Intel X299 CPU's NOT soldered!

[done12many2]

20 minutes ago, MageTank said:

There are also alternatives to CLU, that do not dry up at all when exposed to air, such as Thermal Grizzly Conductonaut, but this particular LM is far more liquid-y than CLU, which makes it slightly riskier to use if you do not take proper precautions. Always use Kapton tape or clear-coat nail polish on any exposed contacts near the die, and re-seal the IHS to try to prevent leaking. Bonus points if you use the Bitspower IHS, as it's designed to help hold your LM in place without being resealed.

 

That aftermarket IHS is pretty sweet.

[WereCat]

@MageTank

I am considering to delid my 4770k. Right now I am at 4.6GHz and temperature spikes in low 90°C when running Aida64 (haven't even bothered trying AVX with P95).

The difference in temperature between the hottest and coldest cores is over 10°C so I am sure the delid would help.

 

However, I am afraid to use LM as on Haswell there are capacitors right next to the die and that would be quite a huge risk as I don't plan on resealing the IHS.

I have spare MX-4 paste, do you think it makes sense to go with that one?

[MageTank]

Just now, WereCat said:

@MageTank

I am considering to delid my 4770k. Right now I am at 4.6GHz and temperature spikes in low 90°C when running Aida64 (haven't even bothered trying AVX with P95).

The difference in temperature between the hottest and coldest cores is over 10°C so I am sure the delid would help.

 

However, I am afraid to use LM as on Haswell there are capacitors right next to the die and that would be quite a huge risk as I don't plan on reselling the IHS.

I have spare MX-4 paste, do you think it makes sense to go with that one?

MX-4 does work, but the best paste (from a longevity standpoint) is Dow Corning. If you don't mind repasting every year or two, Kryonaut or KPx will yield the best thermal results for conventional pastes. That being said, I've even used AS5 on my bare die before, and it still yielded a 3-5C drop in temps (and helped with per-core temp uniformity). So anything will help.

 

You can use Kapton tape, or Clearcoat nail polish on your caps and it will be just fine. I've done that on my G3258, both of my brother's CPU's (4690k, 4790k) and friends of mine have also done so with their Haswell CPU's. Just paint a nice thin layer of clearcoat over top of the caps (or just use capton tape around the entire area of the die, whichever method you prefer) and you are good to go. 

[done12many2]

12 minutes ago, WereCat said:

@MageTank

I am considering to delid my 4770k. Right now I am at 4.6GHz and temperature spikes in low 90°C when running Aida64 (haven't even bothered trying AVX with P95).

The difference in temperature between the hottest and coldest cores is over 10°C so I am sure the delid would help.

 

However, I am afraid to use LM as on Haswell there are capacitors right next to the die and that would be quite a huge risk as I don't plan on resealing the IHS.

I have spare MX-4 paste, do you think it makes sense to go with that one?

 

In addition to the options for shielding the caps on the substrate provided by @MageTank, liquid electrical tape works fantastic.  I've used clear nail polish and liquid electrical tape and both work great.  The only difference is that you can easily remove the electrical tape should you decide to undo anything.  

 

I'd definitely recommend CLU or Conductonaut over any paste.   

 

Good luck and have fun with it.

[MageTank]

Just now, done12many2 said:

 

In addition to the options for shielding the caps on the substrate provided by @MageTank, liquid electrical tape works fantastic.  I've used clear nail polish and liquid electrical tape and both work great.  The only difference is that you can easily remove the electrical tape should you decide to undo anything.  

 

I'd definitely recommend CLU or Conductonaut over any paste.   

 

Good luck and have fun with it.

I've switched over to Kapton tape, as I find myself soldering more and more these days (and I needed it for that hobby). That being said, I'll definitely be picking up a jar of that liquid electrical tape. Should help save me money on Kapton tape, as that stuff is certainly not cheap, lol. 

[Coaxialgamer]

23 hours ago, ARikozuM said:

Expansion and contraction of solder was found to possibly cause breakage of the core as micro-fissures formed.

these chips are far from being small though...

[The Benjamins]

2 minutes ago, done12many2 said:

 

In addition to the options for shielding the caps on the substrate provided by @MageTank, liquid electrical tape works fantastic.  I've used clear nail polish and liquid electrical tape and both work great.  The only difference is that you can easily remove the electrical tape should you decide to undo anything.  

 

I'd definitely recommend CLU or Conductonaut over any paste.   

 

Good luck and have fun with it.

Do you think petroleum jelly would work?

they use that for LN2 prep to protect them from condensation.

[done12many2]

11 minutes ago, MageTank said:

I've switched over to Kapton tape, as I find myself soldering more and more these days (and I needed it for that hobby). That being said, I'll definitely be picking up a jar of that liquid electrical tape. Should help save me money on Kapton tape, as that stuff is certainly not cheap, lol. 

 

There are guys on OCN and HWBOT covering complete GPUs and other PCBs with the stuff.  It's great because it just peels off mostly in one complete piece when you're done.  

 

So far I've used it to cover PCB surrounding shunts on GPUs, around GPU dies, and for contacts on CPU substrates.  The stuff sticks awesome.  You can coat it with as many layers as you want for added insulation/security.  All you have to do is wait a few minutes between layers.  

 

Here's a photo showing the liquid tape coating the area surrounding the liquid metal coated shunts.  Disregard my ghetto fabricated plastic barriers below the shunts.    My cards are vertically mounted on a horizontally mounted board, so I wanted something in place in the event of some liquid metal run off.  

 

Once the unlocked XOC bios for the 1080 Ti cards was released, I was able to undo the liquid metal shunt mod by simply wiping off the CLU and peeling up the liquid tape.  It took 10x longer to disassemble the blocks then to undo the shunt mod. 

 

[Senzelian]

wow, so many pissed off people because of a non-soldered CPU.

 

 

[done12many2]

3 hours ago, The Benjamins said:

Do you think petroleum jelly would work?

they use that for LN2 prep to protect them from condensation.

 

Sure it will, but who wants to smear that all over their PCBs?  

[dexT]

3 hours ago, done12many2 said:

 

That aftermarket IHS is pretty sweet.

bartx makes a bare copper one http://www.bartxstore.cba.pl/pots.php

[Foxxer]

 

GGWP.

[done12many2]

2 minutes ago, dexT said:

bartx makes a bare copper one http://www.bartxstore.cba.pl/pots.php

 

That's even nicer.  Do you happen to know what they cost?

 

[Princess Luna]

@done12many2 Do you think something like the Cooler Master V8 GTS would be enough for the i9 7960x/7980xe on stock? Or these are pretty much mandatory that you go liquid? (considering no delid)

[Senzelian]

2 hours ago, Sampsy said:

We'll ultimately have to wait for reviews. Buuut... these are big CPUs which are going to kick out a LOT of heat. Them not being soldered could genuinely be a problem. 

Then don't buy one... 

Do you know how many bad products are out the on the market? One or two more won't let the planet explode.

[dexT]

Just now, done12many2 said:

 

That's even nicer.  Do you happen to know what they cost?

 

No but his contact is right there. He's still active I believe and a good guy from what I understand.

[done12many2]

Just now, Princess Cadence said:

@done12many2 Do you think something like the Cooler Master V8 GTS would be enough for the i9 7960x/7980xe on stock? Or these are pretty much mandatory that you go liquid?

 

Sure that would be enough.  More then plenty for use at stock clocks.  The clocks speeds / voltage when all cores are firing will be low enough at stock speeds.  The very second you start increasing voltage, even moderately, you'll begin having thermal issues for sure.  

 

Of course, I truly have no real idea until I get one myself.

[PCGuy_5960]

11 hours ago, tom_w141 said:

They absolutely do not have to. Their margins are insanely profitable as it is. This is cheaping out, please don't defend them on this it's wrong of Intel and you know it.

I am definitely not defending them! They should have soldered the 6+ core CPUs! I just can't think of any reason why they would do something like that....

10 hours ago, Pohernori said:

You think intel cares enough to make their HEDT CPUs affordable? 

That's the last thing they would do. 

Maybe, because Ryzen is very competitive...

[PCGuy_5960]

23 hours ago, DocSwag said:

Yeah ik, hence why I'm still a bit skeptical about that explanation  

 

The part about the Pentium 4 does seem to make a bit of sense in his favor though, since they were soldered and had smaller dies. At the same time though, I just looked it up and it looks like the P4 had a max operating temperature of around 70C instead, which could explain why (thermal cycling isn't as intense).

But the Pentium 4 was a 130nm CPU tho.... The 7700K is a 14nm CPU

[79wjd]

3 minutes ago, PCGuy_5960 said:

I just can't think of any reason why they would do something like that....

I can only imagine that it was for cost savings and I hope it was because soldering was damaging many chips and not to just save a few dollars on each chip.

[PCGuy_5960]

Just now, djdwosk97 said:

I can only imagine that it was for cost savings and I hope it was because soldering was damaging many chips and not to just save a few dollars on each chip.

Yeah, that's what I thought as well... They are trying to compete with Ryzen so a $1000 8 Core CPU wouldn't sell too well

[79wjd]

3 minutes ago, PCGuy_5960 said:

Yeah, that's what I thought as well... They are trying to compete with Ryzen so a $1000 8 Core CPU wouldn't sell too well

A $600 CPU (on a more expensive platform) that only has an IPC advantage won't sell too well either -- it really needs the clockspeed advantage as well. If Intel didn't think Ryzen was a threat then they wouldn't have slashed the price of a 7820x 40%.

[PCGuy_5960]

4 minutes ago, djdwosk97 said:

A $600 CPU (on a more expensive platform) that only has an IPC advantage won't sell too well either -- it really needs the clockspeed advantage as well. If Intel didn't think Ryzen was a threat then they wouldn't have slashed the price of a 7820x 40%.

AVX? Ryzen's AVX performance is quite bad compared to Intel's....

[79wjd]

Just now, PCGuy_5960 said:

AVX? Ryzen's AVX performance is quite bad compared to Intel's....

I'm not saying reasons don't exist -- HEDT also has quad channel memory and more PCIE lanes, but that's not enough for most people to warrant paying more than double the price. 

[PabloUS55]

This of not being soldiers also applies to the series xeon w.
Since the xeon w-2155, is basically the i9-7900x with support for 512GB of ECC ram and 4 lines pcie more