Preparing to delid 4770K

Chipfiref

Member
Aug 1, 2013
102
0
71
Hi guys new to the forums

This machine:
-------------
ASRock Extreme6
4770k
Stock cooler ...> Noctua NH-U14s after delid
32GB GSkill 1600
Samsung 840 Pro SSD + whatever WD drives
Corsair HX750w
all in a Coolmaster Cosmos circa 2008? (on sale at CompUSA)

I had the case and PS and wanted to go with a Haswell. Never really overclocked before, but I anticipate going to 4.3 or 4.5 - and really want to get cooler in any case.


currently I am burning in and benchmarking, just sitting here I am getting 51C on the stock cooler.

When I played some Civ V on this Intel 4600 graphics I went to 65C and that was just mid-game.

So needless to say I want to do the delid with CLU...
 

Chipfiref

Member
Aug 1, 2013
102
0
71
I am going to post some pics of my progress on this project - that's what I like to see so that's what I try to do.

IMG_20551.jpg




kind of boring but I will get some better ones soon.
 

Chipfiref

Member
Aug 1, 2013
102
0
71
I also have these other chips :

i7-930
i7-860
i7-2600K
a couple of i5s

I am an "IT professional" so I have collected a few toys :)

my first PC was a used 286 and my second was a used 386sx (worse than a DX)
 
Last edited:

oneofusjustin

Member
Feb 18, 2008
161
0
71
i did the sloppiest most unprofessional careless delid in history and went from 96c to 76c load temps with CLLP
 

Chipfiref

Member
Aug 1, 2013
102
0
71
wow i would be freaked out haha

nobody thinks that adding such force to the side of the CPU won't harm it?

I think its the force exactly parallel with the pcb, which is greater than the silicone glue force on the pcb, but not enough to distort the IHS (if you do it just right). The IHS won't harm the cpu die because the pcb wants to pop off when the glue resistance is overcome. Plus the cpu die is a tough booger, as Idontcare explained.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
wow i would be freaked out haha

nobody thinks that adding such force to the side of the CPU won't harm it?

My concern would be the risk of delaminating the PCB layers, as that will enable oxidation of the copper trace lines in time and your CPU becomes dead in weeks or months :(

I guess it is all a matter of hitting it just right.
 

Chipfiref

Member
Aug 1, 2013
102
0
71
Here is my beautiful delidding vise.

IMG_20561.jpg


IMG_20581.jpg




This is a Drill Press Vise sold by Home Depot for $20 - what a great deal. It can be used for all kinds of projects on a desk. Very heavy and a 41/2" opening! It will last me a lifetime. :)

IMG_20591.jpg
 
Last edited:

Chipfiref

Member
Aug 1, 2013
102
0
71
CLU just got here this afternoon.

Doing a little benchmarking before I delid...keep in mind that this is the stock cooler only

Aida64CPUtempsFanspeddatidle.jpg


Before I started benching with Aida64

Aida64Sress1.jpg


My craptastic chip

Aida64Sress2.jpg


and...that's enough

Aida64Sressgettingbacktoidle.jpg


luckily things are calming down fairly quickly






Aida64Sress2ZOOM.jpg


Better view just before I quit




So what's the deal with the one core that stayed at 42? Did I exclude it or is something wrong? Sorry I'm a noob at Aida64 and all the others.

|
V



Aida64SressStatsview13sec.jpg


Well that's special - the stats view indicates the core is NOT hung at 41C
 
Last edited:

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
i7-930
i7-860
i7-2600K

You listed these as if you were considering maybe messing with them too. However, if I remember correctly those are soldered, so please do not try to delid them. I believe it's only Ivy Bridge and up (i[3/5/7]-3xxx and higher). Unless I'm wrong, then someone please correct me. :p
 

Chipfiref

Member
Aug 1, 2013
102
0
71
You listed these as if you were considering maybe messing with them too. However, if I remember correctly those are soldered, so please do not try to delid them. I believe it's only Ivy Bridge and up (i[3/5/7]-3xxx and higher). Unless I'm wrong, then someone please correct me. :p

Yes, those are not going to be delidded since they are soldered
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
I'm a bit interested in how well the vise works. When I was looking for one, I figured that any one that has bolt holes/grooves would probably shift around too much; however, the guy in that German video used one and it was fine. I went with a table vise, which essentially has a clamp on the bottom. I think it was only slightly more than the one that you mentioned, but it doesn't open nearly as far!

Yes, those are not going to be delidded since they are soldered

Okay. :) Just making sure!
 

Chipfiref

Member
Aug 1, 2013
102
0
71
Thanks for the warning anyway - I feel the need to do the same after what I am about to do!

I have monitored the temps on this box while playing some Civ 5 now that I have Aida64 up and running on this box, and am more familiar with it. I am surprised that I get temps of 80C on the 4 cores, just playing Civ 5 on Intel 4600. Would the temps be lower if I had a discrete GPU? I am not too sure.

What I do see is that any kind of stress test on this chip except for CPU only brings me to throttling within 1 minute. And CPU only looks like I may hit it in 10-15 minutes...what I am saying is that if I uncheck "stess CPU" and "Stress FPU" (leaving the other items checked) and only run with "stress GPU", I will get throttling in 1 minute! Even less time if I only check "stress FPU", which is what you would expect from seeing others' stress tests.

Aida64SressCPUonly.jpg


And I now realize the 41C line was not a core but the CPU package, I just couldn't see all 4 cores on the graph for a while since one was hiding behind the other with the same temp. But I do wonder how the package can stay so cool while the cores are boiling hot.
 
Last edited:

Chipfiref

Member
Aug 1, 2013
102
0
71
So, conclusion at this point? DE-LID! And I can only hope I get a 20C delta, otherwise could I even stress this thing?

In the back up my mind, I am wondering if I popped a plastic cooler pin out of the motherboard off of my most hated PC device of all time (other than cheap external drive enclosures). I spent at least 20 minutes wrestling with this Intel stock cooler until the pins stayed put, and have many times on other builds.

So here is what I will do - attempt to carefully slice into the silicone glue holding the IHS onto the chip pcb. I will avoid scratching the fiberglass pcb to prevent damage to traces. I will avoid cutting into the die or the little row of capacitors further out from the die on the right side. If I have trouble I will use the vise and wood block method to shear the pcb off after hopefully some of the silicone glue force has been diminished.
 
Last edited:

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,087
3,598
126
BE VERY CAREFUL THIS TIME WITH HASWELL..

http://www.xtremesystems.org/forums/showthread.php?286340-4770k-4670K-IHS-Removals
fe5mny.jpg


there are resistors parked right behind that rubber seal.
(u see right below the diag pins...)

One OOPS is a quick way to kill that proc.
Honestly.. thats wayyyy too close for comfort for even me... :\
Ive killed a few AMD chips b4 from stupid impatience... and i had MORE room then that to play with.

If i absolutely had to delid... use a razor and cut into the rubber sideways.
Get the 3 sides off.. or at least 2 sides with the last 2 partial.... the sides which has the resistors.
Then i would pry the IHS up ripping the last remaining rubber attachments.

I would also draw a line on my razor and NEVER EXCEED that line drawn on razor.
Otherwise you might forget exactly how much that razor cut into.

Anyhow good luck... im also waiting to see your results...
I would try to more rip the last part of that top rubber after i cut into as much as i could without breaking it complete.
Once u got the other 3 sides off... u could try to torque the ihsup and rip the rubber off. (like opening a book)
 
Last edited: