Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz

[tigersty1e]

I ended up seeing temperature degradation with the prolimatech Tim (not branded pk-1, but just says prolimatech).

I actually say a few degrees worse 5 hours after my initial application and tests without a shutdown.

I have clu on the die and prolimatech on the ihs. I'll have to check tonight to see how it fares.


Posted from Anandtech.com App for Android

 

[tigersty1e]

if you drop 10c with delidding and everything else is kept constant, does that mean the air coming off your cooler is hotter than before when temps were 10c higher?

it has to, right?

the cooler is soaking up more heat from the increased efficiency, so its transferring more heat to the air that is passing through.

so our cpus are cooler, but our cases are hotter.

 

[Idontcare]

if you drop 10c with delidding and everything else is kept constant, does that mean the air coming off your cooler is hotter than before when temps were 10c higher?

it has to, right?

the cooler is soaking up more heat from the increased efficiency, so its transferring more heat to the air that is passing through.

so our cpus are cooler, but our cases are hotter.

If everything were truly kept constant, meaning power consumption remained the same despite the lowered operating temperatures, then the air coming off the cooler would remain the same temperature because the same amount of heat (same watts) is being dumped into the same volume of air.

A 50W processor is consuming 50W of electricity and producing 50W of heat. That means the air is having 50W of heat dumped into it, and the air will heat up appropriately.

The air temperature is not dependent on the temperature of the heat source, it is dependent on the total amount of heat involved.

Consider for example a lit match, it burns at the same temperature as a bon-fire (or a whole stack of wood burning in a fireplace)...but the burning match is not producing as much heat as the bon-fire, and so the match doesn't really heat up the room (but the bon-fire sure will!).

Now the reality is that your heat source, your CPU, actually consumes less power as you reduce the operating temperatures. (in other words you have not held everything constant in practice) The static leakage is reduced. So in truth your delidded IB is now producing less heat in addition to operating at a lower temperature, so the air temperature will also be lower because it is absorbing less heat overall.

 

[taq8ojh]

I have a dumb question regarding the tools: Where on Earth do you get razor like that? I googled like mad and I can't seem to find anything looking even remotely like that being sold in this country.

 

[Idontcare]

I have a dumb question regarding the tools: Where on Earth do you get razor like that? I googled like mad and I can't seem to find anything looking even remotely like that being sold in this country.

Mine came in a glass-cleaner (scraper) kit.

What you want is called the "standard single-edge razor blades".

 

[dawp]

I have a 3770k that has a temp difference between cores of around 7*. one core is usually around 25 -27* and the other 3 are around 35*. do you think replacing the intel tim would equalize the temps? I redid the paste with the same results.

 

[Idontcare]

I have a 3770k that has a temp difference between cores of around 7*. one core is usually around 25 -27* and the other 3 are around 35*. do you think replacing the intel tim would equalize the temps? I redid the paste with the same results.

It is possible, but there are plenty of reasons why it wouldn't help.

Temperature differences between cores is the "symptom", like coughing it is what we can observe. But just like with coughing, any number of diseases can be the root-cause of the cough, the temperature difference between cores can be caused by any number of root-causes.

One of those possible causes is the CPU TIM under the IHS. You won't know until you replace it. But that doesn't mean replacing it will fix the problem because it could be caused by other issues (inside the cores themselves, unfixable stuff).

 

[tw33k]

Been absent for a while and just catching up so if my question has been answered please forgive me and provide a link. I've been in discussions with people who claim that if left long enough Liquid Ultra will "fuse" with the IHS. I have never heard of this before. I am aware that it may take a little extra effort to separate the two and both surface will require sanding but to actually fuse together? Indigo Xtreme would be more likely I'd imagine but does Liquid Ultra solidify at all over time? <solidify might not be the right word but you know what I mean>

People point to this post onwards on xtremesystems.org as evidence that it does kind of solidify (the author claims it was "semi-solid")

 

[Idontcare]

Been absent for a while and just catching up so if my question has been answered please forgive me and provide a link. I've been in discussions with people who claim that if left long enough Liquid Ultra will "fuse" with the IHS. I have never heard of this before. I am aware that it may take a little extra effort to separate the two and both surface will require sanding but to actually fuse together? Indigo Xtreme would be more likely I'd imagine but does Liquid Ultra solidify at all over time? <solidify might not be the right word but you know what I mean>

People point to this post onwards on xtremesystems.org as evidence that it does kind of solidify (the author claims it was "semi-solid")

Its true, it doesn't require all that much time either.

You can see the mess it made of my H100 in this post from when I mounted my H100 to the bare-die of my 3770k using Liquid Ultra.

I had to lap the H100 to remove it all.

 

[tw33k]

How long would be enough for this to happen? I've just built a test bench and will be doing lots of tests in the coming weeks so if I have a rough time frame I'll make sure I leave the H100i attached long enough so I can see for myself.

 

[Idontcare]

How long would be enough for this to happen? I've just built a test bench and will be doing lots of tests in the coming weeks so if I have a rough time frame I'll make sure I leave the H100i attached long enough so I can see for myself.

What you see there in my post on Liquid Ultra with the H100 happened in the span of about 3-6 hours

 

[tw33k]

Wow! There's a guy on OCN who claims to have completely removed Liquid Ultra from his IHS and heatsink by using metal polish. I'm loath to pull mine apart but I'm seriously tempted to see for myself.

 

[tigersty1e]

I read somewhere that it does stain (get hard) on metal surfaces. but it will stay liquidy on the die and PCB.


Posted from Anandtech.com App for Android

 

[tw33k]

It definitely stains but the staining doesn't appear to impact temps (well it hasn't for me) Once I de-lid my new chip and continue testing different TIMs I'm going to see if I can get the Liquid Ultra off with some metal polish. I want to avowing lapping if possible.

 

[MustangSVT]

Few questions.

My current 3570k is currently running at 4.4g / 1.23v. But when I do 4.5g, it requires at least 1.3 and hits high 90s in linx or any cpu stressing programs.

Q1. So, I think mine has large gap between core and IHS. Is this a correct assumption?

And from reading IDC's data, as he states the issue was space between core and IHS, not the type of thermal paste used that caused poor oc ability of 3570k. And there IS either pump out or some effect that causes most thermal paste to perform less over period of time. (except Liquid pro).

Q2. Why didn't this happen with intel's original thermal ? was it because of the sealing?

Q3. And this still seems to be an issue with direct die mounting. Why wasn't this an issue back in the days of direct die heatsink mount cpus?

 

[Idontcare]

My current 3570k is currently running at 4.4g / 1.23v. But when I do 4.5g, it requires at least 1.3 and hits high 90s in linx or any cpu stressing programs.

Q1. So, I think mine has large gap between core and IHS. Is this a correct assumption?

I don't know if that is the correct assumption but it seems the most plausible one.

There are other possibilities, such as you may just have a crappy (leaky) CPU to begin with.

But it does seem to be a general rule that you will max out around 4.5GHz with high temps until you delid and reduce that gap.

Q2. Why didn't this happen with intel's original thermal ? was it because of the sealing?

Intel's TIM is very solid, like a plastic shim in terms of texture. It cuts with a razor blade like you were cutting plastic. So pump-out is not possible.

Q3. And this still seems to be an issue with direct die mounting. Why wasn't this an issue back in the days of direct die heatsink mount cpus?

Pump-out is the result of a number of conditions coming together at the same time.

I can't recall the specifics of direct-die mounting from a decade ago when we did it on our AMD chips but one difference I suspect is that we run our OC'ed Ivy Bridge chips much hotter (even with direct-die mounting) than we did with our AMD chips from so long ago.

 

[MustangSVT]

I don't know if that is the correct assumption but it seems the most plausible one.

There are other possibilities, such as you may just have a crappy (leaky) CPU to begin with.

But it does seem to be a general rule that you will max out around 4.5GHz with high temps until you delid and reduce that gap.


Intel's TIM is very solid, like a plastic shim in terms of texture. It cuts with a razor blade like you were cutting plastic. So pump-out is not possible.


Pump-out is the result of a number of conditions coming together at the same time.

I can't recall the specifics of direct-die mounting from a decade ago when we did it on our AMD chips but one difference I suspect is that we run our OC'ed Ivy Bridge chips much hotter (even with direct-die mounting) than we did with our AMD chips from so long ago.

You are... (an oc) GOD. thank you for answers.:thumbsup::thumbsup::thumbsup:

 

[Idontcare]

You are... (an oc) GOD. thank you for answers.:thumbsup::thumbsup::thumbsup:

LOL , I wish, but sadly I must confess you are vastly over-estimating my capabilities I'm just a dude with a hammer and a dremel

The real OC'ing gods are the people who make the TIMs, the mobos, and the CPUs and tune/tweak their products to be capable of OC'ing as they do once they get into our hands :thumbsup:

 

[amenx]

Delid without a razor. Seen a couple of peeps in another forum do this as well.

http://www.youtube.com/watch?v=2jjxvY6Y-P4

 

[bgt]

Delid without a razor. Seen a couple of peeps in another forum do this as well.http://www.youtube.com/watch?v=2jjxvY6Y-P4

Ahh.....thats a nice way to do it.

 

[Idontcare]

Delid without a razor. Seen a couple of peeps in another forum do this as well.

http://www.youtube.com/watch?v=2jjxvY6Y-P4

I'm curious if this works often or just some of the time. And by "works" I don't mean "gets the IHS removed" because that is clearly going to happen provided enough lateral force is applied.

Rather, by "works" I mean to say I wonder how often the CPU PCB itself is going to be damaged to the point of being broken with delaminated layers and so on.

Personally I'd be a little uncomfortable slamming my PCB that hard with the butt end of a 2x4.

 

[amenx]

I'm wondering if he pre-cut the rubber glue to make it come off easier.

 

[Idontcare]

I'm wondering if he pre-cut the rubber glue to make it come off easier.

I too wondered if some similar kind of fakery was going on. There is a lot of surface area held together by the black adhesive.

Considering how hard it is to push a blade through the adhesive, seems a bit suspect that a simple smack with a block of wood is going to delaminate the entire IHS. :hmm:

 

[ehume]

I too wondered if some similar kind of fakery was going on. There is a lot of surface area held together by the black adhesive.

Considering how hard it is to push a blade through the adhesive, seems a bit suspect that a simple smack with a block of wood is going to delaminate the entire IHS. :hmm:

In general, I would agree with you. But some materials are stiff when shocked -- while they can deform at lower speeds, they can't deform fast enough at high speed, so they separate.

That would explain what we see in the video.

 

[Idontcare]

In general, I would agree with you. But some materials are stiff when shocked -- while they can deform at lower speeds, they can't deform fast enough at high speed, so they separate.

That would explain what we see in the video.

You are thinking of non-Newtonian liquids which would apply to adhesives that are below their Tg (glass transition temperature), as the adhesive most certainly is.

I'm willing to have my skepticism eliminated via education by numbers, lets see a few more reports of this actually working before we just assume it isn't fake.

Any volunteers?