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

[mastabog]

Initially I removed the IHS on my 3770K (batch L221A999) and replaced the TIM with MX-4 between the core and IHS and also between the IHS and the heatsink -- an almost 2 year old Thermalright Venomous X.

My max Prime95 temps for 4.7 GHz @ 1.30v, ambient 26C were 91C on the hottest core. Quite disappointing.

Then I bit the bullet and purchased and applied that Coollaboratory Liquid Ultra stuff that people kept raving about. I'm stunned.

Hottest core temp under Prime95 decreased from 91C to 75C (same 4.7 GHz @ 1.30v, ambient 26C) ... that's a freaking 16C drop just from changing the thermal paste! (I applied it again both between the core and the IHS and between the IHS and the heatsink).

I just upped vcore to 1.50v and clocked to 5.0 GHz and wPrime 1024M with 8 threads ... max temps are 87C with an ambient of 26.3C. This is absolutely great.

One reason the temps dropped so much might be that Thermalright builds (or used to build) the heatsink base with a small bump in the center, in an attempt to draw heat from the center of the CPU at an accelerated rate. See a pic here:

http://www.xbitlabs.com/misc/picture...x_qrov_big.jpg

Maybe due to that, the MX-4 was thicker on the outside of the CPU, maybe withholding heat. This Coollaboratory stuff being effectively metal, conducts heat much better than the MX-4 when the layer is thicker.

Another reason might be the plating itself. It's nickel plated but ultra shiny and flat -- no sign of lapping whatsover, see here:

http://www.xbitlabs.com/misc/picture..._q_big.jpg&1=1

This cooler also has a screw in the center which, when tightened, applies a lot of pressure. It may just be that with so much pressure, the thicker MX-4 just weeps out of the center (I did notice when removing the heatsink that in the center there were spots n both the cpu and the heatsink where the metal was directly visible).

Maybe the above helps others. In any case, I'm quite happy as you can imagine. Getting 5.0 GHz on air with an old cooler made me grin big time

[mastabog]

I also had removed the IHS and used MX-4 on my 3770K. I just replaced the MX-4 with Coollaboratory Liquid Ultra between the core and IHS and also between the IHS and the heatsink and my hottest core dropped temps from 91C to 75C ... this is crazy. I'm at 4.7 GHz @ 1.30v vcore with 26C ambient, which is fairly high.

I tried 5 GHz but I found I need 1.50v vcore to be wPrime stable with 87C ... prime95 fails within minutes with temps over 94C . Granted, I'm using air cooling with an older Thermalright Venomous X which has that bump in the center of the base which may affect things and could be why the temps dropped so much when using CLU instead of MX-4.

My 3770K has batch no L221A999.

I'm seeing some crazy low voltages and temps in this thread for those using water cooling.

Are the above temps and voltages normal for air? Do you think using water cooling will improve things dramatically?

I'm still on air is because I want the system silent at idle, when my fans spin at 500-700 RPM. I tried the Corsair H100 a while ago and I hated the pump motor noise. Are there any pumps that are as silent as a fan at 700 RPM?

[Daveangel]

Guys

Liquid pro seems to be the favourite for under the ihs. As its good for this why doesn't any one seem to use it between the cooler and ihs? I imagine that gap to be rather small also? So is some other Tim better for that application

I have just covered all my machine in diamond Ic which seems to get slated in this thread??

Any help would be appreciated

[owikh84]

3770K @ 5.0GHz 1.416v

bios set: 1.415v LLC very high

prime95 27.7 needs 0.064v more vCore :stars:
custom blend min FFT 128k

3770K @ 5.0GHz 1.48v

bios set: 1.47v LLC very high

[Church]

Daveangel: most of common users don't use liquid pro between base of cooler & IHS because it has tendency of diffusing into surfaces, thus after some time cleaning it might mean sanding/lapping IHS down with loosing cpu's markings/pristine looks (thus maybe harder to resell). And compared to liquid ultra, pro might be a bit harder to apply. If you don't care about both, go ahead and feel free to use it.

[s3rv3r]

DSCN0426.jpg

[dr_dx]

Guys

Liquid pro seems to be the favourite for under the ihs. As its good for this why doesn't any one seem to use it between the cooler and ihs? I imagine that gap to be rather small also? So is some other Tim better for that application

I have just covered all my machine in diamond Ic which seems to get slated in this thread??

Any help would be appreciated

That's what I did (liquid Pro on all 4 surfaces) and had an average -26C drop. See my previous post #323 in this thread.

[MikeMurphy]

Make sure and "wet" both/all surfaces.

What exactly does this mean? Thanks

[dr_dx]

What exactly does this mean? Thanks

Instead of just applying a small drop to the center of the die or underside of the IHS, you actually want to spread out the drop (very small) on both surfaces to a mirror like finish, then put the 2 together. In my case, I did the same for the top of the IHS (larger surface) and the bottom of my waterblock - hence all 4 surfaces. If you go to Colaboratory's web site, they have video's on how to do it. If you are using Pro, use a very tiny drop; a little goes a long ways.

[Xero99]

OK, first let me say that since the 3770k was released I have been saying that Intel was the biggest douches for allowing such a poor excuse of a processor to be released!
Last week a buddy asked me to build him a new comp with this chip and then asked me to do this mod with the liquid Pro! I asked him was he out of his mind but figured why not.

Well slap me on the ass and call me Susan cause after I finishing his I drove to Micro center and bought myself one. Just finished this mod on MY new chip and its registering...what for it...42c max on Prime95 in asus suite and max core of 64c in Real Temp!!!

Before the mod: 98C Real Temp @ 4.7 GHz @ 1.29v, ambient 31C On Water
After the mod: 64c Real Temp @ 4.7 GHz @ 1.29v, ambient 31c On Water

34 DEGREES DROP!!!

I followed the above post and applied it on all four sides and intentionally left a extremely tint bit more on the center of both the IHS and the die and the top of the IHS and the block!

Thanks to all of you on the ground work on this!!! Now I'm off to push for 5.0 GHz.
And I still haven't changed my mind about Intel recent decisions!

[Church]

IB works within specs and is better at stock frequencies, especially in laptops. Enthusiasts (probably <<10% from all IBs sold) simply expected more, but didn't make business case to fix things with TIM1 and if alternative would be rising for by $5 for all the IB owners due better TIM used .. i somewhat understand intel's choice. A bit hard to get such bashing and calling names for intel for it's business choice. It's not exactly end of the world, isn't it?

[Sniper]

They do charge quite a premium for the K cpu's, maybe spend some of that on "better TIM" ?

[Church]

Not THAT big of a premium (@newegg 30$ (though 3770 is a bit slower clocked too)), and very probably IHS is put on same line for all. If you sell tens of millions cpus, even meager 5$ saving is enormous. Anyway, yes, when heavily overclocking, IB's cheap/bad stock TIM1 takes its toll, but mod to replace it is simple/affordable enough for me to not care and not start holy wars on subject

[cab]

Not THAT big of a premium (@newegg 30$ (though 3770 is a bit slower clocked too)), and very probably IHS is put on same line for all. If you sell tens of millions cpus, even meager 5$ saving is enormous. Anyway, yes, when heavily overclocking, IB's cheap/bad stock TIM1 takes its toll, but mod to replace it is simple/affordable enough for me to not care and not start holy wars on subject

Ivy Bridge being 22nm technology should run much cooler than my 45nm Phenom II, but in fact my 3570k runs much hotter. That's why I think that Ivy Bridge desktop package is defective. Maybe you enjoy doing mods, but most customers don't want to be unpaid repairmen.

[dqniel]

Ivy Bridge being 22nm technology should run much cooler than my 45nm Phenom II, but in fact my 3570k runs much hotter. That's why I think that Ivy Bridge desktop package is defective. Maybe you enjoy doing mods, but most customers don't want to be unpaid repairmen.

Less die area with which to transfer the waste heat. So, while total heat released should be lower that doesn't mean the temperature should be lower.

That said, Intel still made a mistake by going the cheap TIM route rather than soldering to the IHS.

[ZeroOne]

Just reporting in....

I almost can't believe what I'm seeing.

Originally, 4.6Ghz at 1.275v was topping out at 90C yesterday, stock 3770K with a Raystorm and 27C ambient.

Did the following:
Lapped IHS first
Removed IHS
Cleaned IHS of black adhesive and old paste.
Lapped under side (only the rim where it contacted the green PCB)
Cleaned black adhesive off PCB and old paste off CPU.
Applied Liquid Pro to CPU and VERY thinly to underside of heatspreader (contact area only)
Clamped it down using the force of the socket to secure the IHS, and used PK-3 under Raystorm

The results? Same room temp, same OC.... it went from 90C to 63C!!! Thats unreal!! I expected not quite that drastic an improvement!! 2 of the cores haven't even left the 50s!! That was WELL worth the 20 minutes of "what did i just do to $330 bucks!?"

Seriously though.... this is a completely different processor now. I keep looking at my OC notes, seeing that 90C written down, and rechecking the voltage and speed to make sure I didn't change anything.

[dqniel]

Lapped under side (only the rim where it contacted the green PCB)

What was the purpose behind this? (not a rhetorical question to be a jerk; I really am curious as to the reason)

[Xero99]

ZeroOne is eight on! After the mod it really is a different chip! I'm still running Prime 95 @ 4.9GHz using 1.325v and the hottest core is at 59c!!
This is fricking unreal!!

If 5.0 puts me over 1.4v then I'm gonna leave it at 4.9 especially at these temps!

[ZeroOne]

What was the purpose behind this? (not a rhetorical question to be a jerk; I really am curious as to the reason)

When I put the IHS back on, after removing the black stuff.... it was SO close, I couldn't tell if it was completely not-touching the green PCB. I wanted to take off the slightest layer to be sure the underside of the IHS was evenly touching the die with no interference whatsoever. It may not have been needed, but I felt better being able to see the light of day between the two before reinstalling.

[Zeus]

Really?

I've done two 3770K's now and both had the heatspreader clearly resting on top of the die. Like i said earlier, both had the heatspreader "floating" on top of the die even though the gap between the heatspreader and the pcb is very small.
Your results speak for themselfs so maybe it was needed in your case.

[ZeroOne]

Really?

I've done two 3770K's now and both had the heatspreader clearly resting on top of the die. Like i said earlier, both had the heatspreader "floating" on top of the die even though the gap between the heatspreader and the pcb is very small.
Your results speak for themselfs so maybe it was needed in your case.

Its possible it wasn't needed. Just being overkill I guess. The gap you had between, could you see light through? It felt like it was floating on the die, but i wasn't 100% sure it was doing so evenly, all the way around. I don't have the sharpest finger nails to get that black stuff off either.

[dqniel]

Its possible it wasn't needed. Just being overkill I guess. The gap you had between, could you see light through? It felt like it was floating on the die, but i wasn't 100% sure it was doing so evenly, all the way around. I don't have the sharpest finger nails to get that black stuff off either.

OK. This makes sense. I tested to see if it was resting completely flat on the die by spinning the IHS (without paste) on the die. It spun freely without any tilting or rattling so I assumed the contact was smooth and flat

[Zeus]

I haven't held the cpu against the light but i'm pretty sure there would have been a small ray of light coming through if i did.

[mean6]

Hey everyone. Just wanted to share my result from de-lidding my IHS off of my 3770K. I am currently running it @ 4.5 with 1.32v. using a corsair H100 watercooling unit. When running Linx temps got really hot. I wish I had of taken screenshots of before and after, but I did however write down the temps using coretemp.

So, after running 10 mins of Linx, my temps were as follows

Core0: 83
Core1: 97
Core2: 96
Core3: 90

also note, I was using MX-4 between the IHS, and the H100

After removing the IHS, and using Liquid Pro between the die, and the IHS, and the H100, running Linx for 10 mins using the same overclock and volts were as follows:

Core0: 55
Core1: 71
Core2: 75
Core3: 68

It actually averages almost 23 degrees cooler overall. I can't believe the difference. Anyways, I just wanted to post my findings. Now off to overclock this b****h! lmao

[Xero99]

Ok I just finished a 22 hour Prime95 marathon @ 5.0Ghz!!

Max temp in Asus Suite 46c
Max temp in RealTemp 81c
Ambient temp 32c

All on Water!!!
I have now modded 8 chips for various friends (do at your own risk!!!) and ill just post the worse of those numbers.

Chip #6 (Costa Rica) @ 4.8Ghz
Max temp in Gigabyte 61c
Max temp in RealTemp 92c
Ambient unknown


Now for on my chip 5.1Ghz!

Again thanks to all posters on this thread!!!! Your Xtreme-ness makes me giddy!!!