CPU Delid

[Light-Yagami]

2 minutes ago, MageTank said:

It will still be a reduction of temps, just not as drastic as the liquid metal TIMs. I know people that use Kryonaut on their bare dies and still see a 10-15C drop (again, mostly because the glue itself is bad). Use whatever you feel comfortable using. I personally prefer CLU, because liquid metal is superior than silicone based pastes by the tune of about 5-10C depending on which pastes you use. Most DICE or LN2 enthusiasts don't use liquid metal TIMs because gallium is not rated for under 10C usage, and tends to ruin when going subzero. 

 

If you have the time and money, I recommend testing both (maybe start with silicone paste first, then try CLU) and see the differences for yourself. That's the easiest way to know which is better or "worth it" for you. 

What about this?  http://www.coollaboratory.com/?s=liquid+metal

If I place this on the cpu die? will it do the same?

[MageTank]

1 minute ago, Light-Yagami said:

What about this?  http://www.coollaboratory.com/?s=liquid+metal

If I place this on the cpu die? will it do the same?

I have never used it. Wouldn't know honestly. Spreading the liquid metal on the die isn't really difficult at all, it's the IHS you will have to worry about. It's extremely tedious to spread gallium on top of nickel, which is why I recommended sanding the underside of the IHS down (or at least marring it a little) to make it better adhere to it. Looking at the safety data sheet for that liquid metal pad, it seems to be primarily composed of copper and indium, rather than gallium. I can't imagine it will perform as good as standard CLU or CLP. 

[Teddy07]

24 minutes ago, Light-Yagami said:

I have everything exept for liquid metal paste. I consider myself capable enough to delid a cpu without taking a shit while doing it. The only thing I want to know now is if regular thermal paste is also okay and how much decrease of preformance there is compared to liquid metal paste

liquid metal is way better there is no discussion. You delid while you can do it so just use liquid metal

[just tilt]

In a video Linus did it only dropped by 3 or 4 degrees. He said something along the lines of
 

Quote

The CPU temps gone down by only a few degrees. If we put some of the more exotic liquid metal TIM (thermal interface material) we could expect a few degrees more. That's why I never condoned CPU delliding; it's just not worth it.

 

[MageTank]

2 minutes ago, just tilt said:

In a video Linus did it only dropped by 3 or 4 degrees. He said something along the lines of
 

 

Yeah, and he was notoriously wrong. Part of the reason why very few enthusiasts take him seriously.

[Light-Yagami]

1 minute ago, MageTank said:

Yeah, and he was notoriously wrong. Part of the reason why very few enthusiasts take him seriously.

ohhhh GET REKT. It's been said. Insert THANK YOU gif here!

[just tilt]

3 minutes ago, MageTank said:

Yeah, and he was notoriously wrong. Part of the reason why very few enthusiasts take him seriously.

Ohhhhhh okay, sorry for the missinformation then. Never delidded a CPU myself, that's why I replyed based on his video.
Can you explain what he was wrong at and what kind of results should one get when delidding a CPU and replacing the TIM with some quality stuff?

EDIT: @Light-Yagami lol wouldn't say i got rekt. He just has waaaay more knowledge than I do in that area.
@MageTank Sorry bud, didn't see you already posted the explaning part of why Linus is wrong. I'll check that out right now. Thank you for the thorough explanation!

[MageTank]

1 minute ago, just tilt said:

Ohhhhhh okay, sorry for the missinformation then. Never delidded a CPU myself, that's why I replyed based on his video.
Can you explain what he was wrong at and what kind of results should one get when delidding a CPU and replacing the TIM with some quality stuff?

There was quite a few things he did wrong, but some of the most important being:

 

• He failed to acknowledge the difference delidding made in his per-core temp variance (one of the biggest reasons we delid)
• He used Coolermaster Maker Gel (notoriously bad for turning to liquid under thermal shock, similar to Noctua's NT-H1). These pastes are not bad when used on top of the IHS, but are awful when used on the bare die.
• He didn't manually spread his awful paste (can't just let the IHS spread for you on the bare die, it's not the same as letting the pressure of a heatsink spread for you). The lips on the outer edges of the IHS won't allow for enough pressure to evenly spread the paste with the IHS alone unless you use copious amounts of paste (he did not). 

Now, a lot of people complain about his use of a stock cooler, and about his refusal to overclock (to see if thermal headroom allowed for gains in performance) but I don't really see it as that big of a deal. What I take issue with, was his blatant misinformation that has spread rampantly through not just this community, but every tech forum in the world because they view this man as the authority on all things tech. Now we have people telling others not to delid, because it's "only a few C difference" when that's just plain wrong. People are now attacking Dow Corning, as if it's a bad paste, when it's actually a phenomenal paste for what it is designed to do. DC-5688 is some of the most resilient paste ever created, and it's longevity is second to none. There is a reason why it's used. It's thermal capabilities as far as temperatures go isn't the best, but it outlasts the CPU itself. You can delid, remove the glue, and replace it with the exact same DC-5688 and still see a solid reduction in temperatures. Blaming the paste is just wrong.

 

Linus did just that, and in doing so, let his clueless army loose upon the world to drive that misinformation into peoples heads. He has also refused to go back and re-test it with a proper methodology (same with his memory tests) because he seems to believe he is right, despite the overwhelming evidence proving otherwise. I advise people to check with multiple sources, and even try to validate the claims for themselves before actually spreading the information as fact. That way, we at least get more valuable information and additional sources, and can prevent misinformation from slipping through the cracks. 

[Teddy07]

I just ordered CoolLaboratory Liquid Ultra liquid metal 1g

[done12many2]

2 hours ago, PCGuy_5960 said:

Depending on the Thermal Paste, you can get 3-5C improvement. Not worth it IMO

 

More like 20c to over 25c depending on how bad the gap between die and IHS is, which is caused by the horrible silicone sealing of the IHS at the factory. 

 

2 hours ago, PCGuy_5960 said:

With an AIO, delidding will not result in 20C lower temps...

 

I'll be doing another 7700k tonight and I'm willing to bet a 7700k that it drops in excess of 20c with the AIO on my test bench.  I've delidded enough of them to know that it's just going to happen.

 

2 hours ago, PCGuy_5960 said:

 

 

 

 

One of the worst examples that exists.

 

1 hour ago, Light-Yagami said:

What about this?  http://www.coollaboratory.com/?s=liquid+metal

If I place this on the cpu die? will it do the same?

 

Great stuff.  I prefer the Liquid Metal Ultra to the Pro.

 

[PCGuy_5960]

2 minutes ago, done12many2 said:

I'll be doing another 7700k tonight and I'm willing to bet a 7700k that it drops in excess of 20c with the AIO on my test bench.  I've delidded enough of them to know that it's just going to happen.

You don't have to

Hardware Unboxed already did that and got a 25C+ drop... (with an H100i)

I just thought that the 6700K used better thermal paste.

[done12many2]

1 minute ago, PCGuy_5960 said:

You don't have to

Hardware Unboxed already did that and got a 25C+ drop... (with an H100i)

I just thought that the 6700K used better thermal paste.

 

No worries bud.  I am delidding another great 7700k that I found so I was going to delid it anyways.  I'm in the process of binning dozens of these chips.