3770k IHS Removals - CPU temp dropped from 79C to 71C

[SonDa5]

Updating temp post.

My IC Diamond seems to be cooling better after a few days.

Also I have been able to tweak my voltage a little more than before for stability.

Under load at 4.5 GHZ my 3570k is stable with 1.13v.
http://valid.canardpc.com/show_oc.php?id=2537884

10 pass in diagnostic mode with maximum memory Intel Burn Test
Ambient Temps 24C.
Temps under IBT 63C 72C 71C 66C




My temps at 4.5 GHZ have now lowered to 68C average per core. That is a 19.25C per core average temperature drop so far since I de-lidded.

I didn't think IC Diamond had a break in period. The only things I have done is lower my CPU voltage from 1.17v to 1.15v in BIOS and I have raised my PLL voltage from 1.5v to 1.60v and lastly I hand tightened my cpu mount set screws just a little where they felt lose. I'm being as careful as possible to not crack the die with the block. (NO IHS)

As I tweak it for lower temps I will update with changes.

[ChaosAD]

So i guess folding at 4.22ghz with 1.12v and temps between 65-69c (24c ambient), is not an issue with my water setup or a bad mount, but an urgent need to delid?

Sent from my GT-I9300 using Tapatalk 2

[Conquistador SW]

So i guess folding at 4.22ghz with 1.12v and temps between 65-69c (24c ambient), is not an issue with my water setup or a bad mount, but an urgent need to delid?

Sent from my GT-I9300 using Tapatalk 2

Very likely. I tried 4.7 with 1.32v but even with my watercooling it would go over 100C when folding, so 70C with your settings looks possible if you got a particularly bad chip (temperature wise).

[sonnet]

Can someone pls explain few basic things?

The IHS is removed and the thermal paste between the IHS and the CPU core replaced (my inderstanding is that collaboratory liquid pro is above the others in performances, although it's not clear to me how much).
I get that.
What is unclear to me, is if you guys after having replaced the thermal paste, you re-attach the IHS to the CPU (with the help of some glue , in that case which kind of glue would be best?) or you just place it above it and keep it fixed with
the pressure of the heatsink.
Can anyone clarify his to me, pls?

[Zeus]

Most people tend to use the clamp down mechanism that is on the board. So just place CPU without IHS in socket, place IHS on top, align IHS and close clamping mechanism. On top of that you just place your waterblock or heatsink.

Personally i've used some rubbery glue and placed the IHS back on. This makes replacing the CPU a lot more convenient.

[Conquistador SW]

Can someone pls explain few basic things?

The IHS is removed and the thermal paste between the IHS and the CPU core replaced (my inderstanding is that collaboratory liquid pro is above the others in performances, although it's not clear to me how much).
I get that.
What is unclear to me, is if you guys after having replaced the thermal paste, you re-attach the IHS to the CPU (with the help of some glue , in that case which kind of glue would be best?) or you just place it above it and keep it fixed with
the pressure of the heatsink.
Can anyone clarify his to me, pls?

Liquid Pro is indeed one of the best (probably the best, but there are no comparison tests done yet). The difference between Liquid Pro and for example MX4 or AS5 can be quiet large (more than 10C), especially if your overclock it somewhat high.

You can use some glue, but I personally just closed the socket and that holds everything in place. Once the socket is closed, it will not slide around anymore, just make sure that while closing the socket everything is aligned properly.

[andressergio]

Liquid Pro is indeed one of the best (probably the best, but there are no comparison tests done yet). The difference between Liquid Pro and for example MX4 or AS5 can be quiet large (more than 10C), especially if your overclock it somewhat high.

You can use some glue, but I personally just closed the socket and that holds everything in place. Once the socket is closed, it will not slide around anymore, just make sure that while closing the socket everything is aligned properly.

i keep saying this and will sustain it...i changed pate to 6 chips and used MX2 and ALWAYS temps went 10 or more C low...i dont know if you are applying too much or what but for worked as a charm always

[SonDa5]

I just got done doing my first 27 hour Prime95 v27.7 with AVX long FFTs (hot selection).


Ambient temps ranged from low 20C to high 27C.
3570k@4.5GHZ 1.15v.
TIM IC Diamond 24 De-Lidded NO IHS



Hottest temps were recorded during the 15th hour of testing.

Max Temps recorded

64C 72C 69C 64C

[st0ned]

Can someone pls explain few basic things?

The IHS is removed and the thermal paste between the IHS and the CPU core replaced (my inderstanding is that collaboratory liquid pro is above the others in performances, although it's not clear to me how much).
I get that.
What is unclear to me, is if you guys after having replaced the thermal paste, you re-attach the IHS to the CPU (with the help of some glue , in that case which kind of glue would be best?) or you just place it above it and keep it fixed with
the pressure of the heatsink.
Can anyone clarify his to me, pls?

Liquid metal TIMs, both collaborators lp and phobya Lm are above all others in my opinio for two reasons. 1st of all they won't dry as normal Tim does given they are made out of a metal alloy which is liquid at ambient temperature, this is a big 'pro' . 2nd they have the highest heat transfer coefficient then any of other Tim this gives you the best heat moving per area so whilst transferring heat from the IHS to the cooler base you are dealing with a 'big' area here, Die -> IHS you are dealing with a small area thus making such differences becoming bigger, in the end they will withheld better temps. I would say 10°C over normal tims.

i keep saying this and will sustain it...i changed pate to 6 chips and used MX2 and ALWAYS temps went 10 or more C low...i dont know if you are applying too much or what but for worked as a charm always

Mx2 is obviously better then any Tim intel used there but it is much worst then liquid Tim's for this propose.

[SonDa5]

Liquid Pro between DIE and Copper block will tarnish the copper block is what I have read. This is when there is no IHS in the mount.

[hersounds]

Today the patient was admitted to the operating room


1.jpg

2.jpg

3.jpg

5.jpg

that was easy!


Before test

IHS STOCK INTEL.jpg

After test ( same test as before )

IHS OFF.jpg

Before 4.3 Ghz - 1.13 vcore

Core 1 : 93°C
Core 2: 100 °C
Core 3: 105 °C
Core 4 :101 °C

After 4.3Ghz - 1.13 vcore

Core 1: 62 °C
Core 2: 67 °C
Core 3: 66 °C
Core 4: 61 °C

the droop was
-31°C
-33°C
-39°C
-40°C

successful surgery!

Phobya LM was the way

[Zeus]

Awesome result Hersounds!! Told you it was easy!

[hersounds]

Awesome result Hersounds!! Told you it was easy!

yes, yes, I'm surprised because I did not expect too improve buddy!

[Conquistador SW]

i keep saying this and will sustain it...i changed pate to 6 chips and used MX2 and ALWAYS temps went 10 or more C low...i dont know if you are applying too much or what but for worked as a charm always

Any TIM will do much better than the Intel stock TIM, but if you were to use Liquid Pro, you would likely see another 10C drop.

Look at this: http://foro.noticias3d.com/vbulletin...d.php?t=391334 (in Spanish, but the graphs say it all).
The difference between AS5 and Liquid Pro is huge; more than 10C when overclocked.

As far as I have seen everybody that used a liquid metal TIM got a somewhere between 20C an 30C drop in temperatures (or even more on the hottest cores), while people using other pastes see a drop around 10C to 15C.

Before 4.3 Ghz - 1.13 vcore

Core 1 : 93°C
Core 2: 100 °C
Core 3: 105 °C
Core 4 :101 °C

After 4.3Ghz - 1.13 vcore

Core 1: 62 °C
Core 2: 67 °C
Core 3: 66 °C
Core 4: 61 °C

the droop was
-31°C
-33°C
-39°C
-40°C

successful surgery!

Phobya LM was the way

Looks like we can add Phobya liquid metal to the TIM's recommended for the best results
Very nice job hersounds

[ChaosAD]

Thats a great improvement. Did you use LM also between cpu and hsf? I think its time to order some LM for some weekend action

Sent from my GT-I9300 using Tapatalk 2

[SonDa5]

Awesome temps hersounds. Congrats.

[hersounds]

thanks guys!

Thats a great improvement. Did you use LM also between cpu and hsf? I think its time to order some LM for some weekend action

Sent from my GT-I9300 using Tapatalk 2

yes i did but only a little very little touch of phobya LM


Update the temps continue to drop as the hours pass ... like all metal paste ( liquid or not ) should seat

Iddle 4.6 Ghz - 1.22 vcore



Full load 4.6 Ghz - 1.22 vcore



Bonus searching Vcore for 5.0 Ghz ( i think 1.48 is too much but i don't have time to testing right now)

[ZoLKoRn]

Wow !!! 40C droop seem like Phobya LM ways to go and time for i'll lost some money again

Thx for sharing bro it helpful for ways to choose

[hersounds]

Thanks to you bro...and yes after read much forum pages, i see and i decided for phobya LM like best choice to put between Die & ihs... please becareful with aluminum pieces / base this is gallium...

...gallium eat aluminum

[MaynardX]

Hersounds

I use exactly the same 1.488 vcore for my crappy 3570 (unfortunately it hits the 4.9 wall), but in my case I see 74 Celsius while using Intel Burn Test v2.54. I applied Liquid Ultra on the die end the IHS.
Could you please verify your results with 1.488 vcore using IBT?

[hersounds]

Hersounds

I use exactly the same 1.488 vcore for my crappy 3570 (unfortunately it hits the 4.9 wall), but in my case I see 74 Celsius while using Intel Burn Test v2.54. I applied Liquid Ultra on the die end the IHS.
Could you please verify your results with 1.488 vcore using IBT?

Sure buddy what do you need exactly ? my 3770k @ 4.9 Ghz -1.48 vcore? and then run sp32M & intel burn test?...just tell me and I do it

[MaynardX]

Hersounds

Yes, that is exactly what Im asking for. Just set 1.488 vcore all other volts in bios default (mine are like that) and run IBT. That should do the trick.
Sameone correct me if Im wrong: Im not sure if 3770K produces same amount of heat vs. 3570K considering same voltage given. Both used to have the same TDP which is 77W though.

[hersounds]

Hersounds

Yes, that is exactly what Im asking for. Just set 1.488 vcore all other volts in bios default (mine are like that) and run IBT. That should do the trick.
Sameone correct me if Im wrong: Im not sure if 3770K produces same amount of heat vs. 3570K considering same voltage given. Both used to have the same TDP which is 77W though.

Ok understood! I will try later...Remember aditional cache in 3770k and that i'm using HD4000 Igpu

[MaynardX]

Hersounds

Affirm, multigpu enabled here too.
I think CPU die size matters more thank the cache itself?
But good point mate you mentioned this and I think you started good discussion here. What really matters considering heat emission with given vcore? We know both CPUs has same die size, same TDP.
I'd appreciate your comments.

[hersounds]

Also remember that each chip is a different world ... even sharing the same batch, minimal changes as climate, city and region of the world is affecting the chip a positive or negative way.

the tests are only comparative way but mean nothing, because real environment of a pc in Buenos Aires not be the same as another in Alaska.

for 4.9 my machine needs unless vcore of 1.48 But if you still want to try, please wait a few days when I have free time i'm update this thread