Holy fuck!

Holy fuck!

The new i9 series are just overclocked i7s with huge power consumption and heating issues.

How the fuck could Intel think this was a good idea?


>"I didn’t believe it at first, but the Core i9-7900X is using more than 70 watts additional power than the Core i7-6950X!"

>"But woe, if you have all the cores really work out properly! With Prime95 and Luxrender on all cores for the Core i9-7900X, we have determined absolute peak values for the power consumption. The Core i9-7900X achieves remarkable 48% more performance in the render score (Luxrender, console), but also takes up 58% more of the electrical power! This is almost the famous blue breaking bar."

>The system ran stable with 1.4 Vcore in 10 runs and the CPU took an average of 261 watts. Single peaks, however, were load-dependent at up to 293 watts, which is very close to the limit. A test carried out at 4.8 GHz with 20 instances of a shading-all-year calculation for a complex photovoltaic roof system including a yield calculation allowed the power consumption to rise to 335 watts. When starting Prime95 without AVX limitation then switched off the mainboard. The last, still logged value was 364 watts."

>"The real reason is Intel's incomprehensible fun brake in the form of unsuitable (but probably much cheaper) heat transfer paste instead of a sensible solder."

>"Up to 65 ° C Tcore with a heat spreader temperature of approx. 24 ° C results in a delta of more than 40 ° C. We reached this value with a power loss of just under 230 watts. As soon as you move above the 300-watt limit, which can be reached even with simple rendering programs from 4.6 or 4.7 GHz and the necessary voltages (depending on the chip quality), the chiller is hardly anything. At the maximum values of 300 watts reached by us the CPU is already permanently thermal limit of 100 ° C, shortly thereafter comes to the consequential shutdown."

Other urls found in this thread:

i.imgur.com/rK16YZC.mp4
fisica.unipg.it/luca.gammaitoni/fisinfo/documenti-fisici/physical-limits-silicon.pdf
guru3d.com/articles-pages/intel-core-i9-7900x-processor-review.html
pcper.com/reviews/Processors/Intel-Core-i9-7900X-10-core-Skylake-X-Processor-Review/Thread-Thread-Latency-and-
twitter.com/NSFWRedditImage

Housefires incoming lmao

Closed loop liquid nitrogen coolers when?
Hell, why don't we just stick the whole thing into a freezer at this point?

Man, just looked at the benchmarks.

>i9-7900 does 10% better than a i7-6950X
>base clock of 3.3GHz compared to 3.0GHz with the 6950

It's literally a 6950X with an overclock hahaha what the fuck

It performs worse than 6950X in many tests.

all this failure is seeming kinda suspicious though

I feel like there's something wrong here.

Is this gonna be like ryzen where nothing worked at first?

It doesn't matter how well you cool the heat spreader, if thermal jizz isn't transferring the heat from the die to the spreader. You'll end up with a IHS temperature below 0° and a core temperature above 100°.

but muh cinebench!

Intel is actually pushing these on people lul

Partially driver issues for sure, but consider what exactly "i9" is.

It's a Xeon die, with server features like ECC ram stripped off. Glue it to a HEDT socket. Overclock it and pray it works. They also shrunk the L3 cache, which might hurt some applications but the extra L2 might help in others.

I've also heard there's core to core communication latency issues. Like Ryzen's issue, except it effects all of the cores instead of half of them.

Finally, this is still a ~2000 pin socket. Threadripper will use a 4096 pin socket. That's more IO, more bandwidth, more power transfer...

we need hybrid liquid peltier air cooling

Dont forget the rumor going around intel is going to have to either gimp the clocks on the high core count dies to Xeon levels to maintain a power level that the socket can handle, or they'll put out yet ANOTHER socket JUST to handle the MCC i9s.

X99 is a grade a clusterfuck.

That's what I was also thinking. Slap a peltier cooler on the radiator, use a liquid with plenty of antifreeze and you can manage the condensation away from the components.

>when Intel plays catch-up with memes

>disable thermal limit and emergency shutdown
>delid
>now you have a 300w portable heater that functions as a personal defense weapon

>Holy fuck!
calm down, you hysterical spastic

They definitely pushed it out too quickly. But I doubt there's very much they can fix with software updates.

OwO what's this

>But I doubt there's very much they can fix with software updates.

will they fix the raid dongle through software

The paste is going to turn to sand in a matter of weeks. What a new record, Intel!

>4gb gddr5
user pls

how did they manage 4k pins?

Kaby lake was a lie, it was literally rebranded skylake.

I love how smug they were in January, claiming to their investors that from here on, its all about optimization and not about nanometers, cores or ghz.

I think we need to get some more nuclear power plants online a.s.a.p

>The new i9 series are just overclocked i7s
What else would they be. Retard.

>CPU took an average of 261 watts
>CPU took an average of 261 watts
>CPU took an average of 261 watts
>CPU took an average of 261 watts
>CPU took an average of 261 watts
2-6-1-watts

It's fucking huge (and I mean that in the best way)

I've been saying for ages that we need larger CPUs. It's not like we're limited in the tower space or anything - just fucking make everything huge.

Apparently the talk is that there's a new socket for the 12 core and up chips because this one has a thermal limit.

So in other words, i9 is a gimmick that does more damage than good and will be gone before most people even notice it was there.

It's the fucking kaioken of processors.

I couldn't take that off without making it super ugly, I am not good at paint

With what appears to be the largest mass produced socket ever.

>Pull the plug, pull the towers
What did they mean by this

Not POWER8?

Nope, hell even LGA 2011 is bigger than POWER8's socket.

That doesn't sound right, I remembered that POWER8 was the size of fist.

Well, Socket SP3 is even bigger

I cant even tell the difference between the shops and the real pics at this point

that's the real one

I pulled that one from Guru3d's site. GIS "Threadripper", second row, 2nd image for the same image. If you look around on the same GIS search page as above you can see more shots of the Threadripper package from different angles.

There is a limit, of course. At a certain point, the electrical signal won't be able to travel far enough to perform logic operations between clock cycles.
See pg. 2727 (Propagation of pulses on wires) in this awesome paper for more info.

Yeah, except you can't patch out hardware issues like with Ryzen.

Multiple cores acting autonomously

Checkmate, Moore

H Y P E R B I P E L I N E S

Pentium D 2.0?

M A G I C B I N G B U S : D D D D D

50+ stages

pumped at 14ghz

jayhawk is gonna rock

i.imgur.com/rK16YZC.mp4

They still need to communicate with one another, unfortunately. I'm all for breaking what we understand to be the laws of physics, but we'll need a ton of incredibly clever people to do it.

If you're at all interested in the physics behind what's going on inside of your CPU, you must read this paper. You should be able to process the broad strokes if you have any sort of background in physics. If you've studied physics/engineering (barring you civE fucks), at the university level, you'll likely get a great understanding of what the hell is actually happening inside that box of yours.

fisica.unipg.it/luca.gammaitoni/fisinfo/documenti-fisici/physical-limits-silicon.pdf

yeah good idea to disable automatic shut down on a chip that is going full throttle into a nuclear runaway scenario

Are you really that incompetent that you couldn't make a webm/gif and post that here? Reddit pls go

On that note:

>Every reviewer and their mothers get ludicrous temps even at stock speeds on liquid cooling
>Linus "unbiased" review gets load temps of 55° at stock and 80° OC'd at 4.5Ghz

I mean I could have seen it wrong, but really Linus? That's your unbiased review?

guru3d.com/articles-pages/intel-core-i9-7900x-processor-review.html

>It has been and still is a complete and utter mess with Intel, I can honestly say that as for this entire X299 / Core X series release Intel has not send any info or samples. Our hails through emails are not responded to, and the phone-numbers we dial of contacts are not answered. Here in the EU Intel has killed (with a few exceptions) marketing, press and PR activities. Hence 98% media all over Europe is not receiving any information from Intel anymore

I'm guessing Intel is going all-in on burgers too stupid to realize this is their 9590

>Physics papers on Sup Forums
wat

Remember all the Intel shills bashing Ryzen both before and after launch?

Remember all the Intel shills saying how great the i9's would be before launch?

Now it's actually launched and reviews are up and the shills have gone so so silent.

Paybacks a bitch.

Just wondering, doesn't this wear the CPU out a lot?

Power consumption that high would never survive in the long run right?

Yeah it's upsetting that the fact that Sup Forums seems to be more closely related to Sup Forums than /sci/ isn't upsetting anyone.

Sup Forums - Smartphone VR Goggles

Is this a solder thing then?
The overclockers who dun delidded seem to do ok
I think Der Bauer had SKylake-X overclocked to 5GHz with exactly the kind of CLC that has been running too hot for reviewers

those are NOT a heating issues
this CPU just tryies to go SSJ2

>>Sup Forums

There were even space enthusiasts here before. Not the Musk fanboy type mind you.
Almost every shuttle launch had a thread.

if i9 is SSJ2 what is ryzen?

technology is deprecated
we telephones and gimmick of the week now

Brolly when you ate his favorite dessert ??

>Trend of hippies producing ULV processors almost universally regarded as gutless shit
>OH NOES, NEW PROCESSOR IS USING TOO MUCH POWER NOW PLS FIX

Never change, Sup Forums.

Hey, I like a powerful chip just as much as any other enthusiast. My (and everyone elses) problem lies in the fact these things are absofuckinglutely impossible to keep cool, and they suck up a fuckton of power while they're at it.

Not even Bulldozer and Piledriver was this bad, as while they had the high power consumption they didn't have their temps push 100C or require a high end CLC just to barely function within their specified temperature range.

thats not all of it

>AMD finally won for once
feels good

>they didn't have their temps push 100C
Probably that solder right?

That likely has a significant part to play in it. The biggest problem again comes back to Intel being cheap fucks with their own chips and not spending the $5 or so on the materials to actually properly solder their chips to their heatspreaders.

>gigant meteor destroys earth
that's because they bent reality
it's all your fault faggots

...

nothing of value was lost
good people will go to heaven anyway

The real problem here isn't even the power consumption but the temps. Needing a high-grade CLC just to run the CPU at stock clocks is just ridiculous.

>nothing of value was lost
b-but muh Threadripper ;___;

thats not a problem for intel..

they probably lost a lot of money selling K cpu's
and now they are trying to lock people on a platform that is made to run hotter than sun and hotter than a blue giant if you overclock it
either way intel made them so that they are broken down faster the perfect planned obsolence that puts even nvidia on shame
you cant overclock cause it will void the warranty
you cant get maximomum perfomance because of thermal throttling
literally a cpu made to make people a tech cuck

Neo-Sup Forums in a nutshell.

I miss the old days when deep discussion of tech was commonplace. Thanks for sticking around/being interested.

Honestly, I'm all for consumerism. The money helps drive more R&D and advancement. That being said, it is a bit unfortunate that Sup Forums has mostly become a price comparison forum.

In grad school I took a few courses taught by a man who is likely the #1 expert in the world on atomic layer deposition. ALD has been instrumental in the development of sub-14nm lithography. His (S. M. George) courses were incredibly interesting, and drove home how amazing even the current fabrication processes are.

If you're interested in how modern semiconductor devices are fabricated and controlled, I encourage you to read some of his papers.

If we have the Kaioken now, I'd love Super Saiyan here in 2018/2019... Unless Intel goes full derp and thinks doing Kaioken x20 is viable (i.e. this but with 500W power consumption, maybe we just need more power to offset the heat from the power guys)

>and now they are trying to lock people on a platform
You know that won't work? People will get cancer as soon as they see they need to dump $300 on a mobo, $150+ on a cooler and over $400 for decent DIMMs before they even get to the CPU.

pic related

...

and good luck if you want RAID

>364 watts
Okay, what the everloving fuck is this? Is this even possible for home computers? Shouldn't motherboard just melt at this point?

i cant imagine the dual setup on those.. 750 watts only for the cpu's
intel is literally going to recomend people to use 2 psu's at this point

CPUs with external power when.

2000

Why does the 6 core have the same TDP and power draw as a 10 core?

they did the same thing with pentium 4 and d.

And* isn't bothering anyone*

Your grammar bothers me.

At least Tom's Hardware needed an extra fan to cool the MOBO, so yeah, pretty much.

i9 is the Trunks version of ssj2 bulky full of muscles not able to perform. Threadripper is the Gohan true ssj2.

>I've also heard there's core to core communication latency issues. Like Ryzen's issue, except it effects all of the cores instead of half of them.
It's true. pcper.com/reviews/Processors/Intel-Core-i9-7900X-10-core-Skylake-X-Processor-Review/Thread-Thread-Latency-and-

>The new i9 series are just overclocked i7s with huge power consumption and heating issues.
like how i3s are just underclocked i7s with disabled cores

Liquid nitrogen wouldn't help in a closed system, needs to evaporate to work.

Intel wasted R&D on this when all they needed to do was cut prices a little.

They did cut prices, compared to their last HEDT CPUs. The 10 core 6950X used to be $1700.

>wasted r&d
they don't have any serous r&d user, they've been milking the same designs for some 10 years

Somebody at Intel still thinks it is late 1990s and early 2000s.

The HEDT and Ethuasiast market doesn't matter much anymore. The real money is mainstream and enterprise market. Intel appears to have mainstream market lock down nice and tie unless Ryzen Moblies prove to be game changers (Not likely). The enterprise market on the other hand is in jeopardy since Epyc chips are going to cheaper to make in mass then trying to fab massive 20+ core monolith dies and AMD no longer being held back by a shitty architecture.

>Intel appears to have mainstream market lock down nice and tie unless Ryzen Moblies prove to be game changers (Not likely).
Are you nuts? Ryzen is insanely power efficient and coupled with integrated graphics that are actually decent I can't see any way for Intel to actually compete on the mobile side. They'll either have to rely on their own sucky graphics and lose on the performance front or use nvidia's discrete graphics which will negatively affect battery life. Also APUs are cheaper to produce so cheaper laptops and desktops all around.