Intel to possibly announce even more impressive chip than 7980XE at Computex

humbug · 7 Jun 2018 at 01:42

Hotwired said:
I'm gonna bet that a bunch of people didn't see the intel demo of their 28 core so I'll sum it up.

It's rolled out on stage and praised for its cinebench performance while still having amazing single threaded performance at 5GHz

https://www.youtube.com/watch?v=21hNs9GKR5c&t=57m20s

Not a single squeak that someone overclocked the nuts off it or that its being piped chilled water from under the table to make it possible. Just in and out, hey look at this 5GHz 28 core cpu.

Yeah, he actually said its a 5Ghz CPU, not even that its an overclocked CPU, just that its a 5Ghz CPU.

Now we know how trustworthy anything Intel say is.

Bluntwrapped · 7 Jun 2018 at 08:05

I don’t really see the problem with the demo. I kind of wondered whether it was live or a pre-recorded run. Can only imagine the embarrassment if they were to get a BSOD, live on stage

It seems like it was a pitch mainly to the HWBot/competitive community. Why? Because those guys are mental enough to blow £3K on a CPU.

sideways14a · 7 Jun 2018 at 08:06

Yes its all over the web and youtube noob reviewers...
Intel has 5Ghz 28 core... they are all lapping it up... at the time there were little mention of the huge overclock and rather spectacular cooling and power requirements.

If they had come out and said here is this 28 core cpu, this is it stock and look how far we can push it on exotic cooling then.. yeah maybe more acceptable. but the demo was pure BS throughout.

Panos · 7 Jun 2018 at 09:09

Orcvader said:
That's assuming it doesn't keep tripping the breaker .

https://www.reddit.com/r/intel/comm...iftly_removed_their_28_core_cpu_demo/e07b42r/

Hahaha.. So basically in USA no home can run that CPU because of the 150V supply. It would require separate home circuit tad bigger than the one connected to the kitchen oven...

But at least there is hope for us in Europe with the 240V (UK 230V) power supplies. We only need to make sure the is 25-30amp fuse, and pray cables do not melt in some old house.....

But my question is what to do with the heat output of almost 2800W. This cannot run in a normal house room. Has to be outside.

LePhuronn said:
Aren't we going a tiny bit too far now though? Nobody knows for sure what this CPU was, especially as all the demo units were pulled. Yes, this is a deceptive stunt on Intel's part, but if they say it's Cascade Lake then why wouldn't it be Cascade Lake? Laugh at Intel's poor judgement on this, enjoy the PR backlash, but anything more is getting a tad vitriolic methinks. Especially that video above, the guy is just banging on the same point about 4 times in the video. It's an unnecessary rant.

And let's be honest, Intel just ran a 28-core CPU at 5GHz across the board without using sub-zero cooling. Within that limited scope, how is that not awesome?

It was with whole fridge cooling, with total system heat output close to 2800W.
And as you read above, no house in USA can run this thing without asking electricians to create a new separate wiring circuit similar to the kitchen oven.....

8 Pack · 7 Jun 2018 at 09:40

Why is everyone saying 2800w needed?? This is ofcourse not true.

Beren · 7 Jun 2018 at 09:45

8 Pack said:
Why is everyone saying 2800w needed?? This is ofcourse not true.

They are adding the draw of the chiller and the draw of the chip and the rest of the CPU to the single circuit it is plugged into.
We are mostly having a giggle, but if the chip was drawing 800w then the chiller must be consuming at least that to keep it cool... That is 1600w then the rest of the machine... It could easilly have been trying to pull 2000w plus from an extension lead.

DrBombcrater · 7 Jun 2018 at 10:33

The motherboard Intel used has power sockets capable of delivering somewhere around 1100W if my reading of the ATX spec is correct. Assuming the presence of 4 8-pin connectors indicates power draw was too high to get away with just 3, then the CPU+Mobo must be pulling close to or in excess of 840W.

Accounting for the chiller's power consumption, the main PSU being less than 100% efficient, and extras like the gfx card, I'd think it's very possible the demo system as a whole could be pulling over 2KW from the wall.

sideways14a · 7 Jun 2018 at 10:38

Whole pc excluding cooling must be well over 1Kw, your going to want some sort of overclocked 1080ti in there and thats going to be asking over 300W on its own.
Add in ram, storage and all the other bits a psu needs to power like RGB (lol) and its going to be needing a nice fat fuse in the plug.
The cooler though takes the biscuit - well over a kilowatt for that on its own lol....

I like to see extreme overclocking runs but this was blatantly put out by intel as a fire extinguisher which didnt work at all.

Hagar · 7 Jun 2018 at 10:42

8 Pack said:
Why is everyone saying 2800w needed?? This is ofcourse not true.

I would love to know what is true in this case. So many unsubstantiated statements, including the one above.

Panos · 7 Jun 2018 at 11:00

8 Pack said:
Why is everyone saying 2800w needed?? This is ofcourse not true.

@nkata
We do not talk about electricity but heat. Electricity is around 2000-2200W without GPU.

The chiller used by Intel draws 1000W electricity and has 1770W cooling capacity that needs to be dissipated somewhere.
Add the 1000W the CPU was burning and you have the hot side trying to dissipate 2770W heat which is close to 2800W. First law of Thermodynamics.

passifid · 7 Jun 2018 at 11:43

Panos said:
@nkata
We do not talk about electricity but heat. Electricity is around 2000-2200W without GPU.

The chiller used by Intel draws 1000W electricity and has 1770W cooling capacity that needs to be dissipated somewhere.
Add the 1000W the CPU was burning and you have the hot side trying to dissipate 2770W heat which is close to 2800W. First law of Thermodynamics.

None of that makes sense. The cooler draws 1000w. But it has moving parts you can hear it etc. It will have far less than 100% energy lost at heat as a lot of it will go to the movement of water. So maybe 600-700w heat and the CPU we will assume is going 1000w heat. It has 1700w cooling capacity not heat production.
A fan less heatsink has a cooling capacity but it doesn't increase the heat production at all

Still a crap demo tho

Rroff · 7 Jun 2018 at 11:47

passifid said:
None of that makes sense. The cooler draws 1000w. But it has moving parts you can hear it etc. It will have far less than 100% energy lost at heat as a lot of it will go to the movement of water. So maybe 600-700w heat and the CPU we will assume is going 1000w heat. It has 1700w cooling capacity not heat production.
A fan less heatsink has a cooling capacity but it doesn't increase the heat production at all

Still a crap demo tho

I am curious where he is getting the extra heat output in wattage over the power draw wattage from heh.

sideways14a · 7 Jun 2018 at 11:48

nkata said:
I would love to know what is true in this case. So many unsubstantiated statements, including the one above.

Well its going to be close, the chiller was pictured and was apparently this thing http://www.hailea.com/e-hailea/product1/HC-1000B.htm
Say circa 800-1000 watts or so...
The cpu would be pulling stupid amounts of power, the board had 4 8 pin aux power connectors and 32 phases didnt it?
Thats got to be worth similar amounts of power to run say another 800w and the board with vga and all the aux kit needed like drives, fans, ram bla bla... wouldnt be hard to imaging a 1kw plus build in addition to the cooler at all especially as it had a monster psu.

So together passing 2kw would be very likely... i dont know about nearing 3kw but i could definitely see it cruising past 2 at full tilt.

Panos · 7 Jun 2018 at 12:14

passifid said:
None of that makes sense. The cooler draws 1000w. But it has moving parts you can hear it etc. It will have far less than 100% energy lost at heat as a lot of it will go to the movement of water. So maybe 600-700w heat and the CPU we will assume is going 1000w heat. It has 1700w cooling capacity not heat production.
A fan less heatsink has a cooling capacity but it doesn't increase the heat production at all

Still a crap demo tho

@Rroff
Is not water cooler to dissipate the heat of the water through the radiator. It works at full blast like your freezer and aircon. The device creates a flat 1770W of cooling for 1000W of electricity, which some could call efficient.

However those 1770W of provided cooling generate 1770W of heat that needs to be dissipated somewhere. Add the 1000W worth of heat from the the CPU and here is the 2770W heat on the hot side that needs to be dissipated.

Nothing to do with electricity, is just physics and precisely bound to the First Law of Thermodynamics. Cannot magically provide 1770W of cooling without generating 1770W worth of heat.
If that was the case here, this fridge/chiller designer should be getting the Physics Nobel price for the next decade annually, and his name put next to Einstein.

Hotwired · 7 Jun 2018 at 12:22

You're doing it wrong @Panos

Lets take your numbers of 1000W from the cpu and 1000W from the chiller

The amount of energy it can dump cannot be more than 2000W

Even if the chiller can dissipate 1700W while using 1000W it isn't being given 1700w from the cpu, it's being given 1000W.

So 1000W from the chiller + 1000W from the cpu = 2000W.

Panos · 7 Jun 2018 at 12:24

Hotwired said:
You're doing it wrong @Panos

Lets take your numbers of 1000W from the cpu and 1000W from the chiller

The amount of energy it can dump cannot be more than 2000W

Even if the chiller can dissipate 1700W while using 1000W it isn't being given 1700w from the cpu, it's being given 1000W.

So 1000W from the chiller + 1000W from the cpu = 2000W.

However it creates 1770W COOLING flat by consuming 1000W worth of electricity, it doesn't work like water cooling to dissipate heat related to the heat applied to it. To create that cooling it must create heat on the other side.

If it was providing 1000W of cooling yes the heat dissipation would be 2000W, but it does generate 1770W worth of cooling for 1000W electricity.

Panos · 7 Jun 2018 at 14:30

OK. The whole Intel cooling issue gets funnier....

The cooler apparently to some posts is a HAILEA HC 1000B. Which is illegal to be sold in Europe & US.
Actually none of the HAILEA HC series cooling solutions can be sold legally to Europe and North America (incl Canada).
Reason is they are using as freezing medium R22 or R134a (depending model) which are banned gases for quite some time as they belong to HCFCs. Especially R22 is banned for 14 years now. :rolleyes:

Also that explains the pretty good efficiency of the cooling at +77%.

But enough of laughing.... Because they are banned I want one now to cool my system

~
GTX1080Ti @ 3Ghz with 8600K @7Ghz here I come.....

Rroff · 7 Jun 2018 at 14:33

Think they can all use R134a?

humbug · 7 Jun 2018 at 14:41

Rroff said:
Think they can all use R134a?

No. it is how Panos said.

https://www.coolingpost.com/world-news/us-places-bans-on-r404a-and-r134a/
https://www.gov.uk/guidance/bans-on-f-gas-in-new-equipment
https://ec.europa.eu/clima/policies/f-gas/legislation_en

sideways14a · 7 Jun 2018 at 15:05

Lol, can this get any worse.