i7 3770K needs de-lidding?

[BimmerCop 332]

Thanks Efrain,

 

Ive already got the H100 so when i get my 3770K i might look at doing this.

 

Cool, looking forward 2 it... im also buying sandpaper as i type this bc i will be lapping the h100's heatsink as well. I want to make sure i cover all my bases.

 

- Grammar *censored*, this was sent from my phone.

[BimmerCop 332]

Alright guys, I have de-lidded the 3770K, lapped the H100's cpu block and used TIM in this sequence: CPU DIE>>>MX-4>>>IHS>>>MX-4>>>H100

 

As a note for all H100 users, LAPPING IS A MUST!!!! The block has a horrific coarse finish to it and once you lap it, it will have a nice,mirror like finish, just like the cpu die has. I cannot stress this enough.

 

I just finished booting up the PC and at stock speeds, the 3770K is showing 30-31C at idle. I did not try the cpu in a lidded condition with stock cooler/fan, so I don't know if 30-31C at idle right now is a good number.

 

Anyone want to chime in?

 

BTW, here is a photo showing how the cooling block ended up after lapping it...

 

 

[BimmerCop 332]

Ok, I ran Prime95's in-place large FFTs torture test for 10 minutes and temps averaged 50C at stock settings.

[alainneedle1 97]

Have you lap the CPU also? Need to lap both for best result., CPU with a nice finish from the manuf. does not mean it's even (flat) all around, you'll be surprised to see the difference from a virgin CPU to a lapped one.

 

Good luck.

[BimmerCop 332]

Have you lap the CPU also? Need to lap both for best result., CPU with a nice finish from the manuf. does not mean it's even (flat) all around, you'll be surprised to see the difference from a virgin CPU to a lapped one.

 

Good luck.

 

I thought about it but just couldn't pull the trigger... Just handling the CPU gave me shivers! haha

[alainneedle1 97]

Since you de-lidded the CPU you know your warranty went down the drain so no harm done by lapping the CPU, anyway if you change your mind let us know..... :Chicken:

[BimmerCop 332]

Since you de-lidded the CPU you know your warranty went down the drain so no harm done by lapping the CPU, anyway if you change your mind let us know..... :Chicken:

 

Doesn't mean I now have to go and push my luck. You are free to try it yourself though,I won't hold you back. lol ;-)

 

BTW, I have it running at 4.8GHz and it's 25 minutes into Prime 95's in-place large FFTs stress test and it's been hovering in the upper 70s and lower 80s, depending on which part of the test is running.

 

4.8Ghz

vCore = 1.35v

CPU PLL = 1.925v

[TechguyMaxC 516]

Drop that PLL big time. You can probably get down around 1.6V with that OC. For the record I see a max. Of 80C in Prime 95 blend testing at 4.9GHz and 1.48V. That's with 2600 RAM speed so much higher IMC, Vdimm, and VTT also. I'll bet your temps would be a good deal lower with liquid metal.

[BimmerCop 332]

Drop that PLL big time. You can probably get down around 1.6V with that OC. For the record I see a max. Of 80C in Prime 95 blend testing at 4.9GHz and 1.48V. That's with 2600 RAM speed so much higher IMC, Vdimm, and VTT also. I'll bet your temps would be a good deal lower with liquid metal.

 

Is this it, Tech? http://www.amazon.com/Coollaboratory-Liquid-Thermal-Interface-Material/dp/B0039RY3MM/ref=sr_1_1?ie=UTF8&qid=1343881313&sr=8-1&keywords=coollaboratory+liquid+ultra

[BimmerCop 332]

Tech, I have ordered the Liquid Metal Ultra... BTW, I am almost stable at 4.9GHz, vCore 1.44v and PLL of just 1.5v! Temps in P95 blend have topped out in the upper 90s at it's most stressful point and low 90s/upper 80s in other parts of the test. I hope the Liquid Metal Ultra helps bring those temps down.

 

For comparison purposes, here are my voltages:

 

CPU CORE - 1.44v

CPU PLL - 1.5000v

VCCIO - 1.087v (VCCSA is set to 1.01250v since my P8Z77-V does not give me the option to control VCCIO directly)

DIMM - 1.56000v

 

With the above voltages, Prime95 is crashing about 40 minutes into the test. Can you share your settings?

[TechguyMaxC 516]

I'm running 1.48V Vcore, 1.25V VTT, 1.635V PLL, and 1.25V VCCsa. I run my RAM at 2600 C10 1T though, so my VTT and VCCsa are much higher than yours need to be, assuming you don't run such high mem speed/ tight timings. What is you running your RAM at? You can try Vcore a bit higher also.

[westman 257]

1.48v vcore 1.19vtt 1.67pll 1.185VCCAsa

Rams 9 11 10 26 1T 2666.

Offtopic tested 2 pair G-Skill 2800 , they was slower than mine

Think à get à Very good pair just lucky

 

 

[BimmerCop 332]

I'm running 1.48V Vcore, 1.25V VTT, 1.635V PLL, and 1.25V VCCsa. I run my RAM at 2600 C10 1T though, so my VTT and VCCsa are much higher than yours need to be, assuming you don't run such high mem speed/ tight timings. What is you running your RAM at? You can try Vcore a bit higher also.

 

I am running my RAM at 2200 9-10-10-24 1T... I can boot with 2400 but it's gonna take a lot of messing around with the timings and voltage to get it stable and honestly, the difference between 2200 and 2400 is so little that right now, I don't think it's worth the hassle. Besides, this memory is rated at 1600, so running it 100% stable at 2200 with such tight timings is a feat in itself. lol

 

I was able to run Prime95 blend this morning for 4.5 hours without any errors whatsover! I stopped it because I was getting bored of just sitting around doing nothing. hahaha

 

Here are my settings as I type this:

 

CPU Speed: 4.9Ghz (100 x 49)

vCore: 1.46v

VTT: 1.01250v

CPU PLL: 1.50000v

DRAM: 1.56000v

VCCSA: 1.01250v

 

Under full load of Prime95 blend test for 4 hours, temperatures were in the upper 80s C to lower 90s C and two cores peaked at one time at 100 C. I am surprised the system remained stable during that torture for 4 hours! I can't wait for the Liquid Metal Ultra to arrive so I can see if I can drop the load temps to the ballpark of maybe lower 70s C. I might, just might, if that is the case, to squeeze another 100Mhz out of this sucker and hit the 5Ghz mark. Let's see what happens!

[shally17 1]

Please excuse my ignorance but does lapping simply mean sanding back to a smooth finish? How coarse should the sandpaper be?

[BimmerCop 332]

Please excuse my ignorance but does lapping simply mean sanding back to a smooth finish? How coarse should the sandpaper be?

 

Yes, it's sanding down until you have a perfectly flat, smooth mirror-like surface. The most important thing is not so much the smoothness, but the flatness. For this reason we usually use a piece of glass as the surface on which we lap the heatsink.

 

When I do my lapping, I use this sequence in coarseness:

 

180

220

320

400

600

800

1000

2000

3000

 

Some people might say it's too many but I always end up with a mirror-like surface and works wonderfully for me.

 

- Grammar *censored*, this was sent from my phone.