Core i9-7900X Skylake-X Review Shows Up

[Deleted member 172152]

View attachment 89238

 

[trparky]

OK, so long as you don't overvolt or overpower the chip you will never fry it, I get that. However the "less than desirable" TIM will result in more throttling since heat won't be able to be pulled away from the chip fast enough. True you won't fry the chip but it could result in less performance. The chip automatically throttles itself down to save itself when it gets too hot.

 

[EarthDog]

OK, so long as you don't overvolt or overpower the chip you will never fry it, I get that. However the "less than desirable" TIM will result in more throttling since heat won't be able to be pulled away from the chip fast enough. True you won't fry the chip but it could result in less performance. The chip automatically throttles itself down to save itself when it gets too hot.

correct.. mostly.

Keep it to 90C while stress testing and throttling will never happen. That leaves you 10c of headroom on these chips for warming ambient or other factors.You typically wont see the same temps a stress test gets either. Certainly not hotter. Thats the point is to make it a 'worst case' scenario.

That said, with my custom loop, i was able to reach 4.5ghz using aida64 stress test and keep it at 90c (7900k). During gaming temps were way less. Rendering 1080p/4k, encoding, compiling, was all several to 20c less than stress test temps.

The tim will only cause throttling if you raise the voltage and clocks past a certain point with your cooling. To put it the same way for the umteenth time, its certainly fine at stock and wont cause throttling. There is headroom for some overclocking as well. As much as many want/what indium solder would give??? Probably not, but the tim isnt really a problem (shit) like you two were incessantly going on about and blowing it out of proportion.

 

[cadaveca]

Yo, Earthdog, have you check power consumption on OC or stock? I can pull the max rated 300W with these chips easy.

Also remember, this CPU has FIVR, so it seems hotter than it should be because of internal voltage regulation like Haswell CPUs. TIM has nothing to do with it.


"someone said this thing sucks, so I'mma gonna go with that since my wallet likes me that way".

 

[EarthDog]

Edot:

4ghz 1.2v p95 v28.1 small fft showed 465w, lol! (System - at the wall)

Stock 265w... 1.15v

 

[cadaveca]

I dont recall stock, but I know i pulled 450W (system - at the wall) at 4.5ghz all cores/threads 1.3V...this was using p95 v28.1 small fft.

you might be over 300W on the CPU. Just enabling XMP and syncing cores gets me well north of 260W, and I use AIDA for load testing. First "normal" tests, the FPU-only. I have to update BIOS though, might be something weird there.

Tell your boss to spend the $100 for you to get a clamp-based amp meter! You can get a really good fluke one for less than $100, even, and the tool is invaluable.


I think it is HILARIOUS that you and I, guys who actually have access to these chips, are the ones that don't seem to be so adamant that there is something wrong with these chips. You'd think
we'd actually used them!

Also worth noting is the orientation of the DIMM slots and cache performance in AIDA64 memory benchmark (also the memory performance, wow!).

Haven't run into any weirdness as was reported by some reviewer...

 

[EarthDog]

I ninjad you... check my edit..

 

[EarthDog]

Ill have new boss soon enough...

Ill message you when not sitting at a stop light..


Yeah, its funny...and a bit sad all at the same time. What happened to listening?? What happened to being open to being corrected? What happened to supporting your talking points in a discussion. Im always happy to learn new things, but, i need support, not some random new people just spewing things out without it...

...sad forums...forums are sad.

 

[trparky]

OK, I was wrong. I get it.

 

[Kyuuba]

So why don't they use it?! I mean come on, we're paying Intel upwards of $300 USD for these damn things, the least they could do is put some decent TIM in these damn things.

That specific TIM has advantages others doesn't have, one of them is the longevity which can keep you from touching/delidding your cpu for almost a decade, i'll say this so you can understand, high temperature wont kill the chip, is the amount of electricity going through the silicon what kills it, even if it's running cool damage done via high voltage would fry it without problem, again that's why Intel uses their 'crappy' TIM as a limiting factor, you just need to understand that bro.

 

[trparky]

Understood.

 

[cadaveca]

 

[trparky]

Whoa.

 

[EarthDog]

From my review now... dat sh1t tim tho............

 

[Solaris17]

More importantly, does anyone know if the I9 case badge is that same fugly ass pink as the box?

 

[notb]

So why don't they use it?! I mean come on, we're paying Intel upwards of $300 USD for these damn things, the least they could do is put some decent TIM in these damn things.

Because it would mean compromising other properties. This is most likely the best all-round TIM they had. And once again: I strongly urge not to compare the specified thermal properties, because - to make it short - consumer products are full of sh...

BTW: when you look at "delidding" results carefully, most of the temp drop comes from significant shortening of the TIM/solder thickness, not from a different material.

I still stand by the idea that running that close to the thermal threshold makes me uncomfortable. I tend to monitor my system temperatures closely and 90c would cause me to freak out.

I'm pretty sure it's not that. It's most likely just the number. Almost 100*C. Almost boiling water. We're programmed to be worried.

 

[trparky]

I'm pretty sure it's not that. It's most likely just the number. Almost 100*C. Almost boiling water. We're programmed to be worried.

This!

 

[BluesFanUK]

Looks like my 5820K purchase a couple years ago just before Skylake released was a great move after all. OC'd to 4.5Ghz, not far off these results and it doesn't get obscenely hot either under an NH-D15.

Don't think i'll be upgrading for at least another few more years.

 

[uplink777]

Why does everyone see Kaby Lake as puzzling? It's not. It's pretty much straightforward. You have Z270 and a limit of 7700K, maybe some 7800K in the future. Now You have cheap mobos for X299 [starting at around 180e w/o VAT] and You can buy Kaby Lake now and in 2-3 years 7900X or go higher. So much for puzzling. They fused mainstream and high end. No big deal.

 

[EarthDog]

I'm pretty sure it's not that. It's most likely just the number. Almost 100*C. Almost boiling water. We're programmed to be worried.

agreed, however, when whitepapers and knowledgable forum memebers alike tell one not to worry, you would think people would listen a lot quicker than some recently have...

 

[notb]

agreed, however, when whitepapers and knowledgable forum memebers alike tell one not to worry, you would think people would listen a lot quicker than some recently have...

I actually kind of approve of what's happening. Yes, 100*C can kill you. Yes, you shouldn't believe people on a forum just because they have 1000+ thanks.

However, I expect more from individuals. Most people here at least seem to be adults, so they must have had some exposure to high-school physics by now. They should know that electronics are not afraid of 100*C.

They should have at least a slight intuition that modern industrial installations are also operated by some sort of electronics, so it's clearly possible to make chips that sustain way over 100*C (AFAIK the current efforts are to reach 300*C).

 

[Erazem]

How is saying ryzen is an all round better design than what Intel has being a fanboy when it's true?

Yes, Intel's cpus do clock a bit higher, get slightly more single thread performance, have a better implementation of avx and the infinity fabric frequency being fixed to the memory frequency isn't ideal which makes Intel's cpus better for some use cases.... but overall ryzens are better, the single thread performance is very close to Intel's, you get a lot more cores, pcie lanes, ecc support, better power efficiency and all that for less money.....

Bulldozer was a bust and the Phenom II while being not as good as what Intel had at the time was okish.

Sandy bridge was (and still is) an awesome cpu but after it Intel just didn't do much (ok, the igpu did get better) since they really had no competition.... and if you want to go further back the Core2quads, Celerons 300A, and probably a few others were also very good cpus for their time. But they also had some pretty crappy things like the P4 too... the Baytrail/Cherrytrail atoms while not very fast are also very under rated if they weren't artificially gimped by having 2 gb of ram at the most in the vast majority of cases (I think some of the tablets with them did have more but they aren't common) and very slow emmc storage they would be great cpus for cheap x86 tablets.....

And on the AMD side the Athlon and Athlon64 were great as well so both manufacturers had their good and not good moments...

 

[EarthDog]

Its a great buy if you can use more cores... and need more pcie lanes... and need ecc memory...

They have an advantage by price and cores. Nothing wrong with free cores, but if you arent using them, like most cant, and wont for years (where 4c/8t starts holding things back)...