22°C diff by changing thermal paste? Tom's.

[Syntaxvgm]

15 hours ago, MonkeyBrainz said:

Intel uses Dow Corning, because unlike these other pastes, it can last decades and only marginally worsen. That's incredibly important for commercial applications, because processors in dummy terminals and office PCs often go unchanged for a decade. My university was on the original core 2 duos in its computer labs before they upgraded to 4790s in my senior year.

 

Mind you, if Intel wanted to go full-bore and make people buy more often, it could use a better performing paste that gives out after 1-2 years. The other issue though is the damn glue they use sometimes produces more spacing between the die and IHS than is optimal, and when you delid, you get rid of that glue, so it's not just the thermal paste at issue.

people miss this point all the time. 

Because of this I really believe that enthusiast cpus shouldn't come with an IHS or come with an easily removable one. None of that solder bullshit (Just enthusiast cpus. )

 

[Syntaxvgm]

Just now, MonkeyBrainz said:

Eh, I'm not so keen on naked mounts if it's all the same.

I would do it every time if delidding was easier and safer. 

[Syntaxvgm]

17 minutes ago, MonkeyBrainz said:

You have no lack of confidence

I'm just not risking a $1000+ CPU to being cracked by a naked mount.

well yah I don't spend 1000$+ on a cpu, I mean I have a 5820k. I'm not sure I'd like to do it to a xeon or something, but as long as you're careful can't really mess up too bad. It's delidding that scares me. 

[JoostinOnline]

16 hours ago, tridy said:

Some time I watched a video from LTT last year: Delidding an Intel Core i7 6700K - Is it worth it!? and it seemed to make sense, or, at least, something one could expect.

That video was nothing more than anti-delidding propaganda.  He surrounded it with tons of warnings, used cheap thermal paste, used a Skylake processor (which doesn't benefit quite as much as Haswell), and just for good measure, he applied the paste wrong.

[Syntaxvgm]

1 minute ago, JoostinOnline said:

That video was nothing more than anti-delidding propaganda.  He surrounded it with tons of warnings, used cheap thermal paste, used a Skylake processor (which doesn't benefit quite as much as Haswell), and just for good measure, he applied the paste wrong.

I wouldn't call it anti-delidding propaganda it's more rushed LTT production without proper research, attempting to err on the side of caution to not encourage viewers to try stuff based on their videos that are overviews at best. 

[JoostinOnline]

2 minutes ago, Syntaxvgm said:

I wouldn't call it anti-delidding propaganda it's more rushed LTT production without proper research, attempting to err on the side of caution to not encourage viewers to try stuff based on their videos that are overviews at best. 

Well he wouldn't admit he did anything wrong.  It was obvious he didn't WANT it to be successful.

[Syntaxvgm]

5 minutes ago, JoostinOnline said:

Well he wouldn't admit he did anything wrong.  It was obvious he didn't WANT it to be successful.

typical ltt lol 

[Bananasplit_00]

its pretty standart temp drop for a delid tbh

[Drak3]

31 minutes ago, JoostinOnline said:

Well he wouldn't admit he did anything wrong.

It takes a big man to admit his mistakes.

 

And Linus is extremely vertically challenged.

And as muscular as a twig.

[JoostinOnline]

1 minute ago, Drak3 said:

It takes a big man to admit his mistakes.

 

And Linus is extremely vertically challenged.

And as muscular as a twig.

Lol, well he's admitted mistakes before.  That's what made this stand out.

[Drak3]

Just now, JoostinOnline said:

Lol, well he's admitted mistakes before.  That's what made this stand out.

99.99999% certain Luke has a Linus puppet for those instances.

[MageTank]

17 hours ago, MonkeyBrainz said:

Intel uses Dow Corning, because unlike these other pastes, it can last decades and only marginally worsen. That's incredibly important for commercial applications, because processors in dummy terminals and office PCs often go unchanged for a decade. My university was on the original core 2 duos in its computer labs before they upgraded to 4790s in my senior year.

 

Mind you, if Intel wanted to go full-bore and make people buy more often, it could use a better performing paste that gives out after 1-2 years. The other issue though is the damn glue they use sometimes produces more spacing between the die and IHS than is optimal, and when you delid, you get rid of that glue, so it's not just the thermal paste at issue.

I do not know where you came from, but in your 17 posts, you've already earned a great deal of respect from me. You reek of enthusiast, lol. Not often that I come across people that are not instantly mystified by the LTT videos that are often misinformed or poorly constructed (memory speed, delidding, etc). 

 

EDIT:

4 hours ago, zMeul said:

Gallium likes to alloy itself with other metals, Copper is no stranger

Nickel plating however is neutral - if everyone would nickel plate the IHS and heatsinks all would be much better for Gallium based TIMs

but until then ... tread with care; Copper Gallium alloy does happen and it's a solid crystalline alloy much like Gallium's natural form

 

 

soldering on small surface die is not going to happen - is explained in previous posts why

You've clearly never tried to spread CLU over nickel plating, lol. It's an absolute pain. I had to lap the underside of my IHS to the copper in order to get the gallium to adhere properly. 

[DildorTheDecent]

Try it yourself OP. You'll wonder why you didn't do it in the first place. 

 

20C off by doing an easy mod works wonders for clock gains. 

 

That LTT video on it is awful though. I seriously wonder what they were thinking. 

[Sauron]

17 hours ago, tridy said:

The further I go, the better Ryzen soldered solution looks. 

Sandy bridge had the lid soldered to the die, then intel moved away from that solution because they were having problems with soldering the smaller dies without having them break with thermal contractions.

[Sakkura]

14 hours ago, MageTank said:

EDIT:

You've clearly never tried to spread CLU over nickel plating, lol. It's an absolute pain. I had to lap the underside of my IHS to the copper in order to get the gallium to adhere properly. 

I do think ease of application is secondary to whether the gallium literally destroys the heatsink you're using.

 

I mean, this is what pure gallium does to aluminium:

 

 

[tridy]

I don't think that demos like that are fair to make any sort of statement. It is similar to Cola and Mentos ones. You know they don't get friendly together and that is why you don't put them together in real life (unless you want to demonstrate that specifically) like, "look what happens if you don't read the manual."

 

Going back to, lets say 15 degrees difference, when deliding and applying liquid metal, it does look like a reasonable hack, if that indeed gives you this difference in temperatures. The examination of time, and if it needs (I have no idea if it needs) to be painfully cleaned and re-applied every 9 months could be a show stopper. But if it would be continuously giving 15 degrees difference under 36 months or so, I might give it a thought. 

[Drak3]

14 hours ago, NibRiddan said:

The Tim on the 7700k is aweful, all of the reports from people delidding their 7700ks have reported this change of 20c.

The issue isn't the TIM, it's actually rather good for the intended purpose for most users. It lasts a significantly longer time than most, if not all, higher performance TIMs whilst maintaining temps that are acceptable for mild overclocks. It isn't until you get to the more extreme overclockers that Dow Corning TIM becomes unacceptable.

 

The issue with the 7700K, as the issue was with the 4770K, is that Intel's gluing procedure isn't as good as it should be, leaving uneven spaces or too much space between die and IHS, and subsequently not spreading the TIM correctly.

[MageTank]

6 hours ago, Sakkura said:

I do think ease of application is secondary to whether the gallium literally destroys the heatsink you're using.

 

I mean, this is what pure gallium does to aluminium:

 

 

Gallium doesn't destroy copper though, which is what your IHS is made of. Let's not change the context of my words, lol. Besides. Anyone that uses gallium on top of their IHS are simply doing it wrong. It makes a huge difference on the bare die, but on top, it's not much better at all compared to traditional, less expensive silicone based pastes. 

[tridy]

1 hour ago, Drak3 said:

The issue with the 7700K, as the issue was with the 4770K, is that Intel's gluing procedure isn't as good as it should be, leaving uneven spaces or too much space between die and IHS, and subsequently not spreading the TIM correctly.

I thought they are doing their homework and thermo testing their chips prior to the shipping. Or do you mean that it shows effects a bit later after N months or so, or due to the shaking during the shipping?

[Sakkura]

24 minutes ago, MageTank said:

Gallium doesn't destroy copper though, which is what your IHS is made of. Let's not change the context of my words, lol. Besides. Anyone that uses gallium on top of their IHS are simply doing it wrong. It makes a huge difference on the bare die, but on top, it's not much better at all compared to traditional, less expensive silicone based pastes. 

Well, the post you were replying to was warning about interaction between copper and gallium.

 

Though I'd be much more worried if you were dealing with aluminium.

[MageTank]

5 minutes ago, Sakkura said:

Well, the post you were replying to was warning about interaction between copper and gallium.

 

Though I'd be much more worried if you were dealing with aluminium.

I don't know what @zMeul is on about, as I've used gallium on copper for a long time now, and have not had any issues. It bonds far better to copper than it does to nickel, making it easier to spread. I do find that circular motions are best when using CLU, as trying to spread it back and forth is an endless nightmare, no matter what you are trying to spread it on.

 

But yeah... Nobody should be using CLU or any other liquid metal paste on top of the IHS. It's simply not worth it from a price:performance point of view. The die is extremely small, and that's where the bulk of your heat transfer happens. Once it spreads out on the IHS, your traditional pastes will do just as good a job, at a fraction of the cost. I say for that, Gelid or Thermal Grizzly Kryonaut are your best bets. 

[Lays]

On 3/25/2017 at 3:43 AM, Sauron said:

That's not normal thermal paste, it's "liquid metal" and can perform significantly better than your average paste.

Well, Linus did an ok job of comparing the stock paste to something you'd actually use.

No he really didn't, the paste he used in the video doesn't hold up to thermal cycles very well.  Thermal cycling happens a lot on the die much more-so than on the IHS, because the die can rapidly heat and cool simultaneously.  It's been known by quite a bit of enthusiasts that have delidded multiple cpu's that Linus's delid video is not realistic at all.

[MageTank]

Just now, Lays said:

No he really didn't, the paste he used in the video doesn't hold up to thermal cycles very well.  Thermal cycling happens a lot on the die much more-so than on the IHS, because the die can rapidly heat and cool simultaneously.  It's been known by quite a bit of enthusiasts that have delidded multiple cpu's that Linus's delid video is not realistic at all.

Yeah, and lets not forget he didn't spread it at all. Expecting the IHS to spread it with mounting pressure is pretty silly once you factor in that there won't (and shouldn't) be enough mounting force to spread it on the die with just the IHS alone with the tiny bit he used. 

 

When it comes to heavy heatsinks spreading paste on top of the IHS, by all means, go nuts. However, your die is awfully sensitive to pressure, and should not be forced to try to spread paste thinly between the IHS and die. Something like that is better suited to manual spreading for the best possible results. Either way, that video is awful, and is getting in the way of real facts. 

[Drak3]

1 hour ago, tridy said:

I thought they are doing their homework and thermo testing their chips prior to the shipping. Or do you mean that it shows effects a bit later after N months or so, or due to the shaking during the shipping?

Intel's standards aren't quite as high or refined as delidders. If, at stock, their glue job and TIM can keep a 7700K from throttling with popular air coolers (hence the move to not ship unlocked SKUs with coolers at all anymore), then that's that and product ships.

As for the Dow Corning TIM Intel (and AMD) use for smaller die CPUs, it takes quite awhile for it to break down. Longer than the realistic, usable lifetime of most chips, significantly longer. It's not horrible either, provided that the IHS is attached properly.

[Jahramika]

On 3/25/2017 at 2:56 AM, iamdarkyoshi said:

I got a 20 degree drop as well when I delidded.

 

Lol 20 degrees and no warrenty Intel the way it should be.