Intel Haswell Delidding Tutorial

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Want to know how to rip the heatspreader off your brand new CPU? Hardware.Info has a guide posted today that should help you void your warranty in no time at all. :)

Intel's fourth generation of Core processors, also known as Haswell, are much more efficient in terms of performance per watt. Like the predecessor Ivy Bridge, they do still get very hot when you overclock them. There is something you can do to prevent that, but it's a risky endeavour. This guide shows you the best and safest way to do it.
 

CreepyUncleGoogle

Supreme [H]ardness
Joined
Mar 10, 2013
Messages
6,871
Exciting news! I bet computer people everywhere are going to totally flip their lids when they learn about it! :D
 

CreepyUncleGoogle

Supreme [H]ardness
Joined
Mar 10, 2013
Messages
6,871
You would figure CPU makers would use better TIM as how cheap it is.:confused:

Good Tims don't just grow on trees, you know.
Tim-monty-python-and-the-holy-grail-591629_800_441.jpg
 

CreepyUncleGoogle

Supreme [H]ardness
Joined
Mar 10, 2013
Messages
6,871
Seriously, how much cost would using a good TIM add to a CPU, especially when you order in bulk and make a shit load of chips? A dollar per chip?

I'm honestly not sure. The only other company I know of that makes large quantities of really good chips that are lidded is Pringles and I'm pretty sure they'll never talk about their processes because such things are trade secrets. :eek:
 

Dekar12

Gawd
Joined
Oct 2, 2003
Messages
745
This has been around since Ivy came out.

Plus did anyone else notice that the people on the site used a TON of new paste/glue?

And it seems that the vice method linked earlier is much easier/safer for your hands then using the razor blade.

Also, pretty sure not using Coollaboratory between the cpu/die after all that work is a waste as the gains are not nearly enough when just replacing paste with paste.
 

daglesj

Supreme [H]ardness
Joined
May 7, 2005
Messages
5,639
Plus did anyone else notice that the people on the site used a TON of new paste/glue?

Yeah looked awful. Less is more. People forget that paste is only required to fill the micro-gaps between the two surfaces. It's not abut making it look like a New York deli filling in a sandwich.
 

Thermite Paste

Supreme [H]ardness
Joined
Dec 17, 2003
Messages
5,651
I've been wanting to delid mine for ages, but has anyone come out with a mounting solution yet without the heatspreader? ie a shim like the old AXP days.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Accidentally the whole Haswell tutorial:

Step 1: look at botched attempt at delidding.
Step 2: cry.
Step 3: swear off Intel forever.
 

rat

Supreme [H]ardness
Joined
Apr 16, 2008
Messages
4,915
Wow - total fail.

1st - Article is a little late - like 9 months.
2st - Don't use this method. Use the vice/wood method.
3rd - The TIM intel uses is fine. The problem is the distance from the chip and IHS. http://forums.anandtech.com/showpost.php?p=34053183&postcount=570

The results on that link are pretty eye opening. Seems Intel is using pretty good TIM material as-is. What explains the thickness/gap? Reserved space, tolerances, heat expansion?
 

Mike89

Gawd
Joined
Jan 27, 2003
Messages
702
Delidding is some crazy shit as far as I'm concerned. No way in hell would I ever attempt it, but hey I'm just a chicken shit that would surely destroy a perfectly good cpu.
 

UZ7

Limp Gawd
Joined
Dec 5, 2008
Messages
171
Delidding is some crazy shit as far as I'm concerned. No way in hell would I ever attempt it, but hey I'm just a chicken shit that would surely destroy a perfectly good cpu.

12349_10104725410161971_1841943742_n.jpg


We just did this two days ago lol, got the 4770K in the mail, tested it.. stress test at stock gave it about 73C, delid, cleaned, put clu, mounted, put nth1 tried stress test again about 54C stock load so yeah it was a big difference. I didnt believe it at first till I saw the results :eek:
 

Seventyfive

[H]ard|Gawd
Joined
Jul 14, 2004
Messages
1,347
12349_10104725410161971_1841943742_n.jpg


We just did this two days ago lol, got the 4770K in the mail, tested it.. stress test at stock gave it about 73C, delid, cleaned, put clu, mounted, put nth1 tried stress test again about 54C stock load so yeah it was a big difference. I didnt believe it at first till I saw the results :eek:

So you put the lid back on but with different TIM or you left the lid totally off?
 

UZ7

Limp Gawd
Joined
Dec 5, 2008
Messages
171
So you put the lid back on but with different TIM or you left the lid totally off?

[nth1]
---------
[ihs]
---------
[clu]
---------
[die]

So while you clean off the old stuff and the black rubber glue :p (pretty sure its what was causing the gap between the die and ihs), you apply clu which is really like liquid metal, you brush it on and its like t-1000 from the terminator. Anyway you brush on CLU, put the IHS back on, mount it carefully on the motherboard then you can apply any thermal paste you want on top of the IHS like you would normally do.
 

DejaWiz

Fully [H]
Joined
Apr 15, 2005
Messages
21,623
I so want to try delidding an IB or HWL and reattaching the IHS with Indium solder (the same 157C melting point stuff that Intel uses).



Been pondering that maybe by placing it in the oven at 315F (approx. 157C) with everything inverted and putting a small bit of weight of some sorts on top of the whole shebang, it would ensure a near factory fitment of the IHS once the solder melts, oven is turned off, and let sit long enough for the solder to cool and joint.



Pin-friendly Weight

|
v

CPU/Die (upside down, of course)

|
v

Solder

|
v

IHS (again, upside down)

|
v

Flat Non-warping Baking Pan/Sheet or Pizza Stone



...if done with enough care (including sufficient cleaning and prep of the solder contact surfaces, de-oxidizing the Indium solder prior to use, and effective alignment of everything), I think this would certainly work. All the benefits of the IHS with none of the thermal conductivity limitations by using any kind of non-solder TIM underneath it!
 

platero

n00b
Joined
Jul 15, 2013
Messages
29
We just did this two days ago lol, got the 4770K in the mail, tested it.. stress test at stock gave it about 73C, delid, cleaned, put clu, mounted, put nth1 tried stress test again about 54C stock load so yeah it was a big difference. I didnt believe it at first till I saw the results :eek:

That's honestly bananas. I haven't had any issues overclocking my i5-3570K to 4.4ghz on air, but I would love lower temperatures when I do so.

Does the de-lidding allow you to use lower voltages when overclocking, or just make higher voltages safer due to the lower overall temperature?
 

ontariotl

Limp Gawd
Joined
Oct 2, 2008
Messages
242
Wow - total fail.

1st - Article is a little late - like 9 months.
2st - Don't use this method. Use the vice/wood method.
3rd - The TIM intel uses is fine. The problem is the distance from the chip and IHS. http://forums.anandtech.com/showpost.php?p=34053183&postcount=570

Agree 100%.

Vice is so much better to delid than knife. Especially with those resistors on the side of the die of the haswell could easily be damaged with a blade. 3 out of 3 attempts for the vice trick. 2 out of 3 with the blade, and will never go back to it.

Is it just me, or do some of the pics show a different delided chip which has the green laminate gouged on one side of the chip? That can't be good.
 

Wshmaster0

2[H]4U
Joined
Jul 11, 2007
Messages
3,659
That's honestly bananas. I haven't had any issues overclocking my i5-3570K to 4.4ghz on air, but I would love lower temperatures when I do so.

Does the de-lidding allow you to use lower voltages when overclocking, or just make higher voltages safer due to the lower overall temperature?

Well not 100% sure on ivy but haswell has a lower max voltage due to the smaller die size so with the added voltage the CPU can only go so far before reaching max safe but with the poor contact it caused a good +10-20C on top of that. The stock was already hitting temps like i7 9xx days. As for getting better voltages you could probably get higher than what you have now, but also take into consideration max safe for your chip. If you're hitting high temps then the contact could be the limiting factor. Remember if you're delidding expect the risks. But if carefully done, its doable and the gains are great :).

If you're looking for higher over clock I would try to test if the chip can handle it first and if temps are the limiting factor then delid could potentially help with that.
 

UZ7

Limp Gawd
Joined
Dec 5, 2008
Messages
171
That's honestly bananas. I haven't had any issues overclocking my i5-3570K to 4.4ghz on air, but I would love lower temperatures when I do so.

Does the de-lidding allow you to use lower voltages when overclocking, or just make higher voltages safer due to the lower overall temperature?

If temperatures were your limiting factor then it would help. Lower voltages would come more from quality of motherboard, psu, ram oc etc...
 

AaronP

[H]F Junkie
Joined
Jan 13, 2005
Messages
11,528
Reminds me of my days of de-lidding Pentium 4s and AMD Athlon 64s...good times!

001.jpg

002.jpg

003.jpg

6000$_keychains_002.jpg
 

SGA76

[H]ard|Gawd
Joined
Jul 23, 2013
Messages
1,955
Yeah looked awful. Less is more. People forget that paste is only required to fill the micro-gaps between the two surfaces. It's not abut making it look like a New York deli filling in a sandwich.

I used to use Arctic Silver Ceramique, squeezed out an amount about the same as 2 1/2 - 3 short grains of rice, spread it out with my old library card. (because hey, with the net, why go to somewhere you've got to be quiet right?) A little goes a long way, too much can cause heat issues.
Always used ASC on my GPUs and north and south bridges as well, gotta love non conductive on anything that could arc.
Its been too long since I built myself a new PC.....
God I feel old.:(
 
Top