Not at Amd they do it free hand because that's how they roll.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Intel Preps Core i7 9700k 8 core 16 Thread Mainstream CPU
- Thread starter yoyo59
- Start date
Soldato
- Joined
- 5 Sep 2011
- Posts
- 12,876
- Location
- Surrey
Well, there isn't anything 'better' than indium as a thermal conductor that works with the type of heat cycling CPU are subjected to. If there were, they would have used it by now. AMD are using indium for that reason, but it is the one and only reason to use it, outweighed by the many not to. Maybe some evidence suggested that due to the increased TPD and current capability of the most recent SKUs, solder was just as prone to micro fracture under enough thermal cycles. Although the server SKUs haven't yet switched, I've no doubt they will. They're not going to be subject to the same levels of current draw as consumer CPUs that are unlocked.
It's easy to sit here and poo poo them for not using it, but there really are a lot of valid reasons not to use indium. There are other compounds needed, too. It's not just indium and gold. You can't just paste it on like dairylea and go home for the day, as it would greatly increase the rate of damage to the silicon overtime. Other metal elements are needed to adhere the solder to the die.
AMD might well be using it now, but that doesn't mean they will be even 2 years from now, or sooner.
Last edited:
Again it's plausible but no more than it being due to financial reasons. We can only really say that when Amd stop using solder, they haven't hinted that they'll be switching anytime soon.
Perhaps Intel will be better off doing what Jigger said, that being dropping the K line altogether. I mean it's not like you gain a massive amount in performance from turbo boost to 5ghz+.
In the meantime Intel are continuing to take a bit of a battering from the enthusiast community over the use of Tim as opposed to solder.
I like sodium's properties. Have they tried to use it?
Indium
Thermal conductivity 81.8 W/(m·K)
Thermal expansion 32.1 µm/(m·K) (at 25 °C)
Melting point 429.7485 K (156.5985 °C, 313.8773 °F)
https://en.wikipedia.org/wiki/Indium
Sodium
Thermal conductivity 142 W/(m·K)
Thermal expansion 71 µm/(m·K) (at 25 °C)
Melting point 370.944 K (97.794 °C, 208.029 °F)
https://en.wikipedia.org/wiki/Sodium
Lithium
Thermal conductivity 84.8 W/(m·K)
Thermal expansion 46 µm/(m·K) (at 25 °C)
Melting point 453.65 K (180.50 °C, 356.90 °F)
https://en.wikipedia.org/wiki/Lithium
Sodium has a melting point of 97degrees so that rules out the 7700k, 8700k and all the Skylake x cpu's.
Soldato
- Joined
- 5 Sep 2011
- Posts
- 12,876
- Location
- Surrey
LOL.
I'll er, leave that for someone else.
Yeah, well as I've already said - I'd love for them to go back to solder. The rest is what it is. I very much doubt you'll get much in the way of an official reply, either.
I'll er, leave that for someone else.
Yeah, well as I've already said - I'd love for them to go back to solder. The rest is what it is. I very much doubt you'll get much in the way of an official reply, either.
No I wouldn't have thought so either just the usual scripted, political style standard reply. Anyway I'm not quite sure how we got so far into the whole solder debate anyway, it only really matters if you have brought one.
Intel's materials science department wrote the book on soldering silicon dies to copper heat spreaders, but in this thread: "I know better, I googled this element with higher thermal conductivity than indium and surely you can solder it to the silicon die without destroying it!"
And for those wondering why I'm saying the Corning polymer TIM is really good, it's because the temperature difference between stock and delid at normal frequencies & voltages is <10C, GamersNexus tried 4.7Ghz with 1.35v:
https://www.gamersnexus.net/hwrevie...vs-ryzen-streaming-gaming-overclocking/page-2
And mind you, you can get up to 10C of temperature difference if you delid and desolder the older Broadwell-E/Haswell-E dies, as shown by der8auer here:
https://www.youtube.com/watch?v=Hf8V_UulpBk
The temperature difference there is explained by the liquid metal TIM layer being thinner than the indium solder sheet used for those dies, thus better thermal conductivity. Same idea applies for current polymer TIM CPUs, the main issue is the polymer TIM layer thickness and if they improve their packaging they might actually be able to reach similar performance to soldered chips.
As to why they don't solder, again, it's fairly easy to explain, but of course that doesn't make it right either since they could just as well solder the K or X chips if they wanted to. But for the sake of the argument:
1) the enthusiast market is minuscule and the polymer TIM they use is good enough if the chips are used within spec;
2) they did release papers which showed cracks appearing in the indium solder layer with intensive thermal cycling, cracks which could damage the die and reduce the lifetime of the CPU, which explains why they didn't solder their mainstream chips since they were all small dies since the jump to 22nm, but it doesn't explain why they aren't soldering the large Skylake-X dies;
3) for the big Skylake-X dies the environmental argument makes sense, and that covers the small dies too. I believe they probably get some grants or tax incentives so it's not 'out of the kindness of their hearts', no corporation does that, plus it might be cheaper to use polymer TIM. Their yields might also improve by using polymer TIM because soldering die/indium/IHS together requires 170C and a few dies probably do get damaged by the process.
The problem right now is that there are no viable alternatives. AMD is running with this soldered marketing line for all it's worth, but when overclocking their chips you always run into the voltage wall faster than you do into their thermal one. Maybe that will change with Zen+ or Zen2.
Either way this entire chain has been pretty silly so far since we're talking about an unreleased chip and we don't know how good the polymer TIM application in it is going to be. Do we even have any palpable info on the supposed 9700K?
And for those wondering why I'm saying the Corning polymer TIM is really good, it's because the temperature difference between stock and delid at normal frequencies & voltages is <10C, GamersNexus tried 4.7Ghz with 1.35v:
https://www.gamersnexus.net/hwrevie...vs-ryzen-streaming-gaming-overclocking/page-2
And mind you, you can get up to 10C of temperature difference if you delid and desolder the older Broadwell-E/Haswell-E dies, as shown by der8auer here:
https://www.youtube.com/watch?v=Hf8V_UulpBk
The temperature difference there is explained by the liquid metal TIM layer being thinner than the indium solder sheet used for those dies, thus better thermal conductivity. Same idea applies for current polymer TIM CPUs, the main issue is the polymer TIM layer thickness and if they improve their packaging they might actually be able to reach similar performance to soldered chips.
As to why they don't solder, again, it's fairly easy to explain, but of course that doesn't make it right either since they could just as well solder the K or X chips if they wanted to. But for the sake of the argument:
1) the enthusiast market is minuscule and the polymer TIM they use is good enough if the chips are used within spec;
2) they did release papers which showed cracks appearing in the indium solder layer with intensive thermal cycling, cracks which could damage the die and reduce the lifetime of the CPU, which explains why they didn't solder their mainstream chips since they were all small dies since the jump to 22nm, but it doesn't explain why they aren't soldering the large Skylake-X dies;
3) for the big Skylake-X dies the environmental argument makes sense, and that covers the small dies too. I believe they probably get some grants or tax incentives so it's not 'out of the kindness of their hearts', no corporation does that, plus it might be cheaper to use polymer TIM. Their yields might also improve by using polymer TIM because soldering die/indium/IHS together requires 170C and a few dies probably do get damaged by the process.
The problem right now is that there are no viable alternatives. AMD is running with this soldered marketing line for all it's worth, but when overclocking their chips you always run into the voltage wall faster than you do into their thermal one. Maybe that will change with Zen+ or Zen2.
Either way this entire chain has been pretty silly so far since we're talking about an unreleased chip and we don't know how good the polymer TIM application in it is going to be. Do we even have any palpable info on the supposed 9700K?
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
Not at Amd they do it free hand because that's how they roll.
It sounds like AMD's got a new Bumper sticker
I do it Freehand because that's how I roll
Yeah it would have made a great meme if Raja was still there.
We are not talking about normal operating temperatures, we are talking about the point at which temperature does become an issue, where the difference can be as high as 15-20 degrees, which is a lot.
Anyway as usual were going around in circles quite a bit with the pro Amd , Pro Intel dynamic. Rather boring one might say.
Associate
- Joined
- 27 Apr 2007
- Posts
- 968
'I like sodium's properties. Have they tried to use it?'
It should work particularly well with water cooling.
It should work particularly well with water cooling.
Soldato
- Joined
- 5 Sep 2011
- Posts
- 12,876
- Location
- Surrey
Everyone will find a better solution than TIM
Multi Milillion Dollar Epiphany #3
'I like sodium's properties. Have they tried to use it?'
It should work particularly well with water cooling.
As would Lithium.
Sodium has a melting point of 97degrees so that rules out the 7700k, 8700k and all the Skylake x cpu's.
It will not when there are preventive measures applied, proper sensors for throttling and other limits, so the silicon never gets to 98°C.
Have you noticed that sodium and silicon have almost identical thermal conductivity coefficient?
142 and 149 W/m.K.Kind of defeating the point though really when what they have now already means they have to throttle to keep temps in check.It will not when there are preventive measures applied, proper sensors for throttling and other limits, so the silicon never gets to 98°C.
Have you noticed that sodium and silicon have almost identical thermal conductivity coefficient?
I'm sure if it was viable they would have already explored it.
I wouldn't be so sure. They are mostly very conservative and use very old but examined methods.
Do you honeslty believe they explored the whole periodic table plus put some effors for some non-invented chemicals, to finally come on the chewing gum/toothpaste/crap TIM?
They need to remove that bottleneck for the heat to be transfered properly from the silicon to the outer surroundings.
Don't know really Amd uses Indium so it's obviously a well proven method. I'm sure their are company's out there who do explore that sort of thing. Where their is a dollar there is a way and all that.
97 degrees is far to close for comfort though. Last thing you want is molten solder dripping down the side of your motherboard.
I'd like to know if anyone has used a liquid metal Tim for an extended period of time and whether that is a viable option for longevity.
97 degrees is far to close for comfort though. Last thing you want is molten solder dripping down the side of your motherboard.
I'd like to know if anyone has used a liquid metal Tim for an extended period of time and whether that is a viable option for longevity.
They already need layers of titanium, nickel, vanadium and gold to be able to solder indium to silicon since if they just put indium naked on top it would diffuse into the silicon die and destroy it.
Sodium would have the thermal expansion issue, which would probably destroy the die. Indium is apparently already difficult enough to solder, it can deform the package.
Pretty sure that after the billions Intels has invested in materials science R&D, indium is most likely the best solution we've got.
Sodium would have the thermal expansion issue, which would probably destroy the die. Indium is apparently already difficult enough to solder, it can deform the package.
Pretty sure that after the billions Intels has invested in materials science R&D, indium is most likely the best solution we've got.