My 4770k delid video (IHS removal)

[Boulard83]

This is no slant against your video, but I wish there was more of a political message against Intel's strategy from us, the consumers in the trenches.

For the second generation Intel decides to devalue their own product, at the cost of terrible thermals and resulting higher power consumption, completely opposite to what they preach.

From under a tinfoil hat it might even look like intel wants to bake/fry their chips to make us have to upgrade in 4 years or so, especially since miniaturisation is about to come to a halt, and there is less relative profit in more silicon and more cores.

Yet the delidders seem to be perfectly content and happy, now having an ability to make a difference, a purpose, something to do with this magical black box of technology. I can respect the experiment, but really this position is kind of of conflicted.
I should have not trusted Anand's "Just wait for Haswell" hype, should have bought Vishera when it was released.

I honestly thought we'd get proper solder after the Ivy Bridge debacle, it would seem we need more of an outcry, boycott and more de-lidding coverage. Even though the best way to de-lid, is not to buy A CPU like that in the first place.

I agree, as fro delid, it's all up to you/all Ivy/haswell chip buyer.

Intel are "sadly" playing the usual corporation game wich is increasing profits.

 

[MisterMac]

You'll see Ivy-E being soldered.


It's another way of trying to force people onto their server tech lines.
In the long run it seems like a better idea to put us there.


Mainstream is not increasing absolute perf - only E\EP\EX lines will do this from now on.

But it will probably be a long while before we get some better price ratios - gotta move the top echelon of performance mainstream into the high end SKUs to drive volume before you lower the prices.

Atleast that's how i see it - short of some technical validation issue they found with solder they didn't wanna risk on TriGate.

(Should check IVY delidders in a year or two - if they're still good).

 

[thilanliyan]

Lol that does look like a really good way of slicing your finger.

But thanks OP for sharing your results.

 

[Boulard83]

You'll see Ivy-E being soldered.


It's another way of trying to force people onto their server tech lines.
In the long run it seems like a better idea to put us there.


Mainstream is not increasing absolute perf - only E\EP\EX lines will do this from now on.

But it will probably be a long while before we get some better price ratios - gotta move the top echelon of performance mainstream into the high end SKUs to drive volume before you lower the prices.

Atleast that's how i see it - short of some technical validation issue they found with solder they didn't wanna risk on TriGate.

(Should check IVY delidders in a year or two - if they're still good).

Saldy, this is true.

Lol that does look like a really good way of slicing your finger.

But thanks OP for sharing your results.

Thanks

 

[Boulard83]

First paste testing.
Noctua NH-1 versus EVGA Frostbite.

Well, the frostbite suck .... loll !
I had good hope on the frostbite paste since it seems to spread well under pressure but it did not deliver good results.


Edit, i also prepared the CPU for the incoming CLP and CLU "TIM".

EDIT 2 : I just redid the Frostbite application but my first ony appeared to be perfect. I had similar result on second application, quickly peaked over 80°c on AIDA. EVGA Frostbite seems is bad at this job...

 

[Boulard83]

Got something in the mail today

 

[cytg111]

Thanks for the video. Also waiting to see if this brings any extra overclocking headroom.

That pic just made my day

- and eagerly awaiting your improved OC results

 

[CrazyElf]

You'll see Ivy-E being soldered.


It's another way of trying to force people onto their server tech lines.
In the long run it seems like a better idea to put us there.


Mainstream is not increasing absolute perf - only E\EP\EX lines will do this from now on.

We already have confirmation that it is a solder on the Ivy Bridge E lines.

http://www.guru3d.com/news_story/intel_ivy_bridge_e_has_solder_under_its_ihs.html

At this point, the only real question is, how well does Ivy Bridge E overclock?

Also, with the VR integrated, how well will Haswell E overclock? Finally, will the $600 CPU get an 8 core or a 6 core (kind of like what happened with the 980X)?

 

[Aikouka]

I tried de-lidding my 4770k about a week ago using the vice method, but I had a rather hell of a time trying to get it to work. Maybe I was just being a bit squeamish about over-tightening the vise, but the chip just kept popping out. I might need to use a better piece of wood.

I thought about using the razor method, but after seeing just how thin the epoxy layer was, I was a bit worried about attempting it! I might tear apart my other HTPC that has an i3-3225, which should also use TIM instead of solder. The way I see it is... if I ruin my i3, that's $130 vs. $330 for the i7.

 

[Zodiark1593]

I honestly thought we'd get proper solder after the Ivy Bridge debacle, it would seem we need more of an outcry, boycott and more de-lidding coverage. Even though the best way to de-lid, is not to buy A CPU like that in the first place.

Here is where we hit a little snag. A boycott would most likely not be very effective given the severe lack of competition. The only competition Intel really has (in the PC segment) is against their older chips. Even if Intel is publicly shown to be committing mass genocide against kittens. they are still the performance king. :\

 

[Idontcare]

I tried de-lidding my 4770k about a week ago using the vice method, but I had a rather hell of a time trying to get it to work. Maybe I was just being a bit squeamish about over-tightening the vise, but the chip just kept popping out. I might need to use a better piece of wood.

I thought about using the razor method, but after seeing just how thin the epoxy layer was, I was a bit worried about attempting it! I might tear apart my other HTPC that has an i3-3225, which should also use TIM instead of solder. The way I see it is... if I ruin my i3, that's $130 vs. $330 for the i7.

At least with the razor you can have complete control over what you cut and how you cut it (just don't try to use the razor to pry the IHS off, you'll cut into the PCB doing that, only use the razor to cut through the adhesive and for nothing more)...whereas with the vise method you are really putting a lot of force into the PCB itself and if you delaminate those thin layers in the PCB your CPU will be compromised.

Cutting adhesive is fine, your CPU doesn't critical depend on the adhesive, but slamming the PCB and risking damaging it seems like a quick way to have problems IMO.

 

[Aikouka]

At least with the razor you can have complete control over what you cut and how you cut it (just don't try to use the razor to pry the IHS off, you'll cut into the PCB doing that, only use the razor to cut through the adhesive and for nothing more)...whereas with the vise method you are really putting a lot of force into the PCB itself and if you delaminate those thin layers in the PCB your CPU will be compromised.

Cutting adhesive is fine, your CPU doesn't critical depend on the adhesive, but slamming the PCB and risking damaging it seems like a quick way to have problems IMO.

I'm not worried so much about the actual cutting of adhesive, but rather the amount of adhesive on my particular CPU seems rather thin. I was sitting there holding it in my hand and wondering, "How in the world would I get a razor under that without damaging it?" I might just be over-thinking it a bit. Well, actually... if history is any indication, I am over-thinking it!

Would any razor blade work? For example, I could just take apart this utility knife that I keep at my computer desk and just steal one of its replacement blades. Although, it's not exactly the highest quality utility knife....

 

[Idontcare]

I'm not worried so much about the actual cutting of adhesive, but rather the amount of adhesive on my particular CPU seems rather thin. I was sitting there holding it in my hand and wondering, "How in the world would I get a razor under that without damaging it?" I might just be over-thinking it a bit. Well, actually... if history is any indication, I am over-thinking it!

Would any razor blade work? For example, I could just take apart this utility knife that I keep at my computer desk and just steal one of its replacement blades. Although, it's not exactly the highest quality utility knife....

That was the case with my first 3770k as well, which is why I used a small tack-hammer to carefully push the blade through the adhesive in very small increments.

 

[JimmiG]

You'll see Ivy-E being soldered.


It's another way of trying to force people onto their server tech lines.
In the long run it seems like a better idea to put us there.


Mainstream is not increasing absolute perf - only E\EP\EX lines will do this from now on.

But it will probably be a long while before we get some better price ratios - gotta move the top echelon of performance mainstream into the high end SKUs to drive volume before you lower the prices.

Atleast that's how i see it - short of some technical validation issue they found with solder they didn't wanna risk on TriGate.

(Should check IVY delidders in a year or two - if they're still good).

The mid-range desktop segment is going away. We basically have:

-Handheld (smartphones, tablets)
-Laptops (including convertibles etc.)
-Workstations
-Servers

The Workstation/Server chips can share most features, while the LGA115x chips have to be modified because the few people that buy them don't want to spend $500+. This adds to the cost for Intel, while at the same time, margins are lower. Switching from solder to thermal goop was just one way for them to keep that cost down and margins up. But it's clear they don't like this segment.

Any update on the OC results? Your 4770K seems very similar to mine - 1.3V for 4.4 GHz, while I'm at 1.218V for 4.3 GHz and 85+C. Would be interesting to see whether all this work results in an actual increase in OC capability or just lower temps.

 

[Boulard83]

The mid-range desktop segment is going away. We basically have:

-Handheld (smartphones, tablets)
-Laptops (including convertibles etc.)
-Workstations
-Servers

The Workstation/Server chips can share most features, while the LGA115x chips have to be modified because the few people that buy them don't want to spend $500+. This adds to the cost for Intel, while at the same time, margins are lower. Switching from solder to thermal goop was just one way for them to keep that cost down and margins up. But it's clear they don't like this segment.

Any update on the OC results? Your 4770K seems very similar to mine - 1.3V for 4.4 GHz, while I'm at 1.218V for 4.3 GHz and 85+C. Would be interesting to see whether all this work results in an actual increase in OC capability or just lower temps.

I'm short in loose time Ive done quite a few paste testing on die/ihs and ihs/hs but i'm not done yet. I'll reupdate soon ! Sorry.

 

[Kenmitch]

Any update on the OC results? Your 4770K seems very similar to mine - 1.3V for 4.4 GHz, while I'm at 1.218V for 4.3 GHz and 85+C. Would be interesting to see whether all this work results in an actual increase in OC capability or just lower temps.

Your temps at 4.3ghz are under what conditions?

Lowering temps can lower vcore demand required for stability.

I can get 4.5ghz stable at 1.235v or so on my 4670k with a H100i but crazy temps with Linpack 11.

The nerd in me says delid my wallet says don't risk it. Nerd will win at some point.

My motherboard shipped with f4 uEFI and I did most of my overclock testing with it. I updated to the f6 uEFI and so far it looks like it will require a little less vcore for stability at the same overclocks. Not a huge drop in vcore but trending around 0.02-0.05v or so. F7 uEFI just dropped for my board so im wondering if it'll have similar results.

I mention the uEFI as z87 is pretty new and most likely running non perfect uEFI currently. As it matures I'd expect better results.

 

[qliveur]

I'm short in loose time Ive done quite a few paste testing on die/ihs and ihs/hs but i'm not done yet. I'll reupdate soon ! Sorry.

It's okay, just don't forget about us. I find this subject fascinating. :thumbsup:

 

[Aikouka]

That was the case with my first 3770k as well, which is why I used a small tack-hammer to carefully push the blade through the adhesive in very small increments.

Ah, I'll have to look into that. I might just test on my Ivy Bridge i3-3225 first just to dip my toe in the water. I could try the vise method with that as well. Although, I was thinking about the whole thin layer of epoxy "problem", and I wondered if they made thinner razor blades. I figured X-Acto might have something, and I stumbled across this product. It does look like X-Acto makes an even thinner blade too.

What's pushing me toward delidding (again) is that my temps just don't seem that good... and I'm not even overclocking (yet)! SpeedFan doesn't seem to be capable of reading Haswell temperatures, but ASRock's A-Tuner reports my CPU as idling at 37C and I hit around 50C while playing a game. Now, that may not seem bad... until I say that I'm water cooled with 9 120mm fans providing cooling! To be fair, it could be related to either (1) my CPU being directly after two GTX 680s in the loop or (2) attempting to delid once. I'm guessing that #2 might have simply made the gap situation worse even though it didn't appear to do anything at first. ...or maybe these poor temperatures are just normal?