Intel X299 CPU's NOT soldered!

[Blake]

19 hours ago, ARikozuM said:

There's no point to soldering, IMO. The risks don't outweigh the benefits.

 

Hopefully they add enough TIM this time though.

 

13 hours ago, ARikozuM said:

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

 

I hate to do this, actually no, I don't.

[Citation needed]

 

Both have their advantages and disadvantages, typically TIM is cheaper and easier to do, where as solder is more expensive but does a better job. So if you using TIM on a 'enthusiast' product, not the best call, solder really should be used here. where as in the datacentre and on mid to low end chips, that are not designed to be overclocked, there is not issues using TIM (however Intel's marketing team likes to screw this up on the low end). 

 

[Flavio hc 16]

IMHO the people are not understanding the gravity of the problem: not even considering the price of those chips, we are looking at chips that at stock speeds consume like a 7700k at 5ghz at 1.35volts, and these settings are not achievable on a 7700k that has not been delidded. Now a lot of you will tell me: "but the chips that we are looking at are way bigger and so they have way bigger dissipation area", that's the problem: the dissipation area, considering the tim application of the 7700k as a baseline, will not be linearly bigger than the one on the 7700k, so we will see cpu's near or at thermal throttling temperature at stock speeds, even with liquid cooling. Because you can have eve 2 360mm rads, but if the heat dissipation between the die and the heat spreader is bad, and with the 7700k is terribad, you will still get thermal issues

[tom_w141]

17 hours ago, PCGuy_5960 said:

I think that the 14+ core CPUs will be soldered tho.... They are probably doing this on the 12- core parts so that they can be (somewhat) competitive with AMD. AMD's CCXs allow them to make cheaper dies, Intel doesn't use CCXs so they have to cut a few corners to make the CPU a bit more affordable

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.

[TOMPPIX]

if intel could solder this bad boy i'm sure they could solder these cpus also.

[Kinda Bottlenecked]

17 hours ago, PCGuy_5960 said:

I think that the 14+ core CPUs will be soldered tho.... They are probably doing this on the 12- core parts so that they can be (somewhat) competitive with AMD. AMD's CCXs allow them to make cheaper dies, Intel doesn't use CCXs so they have to cut a few corners to make the CPU a bit more affordable

 

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

That's the last thing they would do. 

[Sonorpearl]

http://overclocking.guide/the-truth-about-cpu-soldering/

 

Best part:

 

-- Whenever I read sentences like “What a ripoff – Intel doesn’t even solder a 300 USD CPU” or “Why does intel save 2 USD on soldering” I’m thinking

 

[Valentyn]

8 minutes ago, Sonorpearl said:

http://overclocking.guide/the-truth-about-cpu-soldering/

 

Best part:

 

-- Whenever I read sentences like “What a ripoff – Intel doesn’t even solder a 300 USD CPU” or “Why does intel save 2 USD on soldering” I’m thinking

 

 

Doesn't apply to the massively large die 6-18 core processors.
All current X99 Haswell and Broadwell ones are soldered. Sure you can delid them, and get a reduction of 4 degrees.

Where as with Kabylake a delid and repaste drops between 23-30 degrees at max loads.

Normal TIM with a bad application is a potential disaster for 8-18 cores on Skylake X as they all have significantly higher TDPs, and cores.

[Sonorpearl]

14 hours ago, yian88 said:

I still cant believe in this day and age CPU's and GPU's cost 1000$+ and nothing affordable for sub 150$ price range with decent performance, such a small piece of silicon that cost so much, still pisses me off.

This CPU is very powerful. I use it as an All-In-One Solution for, my Smart-Home, NAS, HTPC and Security Server all in one build. Also for Gaming with some Emulators on my TV, older Games run quite good on this. Everything to like Portal 2 runs very well on it.

 

 

5 minutes ago, Valentyn said:

Where as with Kabylake a delid and repaste drops between 23-30 degrees at max loads.

I want to see that test done right.

 

5 minutes ago, Valentyn said:

Normal TIM with a bad application is a potential disaster for 8-18 cores on Skylake X as they all have significantly higher TDPs, and cores.

As if they never ran the CPUs and checked for maximum potential. Do you think they reduced the clock speed and write down higher ones than they can run on? The clock speeds always turned out great, compared to the reasonable temperatures for that power. Also, Intel does know how to apply TIM right.

[done12many2]

Just now, Valentyn said:

 

Doesn't apply to the massively large die 6-18 core processors.
All current X99 Haswell and Broadwell ones are soldered. Sure you can delid them, and get a reduction of 4 degrees.

Where as with Kabylake a delid and repaste drops between 23-30 degrees at max loads.

Normal TIM with a bad application is a potential disaster for 8-18 cores on Skylake X as they all have significantly higher TDPs, and cores.

 

There are theories popping up right and left.

 

I agree that they should have if they could have.  Whether or not they could have with dependability is something we're all just speculating at this point. 

 

 

[79wjd]

5 minutes ago, Sonorpearl said:

Also, Intel does know how to apply TIM right.

You say that, but Haswell and Kaby Lake disagree with that statement. 

[The Benjamins]

10 minutes ago, done12many2 said:

 

There are theories popping up right and left.

 

I agree that they should have if they could have.  Whether or not they could have with dependability is something we're all just speculating at this point. 

 

 

he even says this is still a bit risky which is crazy. if the thermal paste is bad and you have to delid to get a decent OC. I would be too scared to to that to a $1000+ CPU.

5 minutes ago, djdwosk97 said:

You say that, but Haswell and Kaby Lake disagree with that statement. 

That is the reason why I hate my 4770k, it is impossible to put a high OC on due to uncontrollable temps.

 

seeing the price and it not being soldered I will either upgrade to Ryzen or TR.

[Sonorpearl]

11 minutes ago, djdwosk97 said:

You say that, but Haswell and Kaby Lake disagree with that statement. 

 

 

70° C -> 67/68° C

Even tho it is SkyLake - Real gains here. I don´t trust many reviews.

I would do it to my KabyLake Chip. But I don´t want to damage it. Would test it after I got the new i9 tho.

 

4 minutes ago, The Benjamins said:

That is the reason why I hate my 4770k, it is impossible to put a high OC on due to uncontrollable temps.

I loved my i7-4790k. It overclocked like a boss. And the temps were ROCK solid.

[Kinda Bottlenecked]

18 minutes ago, done12many2 said:

 

There are theories popping up right and left.

 

I agree that they should have if they could have.  Whether or not they could have with dependability is something we're all just speculating at this point. 

 

 

 

My fav part: 

 

Quote

"my advice is to always go for liquid metal" 

 

Get ready yer delids

[done12many2]

8 minutes ago, The Benjamins said:

he even says this is still a bit risky which is crazy. if the thermal paste is bad and you have to delid to get a decent OC. I would be too scared to to that to a $1000+ CPU.

That is the reason why I hate my 4770k, it is impossible to put a high OC on due to uncontrollable temps.

 

Maybe we can just wait and see what actual reviewers experience in less then 2 weeks?  

 

Every single thing we're talking about is just continued speculation.  

 

Sucks to hear about your 4770k.  About a year ago, lucked out with a couple of good 4790k's that I bought of of Craigslist (2 x 4790k and 2 Asus z97 boards for $240).  Both chips ended up clocking well on small loops and I was able to flip them for some decent money.  One of the better deals I've ever scored.  

 

Quote

seeing the price and it not being soldered I will either upgrade to Ryzen or TR.

 

I'm thinking about TR too, but I'll do Skylake X first and go from there.  

 

 

1 minute ago, Pohernori said:

 

My fav part: 

 

 

Get ready yer delids

 

That's my motto.  lol

[79wjd]

2 minutes ago, Sonorpearl said:

 

 

70° C -> 67/68° C

Even tho it is SkyLake - Real gains here. I don´t trust many reviews.

I would do it to my KabyLake Chip. But I don´t want to damage it. Would test it after I got the new i9 tho.

There were a lot of problems with Linus' testing methodology, and I said Kaby Lake for a reason -- the 7700k is having problems, the 6700k did not. 

 

 

[The Benjamins]

7 minutes ago, Sonorpearl said:

 

 

I loved my i7-4790k. It overclocked like a boss. And the temps were ROCK solid.

That also makes me mad, they fixed some of those issue on the refresh. I don't want to risk a $300~ CPU to get better temps or buy a replacement CPU.

 

My 4770k will hit 80-90c at 4.3Ghz on my custom loop, my (now dead ;-;) 290x with a 200Mhz OC struggled to get above 65c on my loop AFTER getting water from the CPU block.

 

don't get me wrong it is not a bad CPU it just pisses me off that the reason I can't push the CPU more is the crappy TIM inside the HIS.

 

I also don't care for its longevity I let it hit 80-90c, if it dies I wouldn't even care.

[Valentyn]

35 minutes ago, Sonorpearl said:

I want to see that test done right.

 

https://www.hardocp.com/article/2017/01/19/intel_kaby_lake_i77700k_cpu_delid_relid_results

 

Quote

Replacing our Kaby Lake TIM resulted in a 25.28% decrease in package temperature under a full load for an hour. I am using a Koolance triple radiator and model 360 Koolance water block.

 

 

http://wccftech.com/intel-core-i7-7700k-delid-performance-tests/

 

Quote

After the CPU was successfully delidded and applied with CoolLaboratory Liquid Ultra thermal paste and put under a Kraken X62 cooler, the chip was once again tested at 5 GHz (1.344V). With the fan operating at 50% and pump operational at 65% (silent mode), the chip reported a decrease of up to 30 degrees Celsius. On average, the chip was around 26C cooler than before delid.

 

 

 

[done12many2]

^^  Pretty much what I've repeatedly experienced with 7700k CPU delids.  

[Sonorpearl]

12 minutes ago, Valentyn said:

https://www.hardocp.com/article/2017/01/19/intel_kaby_lake_i77700k_cpu_delid_relid_results

 

 

 

http://wccftech.com/intel-core-i7-7700k-delid-performance-tests/

 

 

 

 

 

No one shows their cooling solution and stats in comparison. Also the long term effects of using a different thermal paste. And the long term effects of using their thermal paste. Read about some Reddit tests that hinted, the performance would decrease to the PONR.

 

Someone also did a delid and tested it against the same stock intel paste. And the results also lowered over time.

 

That is, why I want a video of testing it, over a long time of period.

I know that the over kinds of pastes also dry faster.

Or could simply cause micro cracks.

 

I am not saying it is better or worse. I just want to see really deep testing and comparison.

[79wjd]

3 minutes ago, Sonorpearl said:

No one shows their cooling solution and stats in comparison. Also the long term effects of using a different thermal paste. And the long term effects of using their thermal paste. Read about some Reddit tests that hinted, the performance would decrease to the PONR.

 

Someone also did a delid and tested it against the same stock intel paste. And the results also lowered over time.

 

That is, why I want a video of testing it, over a long time of period.

I know that the over kinds of pastes also dry faster.

Or could simply cause micro cracks.

 

I am not saying it is better or worse. I just want to see really deep testing and comparison.

The problem isn't the TIM that Intel uses; the TIM is fine. It's their inconsistent/insufficient application methods that lead to problems with the 7700k.

[Valentyn]

4 minutes ago, djdwosk97 said:

The problem isn't the TIM that Intel uses; the TIM is fine. It's their inconsistent/insufficient application methods that lead to problems with the 7700k.

 

Exactly, and if they soldered the 6-18 core processors there'd be little chance of any temperature issues at all.

Der8auer found a degree drop in max temps after delidding Broadwell -E, where as with kabylake we can see the huge amounts when normal TIM is applied wrong.

That potential issue on Skylake X could lead to some nasty thermal issues for some people.

[MageTank]

30 minutes ago, Sonorpearl said:

 

 

70° C -> 67/68° C

Even tho it is SkyLake - Real gains here. I don´t trust many reviews.

I would do it to my KabyLake Chip. But I don´t want to damage it. Would test it after I got the new i9 tho.

 

I loved my i7-4790k. It overclocked like a boss. And the temps were ROCK solid.

Please, do not regurgitate misinformation just because you heard it from someone that you believe knows the answers, becuase you yourself do not know the real answers. @LinusTech was wrong. If he still believes he was right, then he is still wrong to this very day. You do not use a paste that is bad under intense thermal cycles. You do not write-off liquid metal because "it's dangerous". The process of delidding itself is far more dangerous, and if you are going to put in that kind of effort, then why not make it worth it?

 

We have posted countless amounts of evidence as to why Linus was wrong, and he never did come back to retest this. It was only recently that he even corrected his misinformed video on ram speed, simply because Ryzen showed significant gains with it. Even then, it was half-hearted. If you want to see my personal gains from delidding, I have a thread on this very forum showcasing my before and after results, along with my testing methodology: 

This was on a stock 6700k, undervolted to 1.12v. This was also in an ITX case using an ITX CPU cooler. Gains were significant for me, just as they are for people using custom loops running at 5ghz+ on their 7700k's. There are entire threads devoted to delidding on Overclock.net, and people delidding various generations of CPU's to show significant gains. To refuse to believe this, because one tech youtuber with a massive audience claims otherwise, is nonsense. Always look for multiple sources to a claim. Find me someone else claiming delidding isn't significant, and i'll look further into that claim (even knowing otherwise from personal experience).

 

5 minutes ago, djdwosk97 said:

The problem isn't the TIM that Intel uses; the TIM is fine. It's their inconsistent/insufficient application methods that lead to problems with the 7700k.

We also have to remember that the fin pitch for the 7700k was altered, and with both Kaby/Skylake, the PCB substrate was made thinner. This reduction in substrate thickness, without a change in the automated gluing process, might be causing the issues as well. Technically speaking, the change in thickness from Haswell to Skylake, may have also slightly altered mounting pressure itself, though I doubt this given the socket retention arm on the motherboards remain the same (and that will absorb most of the pressure).

 

Either way, delidding is no placebo, and soldered TIM has been proven to be superior for the most part. It's less hassle, less of a risk for the consumer (assuming the die is large enough to mitigate voids/microcracks) and handles subzero better than anything (especially Liquid Metal, which is terrible for subzero). The fact that we now have confirmation that the 10c/20t SKU isn't using solder, scares me. Especially if that was indeed a retail chip, and not an ES. The future of the HEDT being marketed towards enthusiasts is looking pretty grim if this is what we have to put up with. 

[Sonorpearl]

32 minutes ago, MageTank said:

This was on a stock 6700k, undervolted to 1.12v. This was also in an ITX case using an ITX CPU cooler. Gains were significant for me, just as they are for people using custom loops running at 5ghz+ on their 7700k's. There are entire threads devoted to delidding on Overclock.net, and people delidding various generations of CPU's to show significant gains.

What paste did you use and how are the results compared to today. Implicating that you have used that CPU and still have the same paste on it.

32 minutes ago, MageTank said:

(assuming the die is large enough to mitigate voids/microcracks)

That is, what I´m worried about. Long term usage.

 

 

Also, most won´t change up their paste every year.

Some of my friends never changed the paste on anything. Not even the cooler. And not even clean their PCs from dust. And use it for like 6+ years.

[79wjd]

20 minutes ago, Sonorpearl said:

That is, what I´m worried about. Long term usage.

He's talking about microcracks that come from (Intel) soldering (too small a die), not from replacing TIM. 

20 minutes ago, Sonorpearl said:

 

Also, most won´t change up their paste every year.

Some of my friends never changed the paste on anything. Not even the cooler. And not even clean their PCs from dust. And use it for like 6+ years.

Again, the TIM Intel uses isn't the problem. TIM can last a very long time -- the problem is an improper application method.  

[MageTank]

27 minutes ago, Sonorpearl said:

What paste did you use and how are the results compared to today. Implicating that you have used that CPU and still have the same paste on it.

That is, what I´m worried about. Long term usage.

 

 

Also, most won´t change up their paste every year.

Some of my friends never changed the paste on anything. Not even the cooler. And not even clean their PCs from dust. And use it for like 6+ years.

I used CLU (coollaboratory liquid ultra) and the results have held up perfectly. There has been little to no change in thermals (maybe 1-2C at best) after having it on the bare die for nearly a year (June 11th makes 1 full year). This is after a ton of thermal cycles, involving Prime95 (512k-4096k FFT, I love to stress ram) and various 48k FFT tests when swapping out the CPU cooler itself. On the bare die, nothing beats liquid metal (Aside from solder). That being said, we do not recommend using liquid metal in between the CPU and the heatsink itself, as that runs the risk of leaking out on to the board itself (gallium bonds to gold contacts, and will short your motherboard and potentially other components). The difference in temperatures when using LM on top of IHS vs your conventional pastes is small, 2-3C at best. Not worth the additional risk, compared to in between the die and IHS, which has consistently netted a 15-30C difference in temperatures for some, depending on how bad their chip was pre-delid.

 

I have a friend with a 4790k that is going on his third year now of having CLU on his die, and he tells me he has not noticed any degradation in temperatures at 4.5ghz, and has never once replaced his liquid metal. That being said, there are other scenarios in which I do not recommend LM on a bare die. Laptops with terrible mounting pressure is one of these scenarios. I used CLU on my brothers XPS 13, and due to the nature of it's heatsinks poor mounting pressure, air was able to get in between the heatsink and bare die, causing the CLU to solidify into this sandpaper-like texture. Thermals suffered immediately (within 15 minutes of a single thermal cycle) and it required some serious force to clean it off the die. I do not recommend CLU for portable devices that do not have an IHS to help re-seal the CPU itself. Not to mention, if it does maintain it's liquid form, you certainly don't want to be moving that around all willy-nilly, running the risk for it to leak out on to other components. In a scenario in which you see yourself traveling/moving your PC often, always reseal the IHS with a silicone glue ( @done12many2 recommends Permatex Black Silicone Adhesive Sealant).

 

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.