Is Intel’s Thermal Paste ACTUALLY That Bad?

[AlexTheGreatish]

By harvesting TIM from old Intel CPUs we can finally find out just how bad it is.

Buy better thermal paste
On Amazon: http://geni.us/CroMAlu
On Newegg: http://geni.us/Ale04G3

 

 

[JR88]

 

It looks baddd man like something Generic and they put too much on imo. Big old blob of it...I lightly put a coat of MX4 on and just hope got a good contact with the Lid.  7700k should do really well...telling ya though go to you're local  Lowe's or home depot and get that clear a little better Silicone. usually a bit bigger tube of the stuff...it's exactly like the stuff used around Bath Tubs/Sinks whatever etc...good stuff dried in about 4 hours and it is stuck on there well...I wouldn't use some of those Glue's some were using on it.

 

Delidded with some 20$ Tool worked real well.

 

We should start a Delidding Service and fix this mishap with it all. 

 

Can't wait to see the Result's.

[Guest The Viking]

yes it's bad. like really bad. Solder the cpus, shintel. thanks.

 

-a very unsatisfied consumer

 

edit: after watching vid, nothing new, intel's paste is worse than toothpaste. 

[scottyseng]

I kind of wonder why they didn't just buy a jar of the same corning dow thermal paste that intel uses?

[JR88]

4 minutes ago, The Viking said:

yes it's bad. like really bad. Solder the cpus, shintel. thanks.

 

-a very unsatisfied consumer

Yeah I don't know why they wanted to change that and went away from the Solder.  different too older chips bigger die and physically bigger and lid bigger more surface area...that's why they were pretty easy and acceptable to cool. Most of them ran pretty cool really.

[Guest The Viking]

1 minute ago, JR88 said:

Yeah I don't know why they wanted to change that and went away from the Solder.  different too older chips bigger die and physically bigger and lid bigger more surface area...that's why they were pretty easy and acceptable to cool. Most of them ran pretty cool really.

it's for money, obviously. You sell 1 million chips. Soldering it costs you, say, 5$. Using ok thermal paste 2.5$. Using shitty one, 1$. Which one would you pick?

 

The shitty one at 1$, obviously. Maybe on low-end chips, fine, but on high end, 350$+ cpus? intel, please.... 

[JR88]

2 minutes ago, The Viking said:

it's for money, obviously. You sell 1 million chips. Soldering it costs you, say, 5$. Using ok thermal paste 2.5$. Using shitty one, 1$. Which one would you pick?

 

The shitty one at 1$, obviously. Maybe on low-end chips, fine, but on high end, 350$+ cpus? intel, please.... 

 

Yeah kind of disappointed in them too about it. O well another mod to perform. 

[Keikavoos]

It's worse than "that"

(I will miss you, my i5 "fireline" 4690K, may your heroic fights with dat stock cooler and his thermal paste be long remembered)

[meenmeen1103]

lmao tunnelbear...

[Enderman]

You can see at 7:20 that the layer of compound is tens or hundreds of microns thicker than that of the previously applied liquid NT-H1.

This is far thicker than normal when it is applied to intel CPUs, as you can see at 2:30.

 

I'm not sure if you guys are aware of how heat transfer works, but the 'm' in w/mk is a measure of distance, so the performance of the paste is inversely proportional to the thickness of application.

If the layer of 'stuff' you put on the CPU was 100 microns thick compared to a 1 micron layer of NT-H1, you can expect a 100x thermal transfer difference.

 

You basically replaced some variables with even more variables in this test.

 

A proper way of doing this test would be to 

A) obtain the thermal compound used in intel CPUs in its LIQUID form, not after it has dried

B) fill two low-thermally-conductive tubes several cm long with each thermal compound, NT-H1 and the intel compound

C) place a heating element/resistor in one end of each of the tubes inside the thermal compound

D) place a temperature probe in all 4 ends of the two tubes

E) adjust the power to the resistors until you get the same temperature values on the resistor ends

F) measure the temperature on the opposite end

G) calculate the w/mk to see the real difference in thermal performance

 

If you can't obtain the real thermal paste in liquid form, then measuring the temperature of the CPU before being delidded, then after delidding with the new compound, is still a much more accurate way of measuring the performance difference.

The thickness of the original layer of material will be almost identical to the thickness when you replace it with the new NT-H1, since the NT-H1 is still liquid.

[79wjd]

57 minutes ago, The Viking said:

it's for money, obviously. You sell 1 million chips. Soldering it costs you, say, 5$. Using ok thermal paste 2.5$. Using shitty one, 1$. Which one would you pick?

 

The shitty one at 1$, obviously. Maybe on low-end chips, fine, but on high end, 350$+ cpus? intel, please.... 

There's more to thermal paste than the absolute lowest temps -- long term stability is FAR more important, and most thermal paste doesn't last long term and will require replacement -- obviously a bad thing. 

 

Of course, you could argue that solder would solve the problem, but it's entirely possible that there is an issue that prevents them from soldering even their HEDT CPUs (damaging too many CPUs for example, inconsistent temps, etc...).

[Lathlaer]

I feel like the fact that it was scrapped and reused had greater part in the equation. There should not have been that kind of difference. I would've bought a 7-10 degree but not 20, it feels like something else is at play.

 

Dunno, maybe you guys didn't know but AFAIK it's considered as well established that Intel uses Dow Corning as TIM so it can't be that hard to get it in normal form?

[NvidiaIntelAMDLoveTriangle]

Ok, not a perfect test. But it does show that Intel's paste is shiiiiit.

[Lathlaer]

13 minutes ago, NvidiaIntelAMDLoveTriangle said:

Ok, not a perfect test. But it does show that Intel's paste is shiiiiit.

There have been discussions about that TIM - on this forum, I think, and also on OCN. Bottom line is, it's not as bad as people think and it has some advantages over liquid metal as well. I think it was a case of crazy temp difference between singular core temps in multicore CPU's that were overclocked? I vaguely remember that discussion and it turned out that TIM offered less variability whereas with solder (or was it liquid metal?) you could have even something like 10-15 degrees difference between the core temps.

 

Why do I even bother, when I can link someone smarter who explains stuff ;-)

 

@MageTank

 

 

[JR88]

7 minutes ago, Lathlaer said:

There have been discussions about that TIM - on this forum, I think, and also on OCN. Bottom line is, it's not as bad as people think and it has some advantages over solder as well. I think it was a case of crazy temp difference between singular core temps in multicore CPU's that were overclocked? I vaguely remember that discussion and it turned out that TIM offered less variability whereas with solder you could have even something like 10-15 degrees difference between the core temps.

 

Why do I even bother, when I can link someone smarter who explains stuff ;-)

 

@MageTank

 

 

Well it looks like something Similar to all the Carbon Based Compounds, just some are better than others...but it's the difference too of thickness applied like mentioned and I think they put way too much. really the die just needs a light coat and that's it and should be sufficient.  So MX4 is an Excellent Candidate for my Testing of this. Will find out the Result's Next week.  

[Lathlaer]

Just now, JR88 said:

Well it looks like something Similar to all the Carbon Based Compounds, just some are better than others...but it's the difference too of thickness applied like mentioned and I think they put way too much. really the die just needs a light coat and that's it and should be sufficient.  So MX4 is an Excellent Candidate for my Testing of this. Will find out the Result's Next week.

To be fair I think that it was too much only because of the state the TIM was in. If it were a normal paste, it should've been fine because it would've spread more and I think it was Jay who showed that with non-conducting TIM it only makes a huge difference if you put too little of it. If you put too much and actually screw the cooler tight, the excess TIM will just go to the sides because it's in much softer form. Here I had the impression that they put too much but it was too solid to spread as thin as it should've done.

[Fiat-Libertas]

1 hour ago, Enderman said:

If you can't obtain the real thermal paste in liquid form, then measuring the temperature of the CPU before being delidded, then after delidding with the new compound, is still a much more accurate way of measuring the performance difference.

I'm pretty sure they've already done that video, but I don't have time to search for it now.

 

Basically we're back to #OutOfIdeas

[Guest The Viking]

56 minutes ago, djdwosk97 said:

There's more to thermal paste than the absolute lowest temps -- long term stability is FAR more important, and most thermal paste doesn't last long term and will require replacement -- obviously a bad thing. 

 

Of course, you could argue that solder would solve the problem, but it's entirely possible that there is an issue that prevents them from soldering even their HEDT CPUs (damaging too many CPUs for example, inconsistent temps, etc...).

true, I understand that lasting a long time is important too, but I'm sure there's another paste out there that is better than this and last longer too.

 

For the soldering, it's just down to money at this point. Until they actually prove, via some independant research (or observers, whatever really), that soldering their cpus is actually impossible, I'll keep calling them cheapos. AMD does it after all, on much more complex cpus. If tiny, 4 billion income a year AMD can do it, then gigantic 63 billion income Intel is more than capable of doing it.

[LeapFrogMasterRace]

The excuse for using third gen CPUs was not good enough, besides most of the severe Intel TIM hate is on cpus after Haswell anyway (probably the quotes they pulled???). Not only that this whole video is just wrong... 

[JoostinOnline]

Why didn't you use fresh pre-applied paste from an Intel stock cooler?

[dalekphalm]

44 minutes ago, JoostinOnline said:

Why didn't you use fresh pre-applied paste from an Intel stock cooler?

That was what I was expecting, to be honest. I'm sure you can order brand new Intel stock coolers (with the TIM pre-applied, still in "liquid" form).

 

I was expecting them to take one or two stock coolers, scrape off the TIM, and use that for the test.

 

This? Well, as @Enderman pointed out, there are some problems with the way they did this test.

Was it interesting? Yes. Was it entertaining? Yes. Was it educational or informative? I would argue no.

 

Hell, they probably could have talked to their Intel rep to either:

1. Get a supply of the TIM used

2. Get a supply of new-stock Heatsinks

3. Get the info on which exact TIM is used (allowing LMG to buy it themselves)

[JoostinOnline]

Just now, dalekphalm said:

That was what I was expecting, to be honest. I'm sure you can order brand new Intel stock coolers (with the TIM pre-applied, still in "liquid" form).

 

I was expecting them to take one or two stock coolers, scrape off the TIM, and use that for the test.

 

This? Well, as @Enderman pointed out, there are some problems with the way they did this test.

Was it interesting? Yes. Was it entertaining? Yes. Was it educational or informative? I would argue no.

 

Hell, they probably could have talked to their Intel rep to either:

1. Get a supply of the TIM used

2. Get a supply of new-stock Heatsinks

3. Get the info on which exact TIM is used

Well for those wondering, I did the same thing with my 4790k.  It was when the K models still came with stock coolers, and I applied the high end paste (MX-4 I think) to the stock cooler.  It shaved around 7C off at load.  That was when it was new though.  No paste is going to last 6 years if you try to remove and re-apply it.

[Lord Nicoll]

3 hours ago, AlexTheGreatish said:

By harvesting TIM from old Intel CPUs we can finally find out just how bad it is.

Buy better thermal paste
On Amazon: http://geni.us/CroMAlu
On Newegg: http://geni.us/Ale04G3

 

 

Following this video, I can say that if you want to make it a liquid again, use White Spirits, or I think they call it mineral spirits in the US, NOT an alcohol. I use white spirits to remove TIM because it's a lot better than even the isoproply alcohol, it will actually mix with the grease in the TIM, since both are oils. I did this with TIM from when Nvidia used IHS's and so had shit tons of it, and found performance did drop a little though, you'll get a about 1°C of a loss. The TIM will instantly turn the white spirits into a grey liquid then a goo, but use a very little amount. Also I5 750 CPUs are cheap and have a lot of TIM on them (they're not soldered) I included a photo of the i5 750 delidded. 

[Levent]

if you were to get a brand spanking new CPU that can be delidded, test it for a day then delid it and replace the TIM with something decent (AS5, MX4 etc not liquid metal) you would still get atleast 10 degree improvement, that thermal paste is just bad, very bad.

[JoostinOnline]

4 minutes ago, Levent said:

if you were to get a brand spanking new CPU that can be delidded, test it for a day then delid it and replace the TIM with something decent (AS5, MX4 etc not liquid metal) you would still get atleast 10 degree improvement, that thermal paste is just bad, very bad.

That's not true.  Just a couple weeks ago I replaced the thermal paste on a 6700k (a three year old CPU) with liquid metal.  There was a 15C difference.