4770k /4670K IHS Removals

[C-N]

These Haswell temps are nuts... might have to de-lid my 4770K on the weekend.

Here's what its making now!!!! & sure I'll update if/when I de-lid



IHS pop update using Liquid Ultra on 2 sides & PK1 between IHS & Kuhler 920 (21oC Ambient)

[andressergio]

*hernan

u delided ur chip ?

[andressergio]

These Haswell temps are nuts... might have to de-lid my 4770K on the weekend.

Here's what its making now!!!! & sure I'll update if/when I de-lid

nice CN i think that deliding it u will be on 30C less for sure just take care of the circuits near the die but use liquid pro to get those 30C down

[hersounds]

*hernan

u delided ur chip ?

Not yet my friend soon! I don't have time

[Rob94hawk]

Before I go tearing into the cpu, what's the longevity of the liquid pro? Especially since putting it directly on the die is permanent. Is it going to start losing it's thermal transfer abilities after a few months?

[Rob94hawk]

What would be the next best TIM, and one that is prferably not electrically conductive?

Got this from anandtech forums who was just using a naked die and a water block. Apparently NT-H1 was the second best choice after liquid ultra but there was a concern about pump-out.

http://forums.anandtech.com/showpost...3&postcount=67

[andressergio]

Before I go tearing into the cpu, what's the longevity of the liquid pro? Especially since putting it directly on the die is permanent. Is it going to start losing it's thermal transfer abilities after a few months?

well i tried all mate and liquid PRO NOT ULTRA is the best, my CPU its at 5GHz HT at 1,5 load since december

[Dumo]

The socket retention bracket didn't end up shipping with the M6E, I'm trying to get some built, is anyone willing to sell / loan one so I can have dimension taken.

Thanks!

Heres nachtfalke using it http://www.xtremesystems.org/forums/...=1#post5194141

[Vivi]

i went from 80C load to 63C load only using GC-EXTREME. cant wait for liquid pro to arrive

[andressergio]

i went from 80C load to 63C load only using GC-EXTREME. cant wait for liquid pro to arrive

Very nice vivi i would keep it that way

[Vivi]

lost about 30c using liquid pro

using corsair H60

[andressergio]

lost about 30c using liquid pro

using corsair H60

told ya same me but on IVY

what did you put on the circuit part to cover it ?

[Rob94hawk]

what did you put on the circuit part to cover it ?

I don't think anyone's putting anything on it to cover it. Why would you need to if you're putting so little on then it hardens?

[andressergio]

I don't think anyone's putting anything on it to cover it. Why would you need to if you're putting so little on then it hardens?

cauz liquid pro is metal and VERY conductive

[Alex-Ro]

thats why you need to spread only a little dot on die,then spread it,if you use the proper amount should be ok.

[Vivi]

what did you put on the circuit part to cover it ?

nail polish :P. and yes its VERY necessary. liquid metal is very leaky and those resistors are tiny and right next to it.

here is some results at 5GHZ 4c 8t. Previously this cpu would not even get close to this and hit 90c, now its barely at 70c

corsair H60 and 18C ambient

5GHZ VANTAGE


5GHZ WPRIME 32M

[Phaseshift]

so will delidding allow for higher clocks? I mean like right now I can boot at 47x on 1.30v rune cinebench and all is gravy. But will it allow me to boot into 48x 49x 50x? I guess what I'm trying to say is, if I can boot into 50x multi at 1.35v but isn't stable will delidding it give it a chance to be stable?

[Vivi]

so will delidding allow for higher clocks? I mean like right now I can boot at 47x on 1.30v rune cinebench and all is gravy. But will it allow me to boot into 48x 49x 50x? I guess what I'm trying to say is, if I can boot into 50x multi at 1.35v but isn't stable will delidding it give it a chance to be stable?

yes, haswell is very stable at IDLE at almost any crazy speed, then when it loads it heats up like madness. so if you are almost stable at 5ghz, and can run half a pi or start a bench and fail. then there is a good chance delidding will allow you to run 5g

[Phaseshift]

yes, haswell is very stable at IDLE at almost any crazy speed, then when it loads it heats up like madness. so if you are almost stable at 5ghz, and can run half a pi or start a bench and fail. then there is a good chance delidding will allow you to run 5g

hmm alrighty then, I need a razor blade lol

[Demo]

Isnt the VRM gonna be facing on the right/left side when the processor is installed on the socket?

[acebmxer]

Isnt the VRM gonna be facing on the right/left side when the processor is installed on the socket?

Left side of cores

[C-N]

Anybody got a clue as to what all the resistors do on the 4770K PCB?

I just broke one off rubbing my fingers over them whilst cleaning the PCB black stuff up!!!

Its the very bottom one in this picture


I quickly tested the chip in the board & it booted so I carried on & applied nail varnish to the resistors liquid ultra to the die & under side of IHS. I also glued the IHS back on with some black mastic.

The chip is now in a board & appears to be ok but I'd still like to work out what the resistor was for.

[Demo]

Did you have to delid to get 4.5Ghz stable?

[C-N]

Did you have to delid to get 4.5Ghz stable?

No it was stable just hot see post 51 at the top of this page.

Its now 4.7 stable & still about 20oC cooler than 4.5.

I have played about with 4.8 & it runs OCCT at about 1.35v however the rig reboots after 10 minutes....

[Kallenator]

The chip is now in a board & appears to be ok but I'd still like to work out what the resistor was for.

It's probably a capacitor, not a resistor. Resistors usually have markings on top of them that display their resistive value, ceramic capacitor on the other hand rarely have markings on them.
Depending on the size it's either a 0402 or 0201, its hard to tell.
My guess is that these caps are used for some kind of decoupling.

0201 (0603 metric): 0.6 mm ? 0.3 mm
0402 (1005 metric): 1.0 mm ? 0.5 mm

I have doubts that you will be able to solder this component back on, so I would put efforts into testing the CPU's integrity through stress-test programs.