Intel leaving the IHS off of would be HEDT, and since those are soldered with the same process as Xeons, it's moot.

I think they could do it for k chips. And yeah, most heatsink bases are thick enough that heatspreading is not really necessary (except for certain HDT coolers).

 

[JEDIYoda]

Actually, IX is the best.

Define best...

 

[DrMrLordX]

Highest k value, lowest temperatures given identical conditions between mount comparisons.

IX tests better than every other TIM on the market, period. It's expensive, it's a PITA to install, and AMD users can't use it. But it delivers.

 

[crashtech]

I think they could do it for k chips. And yeah, most heatsink bases are thick enough that heatspreading is not really necessary (except for certain HDT coolers).

It's easy to forget that the chips like the 4790K are considered mainstream processors that OEMs need to be able to ship in their high line PCs. As such they have to be idiot-proof drop-ins, not finicky variants, even though we might like it better otherwise.

 

[MongGrel]

If you mean Indigo Xtreme, I would agree, but there's the drawback of (a) first getting it to melt by heating up the CPU, and (b) removing it when you need to.

I forgot about that option. It did test out marginally better than anything else. And the reasons I forgot about it were the extra hassles.

On the delidding. There are certainly the risks. But then, you have to weigh the improved "polymer TIM" against using CLU -- which as I recall, was the consensual choice of delidders. I dunno -- ask IDontCare -- he knows.

I've seen this or that member get some decent temperatures with the 4790K -- overclocked -- without any delidding and only extra care and measures in cooling. So personally, I'd just grind off the nickel-plate and otherwise leave it alone.

I was tempted to try out IX and just never went there myself I guess.

 

[DrMrLordX]

It's easy to forget that the chips like the 4790K are considered mainstream processors that OEMs need to be able to ship in their high line PCs. As such they have to be idiot-proof drop-ins, not finicky variants, even though we might like it better otherwise.

They could just make a separate SKU for the top-end k processors (or maybe all the k parts). There are costs associated with inventory control when adding SKUs, but at the same time, having an entire run of chips where they needn't worry themselves with bonding the IHS to the PCB might help them recover some of those costs.

I agree that the needs of the enthusiast sector are not foremost in the mind of Intel management. You'd better believe that the Haifa engineers would do something like that in a heartbeat! They're the ones that allegedly want to go back to solder on desktop chips . . .

 

[BonzaiDuck]

I was tempted to try out IX and just never went there myself I guess.

Looks as though we have a few parallel conversations here; DrMrLordX is currently on a different track at the moment, but he raised attention to Indigo Xtreme.

So -- JediYoda! Been a while, Bro'!

Actually, IX is the best.

Define best. . .

At the simplest level of evaluation, only "performance" rules. Indigo Xtreme trumps CLU by a couple C degrees, which trumps IC Diamond by a degree or two in turn.

On that fundamental criterion, it's all about thermal resistance and its inverse -- thermal conductivity. No more -- no less.

But throw in the other factors: ease of application; re-use-ability; various risks; etc. Then you have a "qualified best" or multiple criteria in the score.

With the Indigo, you have to cure it, and some of the installation advice suggests using a hair-dryer to raise the processor temperature to 80C which will melt the IX "metal pad." Once done, you have the other factor: ease of removal. As I understand, it forms an amalgam with the copper surfaces, so removing the HSF from the IHS is no picnic, either.

Folks may please argue and contest these points, but if I were ever tempted to use IX, these factors -- whether assumed, exaggerated, misunderstood or simply correct -- led me to defer it and abjure it.

Then there was CLU or its predecessor from the same outfit -- Cool-Laboratory. "Little balls of liquid metal rolling around on your motherboard like spilling Mercury on the floor in a sort of pinball frenzy." CLU is apparently less likely to result in any such mishap than what came before it.

So I stuck with IC Diamond -- a thermal resistance which is "way down there" and just a tad higher than the gallium-indium-whatever formulations. It's not conductive. It won't roll around and get lost; it won't wrench your processor from the socket when you want to remove your HSF or waterblock for cleaning; and it can be literally re-used.

It's just a slight bitch to apply, and a tad more expensive. Of course, the criteria change when you alter the cooling configuration: for instance, de-lidding the IHS and either chucking the IHS altogether for a bare-die approach, or reinstalling the IHS and replacing the TIM with one thing or the other. In that latter possibility, damage to the die may occur even from micronized diamond particles. Similarly, with Indigo Xtreme, I'm only guessing that "You're really stuck." Of course, the IX wouldn't form an amalgam with the processor die. But I never saw any attempts to use it in the de-lidded arrangement.

Like they say at the nudie-bar: "Pick your poison!"

 

[MrTeal]

Looks as though we have a few parallel conversations here; DrMrLordX is currently on a different track at the moment, but he raised attention to Indigo Xtreme.

So -- JediYoda! Been a while, Bro'!



At the simplest level of evaluation, only "performance" rules. Indigo Xtreme trumps CLU by a couple C degrees, which trumps IC Diamond by a degree or two in turn.

On that fundamental criterion, it's all about thermal resistance and its inverse -- thermal conductivity. No more -- no less.

But throw in the other factors: ease of application; re-use-ability; various risks; etc. Then you have a "qualified best" or multiple criteria in the score.

With the Indigo, you have to cure it, and some of the installation advice suggests using a hair-dryer to raise the processor temperature to 80C which will melt the IX "metal pad." Once done, you have the other factor: ease of removal. As I understand, it forms an amalgam with the copper surfaces, so removing the HSF from the IHS is no picnic, either.

Folks may please argue and contest these points, but if I were ever tempted to use IX, these factors -- whether assumed, exaggerated, misunderstood or simply correct -- led me to defer it and abjure it.

Then there was CLU or its predecessor from the same outfit -- Cool-Laboratory. "Little balls of liquid metal rolling around on your motherboard like spilling Mercury on the floor in a sort of pinball frenzy." CLU is apparently less likely to result in any such mishap than what came before it.

So I stuck with IC Diamond -- a thermal resistance which is "way down there" and just a tad higher than the gallium-indium-whatever formulations. It's not conductive. It won't roll around and get lost; it won't wrench your processor from the socket when you want to remove your HSF or waterblock for cleaning; and it can be literally re-used.

It's just a slight bitch to apply, and a tad more expensive. Of course, the criteria change when you alter the cooling configuration: for instance, de-lidding the IHS and either chucking the IHS altogether for a bare-die approach, or reinstalling the IHS and replacing the TIM with one thing or the other. In that latter possibility, damage to the die may occur even from micronized diamond particles. Similarly, with Indigo Xtreme, I'm only guessing that "You're really stuck." Of course, the IX wouldn't form an amalgam with the processor die. But I never saw any attempts to use it in the de-lidded arrangement.

Like they say at the nudie-bar: "Pick your poison!"

I've used CLU quite a bit on bare die applications, and other than being a bit of a pain to paint onto the die it's actually not that bad to use. Wetting is excellent, and the performance compared to MX-4 and the like is very, very good. I've never had an issue with the excess balling up, though you do have to be careful while cleaning it off the die when you're removing it.

 

[DrMrLordX]

The best thing about CLU is that you never really have to remove it. Just add a little extra when reseating the IHS and/or HSF, and the old residue "comes back to life". You might want to get rid of some of it to level off the surfaces, but removing it entirely is just not necessary.

 

[BonzaiDuck]

Yeah -- the CLU was unlikely to "run away."

I didn't know that you could actually re-use it, so that's another plus.

If I plan on removing my HSF to give it a real good liquid and air cleaning, I may go that route -- never mind the 3 grams of micronized diamond powder I still have after my purchase that preceded IC Diamond's release.

Hey! A coupla-degrees-C . . . is a coupla-degrees-C!!

Air or water -- take your pick.