Intel Core i7-7700K Delidded, Reapplied TIM Reduces Temps By 30C

[P4-630]

Could Intel Have Messed Up TIM Once Again?
"So an interesting story just popped up regarding Intel’s upcoming Core i7-7700K processor. A member of the Anandtech Forums has managed to delid the processor and posted some great findings after reapplication of new thermal interface."

Intel Core i7-7700K High Temperatures Could Be Caused By Poor TIM – Temps Improve After Delidding
"The forum member who goes by the name of “RichUK” managed to acquire a retail sample of the Intel Core i7 7700K processor. Several overclock runs were posted and the user went as far to delid the chip and apply new TIM (Thermal Interface Material). The results are shocking as it may proof that Intel has once again used poor quality TIM on their processors.

The issue with high temps on Intel processors arose back in the Ivy Bridge generation where the chips were heating up beyond a certain limit. The heat build up would affect overclocks and would result in poor stability. The issue was fixed when Intel started using higher quality packaging materials and improved TIM in the new Devil’s Canyon (Haswell Refresh) lineup. The last three generations of Intel processors haven’t seen issues related to heating but the issue might be returning.http://cdn.wccftech.com/wp-content/...-Lake-vs-Core-i7-6700K-Skylake_Processors.jpg

Several tech sites have already published reviews of the Core i7-7700K prior to the chip’s launch. All sites report higher temperatures for the Kaby Lake part compared to Skylake chips. One reason for that is due to the fact that Kaby Lake ships with higher clock speeds and boards (Z170) are not optimized with voltage ratings of the new processors as seen in Expreview’s review. But the other reason could very well be the application of poor TIM between the CPU die and the IHS (Internal Heat Spreader). More performance leaks can be seen here, here and here."

Intel Core i7-7700K Delidded and Tested (Before and After Results)
"RichUK started off the testing by acquiring his brand new Core i7-7700K processor. This is a retail chip and not an engineering sample. If the issue was on an ES chip, we could have blamed the early sample for being the reason but all tech sites that have covered the reviews got the retail chip for testing. The system used for test is listed below:

• Processor – Core i7-7700k
• Mobo – Asrock Z170 Pro4S – BIOS v7
• RAM – Corsair 2x8GB 3000Mhz C15
• GFX – EVGA GTX 1060 6GB SC
• Cooling Stock – an old Thermalright 6 heat pipe tower, Corsair SP120
• Cooling for OC – Corsair H110i with 2 x Corsair ML140 Pro LED"

Intel Core i7-7700K Stock With Thermalright Cooler:
"On the first day, the user tested the chip in Cinebench R15 at stock settings. The temps during the test were stable around 60C. This testing was achieved with a Thermalright Ultra 120 tower cooler."

Intel Core i7-7700K OC With Corsair H110i:
"The moment the chip was overclocked to 5 GHz (1.34V) and tested through IBT and OCCT, the temps spiked. During Prime 95 test, the maximum reported temperature was 96C. The chip was stable for 15 minutes in Prime 95 v27.9. The chip passed through 6 hours of stability in Prime95 Small FFTs at 4.7 GHz (1.264V) with temps averaging around 83C. Cooling used was a Corsair H110i set to quite mode.
The chip was tested through several benchmarks including Cinebench R15 at the stable clock speeds of 4.7 GHz."

Intel Core i7-7700K OC With Kraken X62 and Delid:
"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.

The user reports that previously, the chip would shoot up to 99C but now it was averaging around 66C in Prime 95 v28.7. Furthermore, it was stable while before, it wasn’t as much.

The results show that new TIM application and delid improves the temps by a great margin. Delidding generally improves the temps but we don’t see such decreases so it is possible that the higher temps on Kaby Lake CPUs could be related to the low quality TIM used. We won’t say that it’s entirely to be blamed on TIM as the chips also ship with higher clock speeds over Skylake but this is a really interesting find by RichUK."

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

 

[Keullo-e]

And one another stupid thing is that the IHS is physically different, so I guess it can't be popped with the Ivy->Skylake delid tools.

 

[Folterknecht]

Same happend to StullenAndi from hardwareluxx.de ... ~25K improvement after delidding.

 

[Ferrum Master]

This is actually a sad thing for such a pricey item...

 

[P4-630]

This is actually a sad thing for such a pricey item...

Yes it is , I don't get it why intel doesn't use quality TIM under it, especially on the more expensive chips.

 

[KainXS]

ehhh they did it again, Intel never learns do they.

 

[JalleR]

that is just sad, but maybe that is how they will beat ZEN when it is launched by adding better cooling paste and running higher clocks. LOL

Funny that on the Extrem models it is good.

 

[infrared]

This is bullcrap. I really really hope Zen hurts intel big time for ripping off their customers while being lazy and cheap with their manufacturing. Hopefully this is the wakeup call they need!

I had to de-lid my 6700k on day 1 because temps were atrocious. Looks like the 7700k is even worse in that respect

 

[alucasa]

There is no excuse really. Intel had virtually no improvements for Kaby lake, so they had nothing to do. Yet they still cheap out on TIM?

Do they take customers for granted? If so, it's time that they are taught a lesson in Zen.

I have no temp issue with my i7-6700 though.

 

[cadaveca]

I have no temp issue with my i7-6700 though.

That kind of the point. Yes, changing TIM can lower temps. But the chips work just fine as is, and are actually likely to live a bit longer since they will throttle before they damage themselves. Change TIM, and then the CPU can seem to run cool, but can then consume so much current you can kill it easily.

People seem to forget that the CPU cores are but a small part of today's modern CPUs and those other parts.. don't report temperatures. It's like you all think Intel, that makes billions, is either dumb, or purposefully releasing stuff of lower quality just to be a bunch of jerks...

BTW, I have 7700K, 7700, 7600K, 7350K. They are nice chips, if you aren't having unrealistic expectations for them.

Also, my chips appear to be retail as well, but they aren't really retail version. You cannot get retail version of CPU that isn't in retail...

I tell people when the question comes up, stay to 1.325V. Here is a user, pushing higher, and overheating? ROFL. NOT surprised.

 

[cdawall]

I tell people when the question comes up, stay to 1.325V. Here is a user, pushing higher, and overheating? ROFL. NOT surprised.

I mean to be fair if he is shooting for a maximum, voltage recommendations are all a joke

 

[infrared]

@cadaveca Sorry, I don't accept that it's ok to use crappy TIM on £300+ K series processors. It's a corner that doesn't need to be cut. They're already over-charging for the processors int he first place, due to lack of competition, they could at least cook us up a decent product.

 

[cdawall]

@cadaveca Sorry, I don't accept that it's ok to use crappy TIM on £300+ K series processors. It's a corner that doesn't need to be cut.

Welcome to buying a midrange cpu. Unluckily this is the same argument as GP104 not being a midrange product. They both fall squarely into the "mainstream" category. Hence why there are models above them.

 

[infrared]

Welcome to buying a midrange cpu. Unluckily this is the same argument as GP104 not being a midrange product. They both fall squarely into the "mainstream" category. Hence why there are models above them.

Yeah, I guess that's a good point. Still leaves a bad taste in my mouth.

 

[alucasa]

That kind of the point. Yes, changing TIM can lower temps. But the chips work just fine as is, and are actually likely to live a bit longer since they will throttle before they damage themselves. Change TIM, and then the CPU can seem to run cool, but can then consume so much current you can kill it easily.

People seem to forget that the CPU cores are but a small part of today's modern CPUs and those other parts.. don't report temperatures. It's like you all think Intel, that makes billions, is either dumb, or purposefully releasing stuff of lower quality just to be a bunch of jerks...

BTW, I have 7700K, 7700, 7600K, 7350K. They are nice chips, if you aren't having unrealistic expectations for them.

Also, my chips appear to be retail as well, but they aren't really retail version. You cannot get retail version of CPU that isn't in retail...

I tell people when the question comes up, stay to 1.325V. Here is a user, pushing higher, and overheating? ROFL. NOT surprised.

Your point is valid for non-K versions. If they did cheap out on K-version, that's a whole different story. Those K-versions are meant to be pushed.

Otherwise, you are right on.

 

[cdawall]

Yeah, I guess that's a good point. Still leaves a bad taste in my mouth.

I would be upset that they are an overpriced mainstream product.

 

[cadaveca]

Your point is valid for non-K versions. If they did cheap out on K-version, that's a whole different story. Those K-versions are meant to be pushed.

Otherwise, you are right on.

Every CPU is made to handle a specific amount of power. It's rated by that, in fact. "K" CPUs have higher limits than "non-K" CPUs already. That added limit is how much MORE you are supposed to be able to push it (65W - 91W = nearly 50% boost). If you want more... look at a different product.

@cadaveca Sorry, I don't accept that it's ok to use crappy TIM on £300+ K series processors. It's a corner that doesn't need to be cut. They're already over-charging for the processors in the first place, due to lack of competition, they could at least cook us up a decent product.

It's not crappy TIM. It is designed purposefully so that should power drawn exceed a specific value, the CPU will overheat, and then throttle, protecting itself. Yes, this does mean it might run "hotter" than it needs to, so that throttle can be triggered at the right point. Change the TIM< you change where that throttle point occurs, and thereby put yopur expensive CPU at risk. CPU cache, which actually produces the most heat, doesn't report temperatures.

I'll say, I don't pay for my hardware, so maybe that lets me see things as they are, rather than wanting more than I am supposed to get for my dollar. If you don't like it, then simply don't buy it.

I'm going to leave this topic now though, and I'll have more to say in the review.

 

[LAN_deRf_HA]

I thought it was figured out awhile ago it's not about TIM. It's about the tiny gap made by the epoxy.

 

[R-T-B]

I thought it was figured out awhile ago it's not about TIM. It's about the tiny gap made by the epoxy.

I have reapplied the epoxy in my delids, and I still get better thermals.

 

[Hood]

It's not clear whether the CPU mentioned is running naked or only had the TIM changed and the heat spreader re-installed. The 30c drop in temps sounds like he's running a naked die, with it's high risk of breakage...

 

[Keullo-e]

That kind of the point. Yes, changing TIM can lower temps. But the chips work just fine as is, and are actually likely to live a bit longer since they will throttle before they damage themselves. Change TIM, and then the CPU can seem to run cool, but can then consume so much current you can kill it easily.

People seem to forget that the CPU cores are but a small part of today's modern CPUs and those other parts.. don't report temperatures. It's like you all think Intel, that makes billions, is either dumb, or purposefully releasing stuff of lower quality just to be a bunch of jerks...

BTW, I have 7700K, 7700, 7600K, 7350K. They are nice chips, if you aren't having unrealistic expectations for them.

Also, my chips appear to be retail as well, but they aren't really retail version. You cannot get retail version of CPU that isn't in retail...

I tell people when the question comes up, stay to 1.325V. Here is a user, pushing higher, and overheating? ROFL. NOT surprised.

Yep, these locked Skylakes are cool as hell (well, that sounds kinda strange.. ), my dualcore G4400 runs under 30C idle and ~50C under load, all case fans at 5V and cooler's (Thermalright Macho Direct) fan is about ~280rpm..

It's just hard to decide do I get a 7350K or 7600K later at spring..

e: btw, is there still that 0.06mm gap between the die and IHS like they had on some previous CPUs?

 

[Folterknecht]

Also, my chips appear to be retail as well, but they aren't really retail version. You cannot get retail version of CPU that isn't in retail...

I tell people when the question comes up, stay to 1.325V. Here is a user, pushing higher, and overheating? ROFL. NOT surprised.

Retail chips

http://www.hardwareluxx.de/communit...gewechseltem-tim-891243-338.html#post25150844

 

[infrared]

It's not clear whether the CPU mentioned is running naked or only had the TIM changed and the heat spreader re-installed. The 30c drop in temps sounds like he's running a naked die, with it's high risk of breakage...

Hmm, it's hard to say, the difference between running with and without a heat spreader on a die this small is negligable. I didn't see a temperature drop going from 'with IHS' to without. Could be dependent on heatsink/wb design though, it's a much smaller area you're trying to move the heat from which negates the benefit of having less thermal resistance.

Edit: 'with IHS' above is already with de-lid with CLU btw. I saw a 20c drop doing initial de-lid.

 

[Recon-UK]

M<eh.

 

[Vario]

Thing is, temps might decrease but the performance is pretty much the same post delid. Maybe you will get 100 mhz more out of it, but even that won't really be noticeable.