4770k /4670K IHS Removals

[Demo]

@ andressergio


Nice delid and great results, I still need to muster up the courage to delid!

[munim]

Oh no i told u to get pro not ultra !!!

Oops! Why is that? Price or performance?

[andressergio]

Oops! Why is that? Price or performance?

performance...

[Fosco]

performance...

How much are on air\water 1-2c ?

Just removed on my 4th haswell

Thanks

[andressergio]

How much are on air\water 1-2c ?

Just removed on my 4th haswell

Thanks

more than 5C...

i saw ppl using liquid ultra and dropped 14C mine drops always same 29C and more with my custom WC

cheers !
Sergio

[Justintoxicated]

I have seen quite a few reviews now that show the ultra working better than the pro, but that's between the IHS and block (where the IHS is not lapped) so I'm guessing this is what accounts for the difference.

I ordered some Phobya Liquid metal myself, supposed to be pretty much the same thing but only $6 for an oz rather than $14. Hope I didn't make a terrible decision.

Now I just need to find a way to use it for direct die cooling without it becoming permanent or is this not possible? I guess I should have bought a nickel plated block instead of a copper one to avoid problems with Phobya liquid metal?

[munim]

My temps dropped 11 to 14 degrees under AIDA FPU stress depending on the core. This is obviously not as much as others have using liquid Pro (I'm using Ultra) but it's good enough and I'm really happy. I top out at around 73 degrees under linX @ 188 gflops 4.5 GHz 1.245 vcore. In games and other tasks that I actually perform I barely reach 50 degrees lol! I'm loving it, I've also reduced my fan speeds so my pc is inaudible.

Also, before I couldn't even boot 4.7 GHz but now I can, with 1.3Vcore but it's not stable and I don't really need a crazy oc like that anyway.

If I had to do it over again I'd follow Sergio's advice more closely and go with Liquid Ultra.

@Justintoxicated:
Liquid Ultra wipes off totally clean from the cpu die but not from the IHS. This stuff is so fine that it literally looks finely painted onto the IHS if you wipe it. The included scuffing pad removes it quickly though. I'm not sure if it's even necessary to totally clean it off the IHS because at the end of the day it's filling in the microscopic grooves in the IHS underside. I haven't used it topside and don't plan to (using MX4 there).

[FaYgOkRiS]

Ok this might be a dumb question, but I have been outta the game for a long time(deployment) is this only for air & watercooling situations, I have an SS being built by Sdumper right now and just want to know whether or not to bother with this.

[todda7]

Ok this might be a dumb question, but I have been outta the game for a long time(deployment) is this only for air & watercooling situations, I have an SS being built by Sdumper right now and just want to know whether or not to bother with this.

If you can bother with an SS, you can definetly bother with relidding. Get the lid of using the vice method (youtube: 4770k dellid vice), remove all the glue, and apply Coollaboratory Liquid Pro. Temp may drop 15-20c or more under SS.

[FaYgOkRiS]

If you can bother with an SS, you can definetly bother with relidding. Get the lid of using the vice method (youtube: 4770k dellid vice), remove all the glue, and apply Coollaboratory Liquid Pro. Temp may drop 15-20c or more under SS.

Wouldnt i need to reseal the IHS to the chip inorder to avoid having a pocket of air form condensation?

[Vivi]

Ok this might be a dumb question, but I have been outta the game for a long time(deployment) is this only for air & watercooling situations, I have an SS being built by Sdumper right now and just want to know whether or not to bother with this.

like todda said, it will be excellent for SS. But for ln2 you will lose about 50mhz.

here some new delid results, 4770k are hot...



4770k - off-forum 24.4c drop new batches are warm.. but clock nice


4770k - Navz 19c drop


4770k polygon -29c drop new warm batch too, but good clocker. i just want to iterate that i am not doing anything phony, i keep same auto fanspeeds, same paste, same cooler..


VPII's 4770k


LOAD - 21 C drop overall not bad not bad, these 4770k's are warmmm

[FaYgOkRiS]

like todda said, it will be excellent for SS. But for ln2 you will lose about 50mhz.

here some new delid results, 4770k are hot...


4770k polygon -29c drop new warm batch too, but good clocker. i just want to iterate that i am not doing anything phony, i keep same auto fanspeeds, same paste, same cooler..


VPII's 4770k


LOAD - 21 C drop overall not bad not bad, these 4770k's are warmmm

Well you have definately convinced me but what about glueing the IHS back down properly to seal it, wouldnt the air pocket cause condensation otherwise

[andressergio]

congrats mate delid on Haswell = WIN !!!

[Vivi]

Well you have definately convinced me but what about glueing the IHS back down properly to seal it, wouldnt the air pocket cause condensation otherwise

i have not tested this myself regarding air pocket! I put nail polish on the little chips of all my cpu's just incase, so i recommend you coat those little resistors with something waterproof

[docwhom]

I put nail polish on the little chips of all my cpu's just incase, so i recommend you coat those little resistors with something waterproof

Do you use clear or coloured nail polish and what brand did you use - Rimmel, Wet 'n' Wild, Essie, Maybelline, Revlon, Max Factor, L'Oreal, Street Wear, Bourjois, Lakme, Chambor, etc....???

[tool_462]

Do you use clear or coloured nail polish and what brand did you use - Rimmel, Wet 'n' Wild, Essie, Maybelline, Revlon, Max Factor, L'Oreal, Street Wear, Bourjois, Lakme, Chambor, etc....???

Stick with clear. Colors might (I don't know for sure) use something metallic or the dyes might carry current. I'm sure any brand is fine.

I use super glue myself, actually a pH neutral cyanoacrylate that is used for fuming finger prints that "expired" at work :nod:

Going on 35 chips sliced and diced and no issues! (I sell a service on eBay)

[docwhom]

I use super glue myself, actually a pH neutral cyanoacrylate that is used for fuming finger prints that "expired" at work :nod:

What a great idea. Superglue takes around 20 to 60 seconds to dry and it's clear. Thank you for letting me know about this valuable little chemical.

[tool_462]

What a great idea. Superglue takes around 20 to 60 seconds to dry and it's clear. Thank you for letting me know about this valuable little chemical.

I use one of the little CoolLaboratory brushes to paint it on, dries super fast!

[hyper_2002]

I do not get it... Do You remount IHS after applying Liquid Metal ? or mount the CPU without IHS ? Then what about mounting the cooler ? I mean the contact/pressure between the surface of core and cooler/WC block etc. :/

[komatsu]

before delid i was having 13c difference between the maximum and the lowest core after delid there is still 10c difference, is this normal?(clu on the die and back of the ihs, gelid extreme on the ihs)But overall i gained 20c

[Alex-Ro]

I do not get it... Do You remount IHS after applying Liquid Metal ? or mount the CPU without IHS ? Then what about mounting the cooler ? I mean the contact/pressure between the surface of core and cooler/WC block etc. :/

You put liquid(pro,ultra or other brands) between die and ihs,then socket should provide adecvate pressure,then u proceed on mounting water block or cooler on the ihs like normal.don't use liquid between ihs and water block for example,not worth it.

A 100% good job for air,water implies:

-remove ihs,clean the margins and die
-use liquid ultra/pro on the die
-isolate internal vrm just in case
-do a little lap job on the side of the ihs that sticks on die
-glue back the ihs using black silicone/adhesive,fix it using the socket so u got the optimal pressure

If you do a clean job warranty is not void since no one at intel RMA will cpu's to check if stock tim or not.

[vortex240]

Just in case anyone finds this interesting. The last transistor near the die on the 4670k came off along with the lid - seems the original glue took it with it. I'm referring to the one closes to the edge - at the same level as the top part of the die.

The cpu is working perfectly fine at 4.7, with normal stability. I was sure I killed it.. my delid only nettet 7 celcius difference. But this is with a crappy thermalright white paste.

[C-N]

Just in case anyone finds this interesting. The last transistor near the die on the 4670k came off along with the lid -

The cpu is working perfectly fine at 4.7, with normal stability. I was sure I killed it..

Did the same thing myself! http://www.xtremesystems.org/forums/...=1#post5195356

Chip is good for OCCT 4.8GHz, runs Super Pi 32M just fine with high BCLK & mem speed http://www.xtremesystems.org/forums/...=1#post5203831

The chip is now in my UD5H 24/7 workstation @ 4.5GHz & working well

[Alex-Ro]

I ended up screwing one transistor of one chip too about 3 months ago ,placed almost in the middle of the pack,cpu ran exactly the same after,same bench frequency,same prime stable,etc...

[vortex240]

How bizzare - well, I guess we got lucky. My original though was that it had to do something with the memory controller for channel 1. The one on my chip is dead from me trying to delid it with the vice method.

Turns out that Costa Rica chips, at least the one I have have really strong glue - might be a different type all together. So I ended up damaging the edge while gently hitting it and I did it properly, not a hack job. The top part of the surface flaked off completely, about 2-3 mm from the the edge, about half way of the side I was hitting.

When I popped it back in, it would not boot. So I tried on dimm slot at a time and it turns out, that channel 2(both slots) work perfectly fine, while if anything is inserted into channel 1 - no post.

So this is my little 'miracle' chip I guess. Few years from now, when I kill a chip, I can see myself saying "They sure don't build them like they used to"