Intel Core i7 8700K Reportedly Reaches 4.8 GHz Easily, 5 GHz+ Requires Delid

[INSTG8R]

Well, with baking, you don't really have anything to lose. The component died already, so if you make it work, great, if not, well, it died anyway in the first place. So, it's always worth a try.

I’m pretty sure my OG Release PS3 Phat went in the oven 6 times and came back before I gave up.

 

[Basard]

To your point about thermal paste drying out, it won't dry out in a sealed environment like a heatspreader, or at least not for decades. Where is the moisture going to go?

Heatspreaders are not sealed environments. On every one there is either a hole drilled in the top or a gap in the bead of glue they use to hold it in place.

You get an "A" for effort, but your whole argument is flawed.

 

[oxidized]

The cracks aren't formed by shock but by wicking of the Indium along the Gold bonding regions. I was pointing out that the cycle has to happen many times for this to become problematic as TIM defenders are making it out to be. Also melting temps do not need to be reached only enough energy need to be provided for the two metals to react causing the atoms to shift in the lattice. When enough atoms shift or are displaced cracks form.

I have no idea what you're talking about but i'll take it , i thought thermal shock made it happen, something like what happens with Pb-Sn solder if it cools down too quickly.

 

[RejZoR]

Things crack at extreme temperature shifts. Something CPU's never really experience. I mean, temperatures we read on Windows desktop are on-die sensor readings. 70°C inside core is a lot lower at the IHS point. It also never goes below ambient temperature which is usually what, 22-25°C, but CPU never ever reaches just that. My 5820K sits at around 35-38°C idle. So, that's roughly 30°C difference (even if it's 40 it's not much).

 

[john_]

They had this in the comment section of wccftech


Yes, I know, wccftech, but it does look like we have seen these reports before. I wonder how much better really are the new 14nm++ and Coffee Lake, compared to the original 14nm and Skylake.

 

[cadaveca]

It didn't , if you search around for 7xxx series cards all the way up to the 200 series you're going to find a lot of defective cards which most likely died this way. There was quite the fuss over this back when this was uncovered.

This problem was a result of RoHS coming into effect, and switching to lead-free solder. Took some time before they found the perfect blend, and this problem was present in nearly every type of electronic you can think of for a few years.

People constantly say that solder degrades. But has anyone actually EVER seen one that has failed or started overheating because of this? Or is this just fluff myth that exists in theory but not in practice? Where with TIM, we know that it's in fact inferior from a brand new processor purchased on day one...

Yes, I have three or four chips that overheat on particular cores after some time to the point of instability, and those cores were fine before. There are also many reports of chips that begin to overheat after a time all over these forums. Not so common these days, but at the same time, people are changing rigs quite often, so it is hard to tell, but for sure reported. Nothing sucks more than having overheating Cores, and no real way to change the TIM.


Oh look, what's this CPU I just got? Oh wait, it's not a CPU, it's two! And an M.2 device? WTF... So... I don't need to speculate.

 

[RejZoR]

But is it really a thing of an IHS solder or something else? Like a fab defect or something within core rather than IHS contacting point?

I mean, even if there is a fracture in there, the CPU heatsink pressure would probably keep it together, making difference unmeasurable. Especially on monolithic designs with Intel. I find it unlikely that a micro crack would affect just 1 core somehow. For that to happen it would have to be a huge crack which just doesn't seem to make sense.

 

[cadaveca]

But is it really a thing of an IHS solder or something else? Like a fab defect or something within core rather than IHS contacting point?

I mean, even if there is a fracture in there, the CPU heatsink pressure would probably keep it together, making difference unmeasurable. Especially on monolithic designs with Intel. I find it unlikely that a micro crack would affect just 1 core somehow. For that to happen it would have to be a huge crack which just doesn't seem to make sense.

100% spot on. So really, it's not just a micro-fissure. The solder actually wicks itself out of place, creating an air gap. What really causes this, other than "thermal cycling causing solder pump", is totally unknown. You shouldn't be able to "reflow" a CPU...

You do also need to keep in mind that these guys are likely using some weird blend of metals that is highly un-tested except in-house. There's probably some dude whose job it is requires him or her to play with this stuff day in and day out, until he retires. I'd love to talk to this guy.

 

[Vayra86]

4 pages about TIM vs solder going nowhere substantial, that to me is a sign that we are discussing a non issue...

Oh well

 

[cadaveca]

4 pages about TIM vs solder going nowhere substantial, that to me is a sign that we are discussing a non issue...

Oh well

I think you missed the news item. It's about exactly that.

 

[RejZoR]

Even with the fracturing risk, I'd take soldered IHS any time over TIM. I think all the CPU's I've had were soldered. And to date I never had issues. And I like to push them to the limit.

 

[Steevo]

100% spot on. So really, it's not just a micro-fissure. The solder actually wicks itself out of place, creating an air gap. What really causes this, other than "thermal cycling causing solder pump", is totally unknown. You shouldn't be able to "reflow" a CPU...

You do also need to keep in mind that these guys are likely using some weird blend of metals that is highly un-tested except in-house. There's probably some dude whose job it is requires him or her to play with this stuff day in and day out, until he retires. I'd love to talk to this guy.

It is known, the electron energy of a mixture of metals allows two metals with a higher melting point to mix and creates a lower melting point. The solder gains energy enough to slowly reform small crystal lattices of its compounds and those differing crystal lattices are what cause the micro cracks or fissures. On a tiny BGA pad this is terrible, and will eventually result in death of all BGA devices unless the crystal lattice forms in such a way that it still allows low impedance conductivity. Note that thermal conductivity is NOT the same as electrical conductivity, as even though both metals may be thermally conductive still halogens still used in the substrate manufacturing may react with the solder metals and form a metal halide that is NOT conductive to electricity but is to heat. http://www.indium.com/blog/halogen-free-vs-halide-free.php https://en.wikipedia.org/wiki/Halide Dave, your cores may become unstable due to voltage drop to specific cores, increase in the resistance of the solder joint causing communication issues, and would also raise the resistance of the thermal probe circuit which would look exactly like a higher temperature , or very remotely possibly that the indium has migrated away from a core if they coated too much of the IHS inside with gold, but the overwhelming evidence anymore is the BGA solder will fail long before solder used as TIM.

 

[RejZoR]

Why don't they use liquid metals then? Even if their production costs are bumped up by $5 per CPU because of it and they charge extra $20 bucks for that to the end user, I'm willing to accept that. When you're paying 500+ bucks for CPU, extra 20 is nothing. This becomes even more true when we step up to 1000 and 1500 bucks CPU's. Adding extra 20 bucks on top of that is nothing if this ensures near perfection.

If TIM is crap and solder is bound to crack, liquid metal is best of both worlds. It can't crack and it has superior heat transmission. Cheaper CPU's can use TIM as it doesn't really matter for their heat output, but with 6+ core monsters, I want best of the best.

 

[EarthDog]

But the tim works....

I dont get this....sure it can be better, but it does just fine at st9ck and overclocked...which probably 1% of people do.

 

[RejZoR]

The "that'll do" approach only benefits product makers, not consumers. By not criticizing it, you're basically encouraging it. Today it's TIM instead of solder, tommor they'll charge you extra to even get the TIM...

 

[Vayra86]

I think you missed the news item. It's about exactly that.

Indeed it is, at the same time, soldered CPUs also cap out at similar clocks, and if you want to go further, you'll need exotic solutions just as well. Then there is a longevity concern, which also is not substantial, so far, I have yet to see credible numbers of Intel TIM CPUs that degrade faster than soldered ones. Reality dictates however, that a CPU is obsolete long before that happens.

The only trend we are really seeing is that Intel is eating up our OC headroom for us because the Core arch is nearing its end of life and a competitor presents a good product that can match its performance. At the same time, it cannot match its clocks by a long shot, while being soldered.

To me that is 'nuff said' with regards to solder vs TIM. What remains is a discussion for the 1% that likes to seek out limits and I am not surprised they find them...

Lol @EarthDog ... Great minds think alike or... I had not even read your response yet

 

[InVasMani]

Even with the fracturing risk, I'd take soldered IHS any time over TIM. I think all the CPU's I've had were soldered. And to date I never had issues. And I like to push them to the limit.

The original 771/775 Core 2 Quads/Duo's still run like champs and were soldered if that's not longevity idk what is and those CPU's were plenty abused in terms of heat and voltage. Honestly Intel claims a lot of things that are misleading to cover their own bottom line less warranty issues to resolve and higher profit margins due excessive overclocking. Unless they are profiteering of OC they don't care about it or appreciate it particularly from a business standpoint.

 

[trparky]

The only trend we are really seeing is that Intel is eating up our OC headroom for us because the Core arch is nearing its end of life

That's my thinking as well. We've reached the end of what the Core Architecture can do. The fact that it's gone as long as it has (8 years since Nehalem) is amazing. Now it's time to go back to the drawing board and develop the next architecture.

 

[EarthDog]

The "that'll do" approach only benefits product makers, not consumers. By not criticizing it, you're basically encouraging it. Today it's TIM instead of solder, tommor they'll charge you extra to even get the TIM...

Why would i criticize something that works? Again, im overclocked to 4.5ghz daily and could do 4.8.... should i cry because i cant get to 4.9 because intel chose tim instead of solder?? If i am one of the 1% of the 1% who wants that extra 100mhz, ill delid. Simple.

Few people, even here give a hoot about that last 100 mhz.

 

[RejZoR]

I don't want to delid my CPU. I had Athlon XP without any IHS and with re-mounting, I know the edges of the core were starting to chip off a bit.

 

[EarthDog]

So put it back on...

Dont split hairs and act stupid. Clearly we are talking swapping out the tim and placing the IHS back on...

 

[cadaveca]

To me that is 'nuff said' with regards to solder vs TIM. What remains is a discussion for the 1% that likes to seek out limits and I am not surprised they find them...

For those that like to seek out the limits, paste is better because you can then remove the metal in the way and go direct-die. IHS just gets in the way, so which is the truly the best enthusiast option?

BTW, that's what it is, paste vs solder, since both are a TIM.

 

[Kyuuba]

TIM solution works just fine, or hasn't it?
Move on as i have and many millions...
You can hate, yell, wish death to Intel but they are geniusess and you... know that, don't try to find the reason behind, you'll never find it, but be sure of this, IT WORKS!

 

[hojnikb]

The issue with the Xbox 360 was the solder used in attaching the chip itself to the motherboard , not the solder used for the heatsink. They didn't even use solder for that.

Do some research and cut back the Intel damage control.

actually, neither of that is correct. The issue was defective chip substrate, where bumps inside simply disconnected. Heating the chip to 130c or so "fixed" the issue temporary. People thought it was solder, because solder melts too.

 

[Agony]

Can anyone prove , that TIM is better or Worst than Soldered for long time runs ? Or its just Intel Vs AMD again ???