Intel Shows Off 28 Core, 56 Thread Core-X HEDT Processor For Enthusiasts, In Market Q4 ’18

[jpiniero]

Now, maybe it's a holdover from a previous designed board, but that's seems highly unlikely considering these are early working samples of the board made just for early testing/demonstration. These aren't the production ready versions of the board. Now, why would a brand new board made specifically for this new cpu be labeled as SKL-SP1S if the new cpu it was designed for was in fact CSX? Possible? Yes, but very, very unlikely.

I would be surprised if the demo wasn't simply an unlocked 8180. The final product will have the appropriate stuff disabled.

 

[coercitiv]

Because posters pointed me to articles that had little sign of skepticism, and posters here immediately were skeptical.

By the time we began discussing this on the forums all kinds of info was readily available. My post was one of the early replies in the thread, and I already had good reason to believe chilled water was used. This was based on a VideoCardz tip they received prior to the presentation, which was ironically ignored because VideoCardz was skeptical it could be done

Nowadays news travels very fast, and maybe this event is the perfect showcase of much hindsight can affect the way we judge our sources of information.

 

[Fir]

The issue with used Xeons aren't usually getting the CPUs themselves, but a compatible board.

Unfortunately Intel has locked down the newer Xeons to only run in boards with Xeon specific chipsets. Which is a bit of a downer for the used market.

EVGA SR-2 is the only choice since it allows overclocking. Those boards are still expensive, must be the mining thing still.
The SR-X, IMHO, was dead on arrival since the CPUs were locked.

In any case with this "new" platform, time will tell. I hope it's out before year's end though. Will make good use of this year's fiscal spend down cash!

 

[dacostafilipe]

I haven't watched the presentation.

Please do before commenting on how people should interpret the presentation.

 

[Dayman1225]

GamersNexus has said the demonstration was Skylake SP with altered uCode to OC but the final product will be based on a the newer Cascade uArch

https://twitter.com/GamersNexus/status/1005096047469707266?s=19

https://twitter.com/GamersNexus/status/1005102983225344000?s=19

Also WTH are you guys using SKL SP board indicators to say it's SKL SP? CCL SP and SKL SP are pin compatible

 

[LTC8K6]

GamersNexus has said the demonstration was Skylake SP with altered uCode to OC but the final product will be based on a the newer Cascade uArch

https://twitter.com/GamersNexus/status/1005096047469707266?s=19

https://twitter.com/GamersNexus/status/1005102983225344000?s=19

Also WTH are you guys using SKL SP board indicators to say it's SKL SP? CCL SP and SKL SP are pin compatible

So the chip that will be delivered "soon" is going to be Cascade Lake? Is it expected to perform better than the 8180?

 

[coercitiv]

So the chip that will be delivered "soon" is going to be Cascade Lake? Is it expected to perform better than the 8180?

At this point there is no confirmation from either Intel or inside sources that is SKL or CL based.

 

[LTC8K6]

By the time we began discussing this on the forums all kinds of info was readily available. My post was one of the early replies in the thread, and I already had good reason to believe chilled water was used. This was based on a VideoCardz tip they received prior to the presentation, which was ironically ignored because VideoCardz was skeptical it could be done

Nowadays news travels very fast, and maybe this event is the perfect showcase of much hindsight can affect the way we judge our sources of information.

Right, VC got an accurate tip and then didn't believe the tipper...
They missed a huge scoop.
I wonder when they began to believe the tipper might have been correct?

 

[formulav8]

Are reporters not allowed to speak at these presentations?
Can they ask, "Hey, how is that being cooled?".

No these are presentations only. No Q & A's. You can hear how they present the 28c CPU @ around 1:10:15

 

[Dayman1225]

So the chip that will be delivered "soon" is going to be Cascade Lake? Is it expected to perform better than the 8180?

I'd assume so because Cascade Lake is essentially Skylake SP on 14nm++, with bug fixes, upgraded IMC(Jedec Spec Upgrade to 2933mhz) and in Silicon Mitigation for Spectre and Meltdown.

 

[Dayman1225]

At this point there is no confirmation from either Intel or inside sources that is SKL or CL based.

FWIW GamersNexus said they corroborated with "sources close to the product" and then later said it'll move to a newer uArch in the final product (Cascade Lake)
https://twitter.com/GamersNexus/status/1005095205345914882

 

[LTC8K6]

I don't understand the end game here.

Let's pretend that the cooling was never revealed, and most people still think this was not done with exotic cooling and they accept Intel's presentation about the chip.

That means that most people then expect Intel to release such a chip this year, right?

But Intel, as far as we know, can't deliver this chip this year, or even next year.

 

[Hitman928]

FWIW GamersNexus said they corroborated with "sources close to the product" and then later said it'll move to a newer uArch in the final product (Cascade Lake)
https://twitter.com/GamersNexus/status/1005095205345914882

They said, "it will eventually be a new uarch." Does that mean the final version of the first chip or that there will be a refresh down the line? If they mean the final version of the first chip, it's hard to think that they'll have anything but a paper launch by the end of the year if they don't have even a single working chip to use for the computex demo.

Perhaps it is a new uarch and they will get it out by the end of the year with actual availability, but intel has been very reluctant to share any details of the chip when press members have tried to follow up. It just feels like intel is doing things on the fly right now and is in pure hype and hope you don't get called out mode. I could be wrong, we'll see.

 

[IEC]

They said, "it will eventually be a new uarch." Does that mean the final version of the first chip or that there will be a refresh down the line? If they mean the final version of the first chip, it's hard to think that they'll have anything but a paper launch by the end of the year if they don't have even a single working chip to use for the computex demo.

Perhaps it is a new uarch and they will get it out by the end of the year with actual availability, but intel has been very reluctant to share any details of the chip when press members have tried to follow up. It just feels like intel is doing things on the fly right now and is in pure hype and hope you don't get called out mode. I could be wrong, we'll see.

PR stunt to generate buzz (which it did). As we all know, there is no such thing as bad publicity.

And as if we needed more evidence it was a PR stunt:

One of the interesting numbers to come out of the initial announcement were that overclockers were seeing 1000W power draw for an 18-core Core i9-7980XE at 4.9 GHz, so one would assume that a 28-core at 5.0 GHz would be more. We know that the demo that achieved 5.0 GHz was using a Hailea water chiller capable of 1770W of cooling, and the system being used had a 1600W power supply with a low power graphics card. That gives us a good idea for low-bound and high-bound. However Intel would not help narrow down our estimates.

Source: https://www.anandtech.com/show/12932/intel-confirms-some-details-about-28core-5-ghz-demonstration

Mission: Accomplished.

 

[HutchinsonJC]

It's an enthusiast part, showcased in such a way that enthusiasts are, by and large, going to be the primary audience...

It's really quite perplexing trying to even think what their goal with the whole thing could have been. They tried to impress an audience that's the most likely kind of audience to pick things apart.

 

[Zucker2k]

It's good PR for Intel: 'We have a 28 Core chip that can overclock to 5GHz,' (the presenter said so) albeit on a chiller. That's something their competition can't say. It's a good selling point for the niche of the niche who overclock these chips. The crème de la crème of overclockers all over the globe probably drooled in unison when they saw this chip pushing those clocks and numbers on chilled water. This demo was not for serious production environment folks, it was for those who take the risks of pushing the limits on performance, and people like me who marvel at the spectacle and the technology it takes to accomplish these groundbreaking feats.

 

[LTC8K6]

If the mission was to look silly, they accomplished it.

Unless they release a chip somewhat close to the chip in the demo, which isn't going to happen.

So I guess we wait and see if a 28 core chip with a 2.7 base clock is delivered, to find out what this chip actually was.

 

[HutchinsonJC]

That's something their competition can't say

The kind of cooling they used, is almost up there with liquid nitrogen, though. And we don't count any liquid nitrogen OCs as any kind of practical/daily driver. Someone above said something about having to do a whole new circuit breaker just for a machine to run both the CPU and the chiller. We're looking at 2500+ watt rig, without counting a GPU, and that's being modest in estimation!

 

[coercitiv]

FWIW GamersNexus said they corroborated with "sources close to the product" and then later said it'll move to a newer uArch in the final product (Cascade Lake)
https://twitter.com/GamersNexus/status/1005095205345914882

I would be both glad and sad to learn this was previewed as SKL and the final product will be CL. The glad part is obvious, the sad part is not hard to decode: this would imply they used previous product to emulate new product, which means the demo was bordeline fake. (either we admit SKL and CL are too similar for differences to matter - hence CL is a rebrand, or the the differences matter and it was all smoke and mirrors).

 

[LTC8K6]

The kind of cooling they used, is almost up there with liquid nitrogen, though. And we don't count any liquid nitrogen OCs as any kind of practical/daily driver. Someone above said something about having to do a whole new circuit breaker just for a machine to run both the CPU and the chiller. We're looking at 2500+ watt rig, without counting a GPU, and that's being modest in estimation!

The chiller is 1100W / 9.6A from the wall.
So, you'd just need to plug the chiller and the PS into different 15A USA house circuits.
You wouldn't need to install a new breaker or wiring.

 

[PG]

Shocking Interview with Intel Engineer about 28-Core 5 GHz CPU

There have already been 2 warnings in this thread for linking to the Intel interview video. This is not allowed in the technical forums.

Daveybrat
AT Moderator

 

[HutchinsonJC]

So, you'd just need to plug the chiller and the PS into different 15A USA house circuits.

I mean, I get that, but how many houses have two different 15A circuits in a single room that also doubles as the room you'd want to use your machine in? Are we talking about running extension cords into an adjacent room? And then remind yourself to never run the vacuum cleaner off either of those circuits, too?

Edit:
I put their whole show case on the same level of getting on stage with liquid nitrogen. Irrelevant for anything but a science experiment.

 

[LTC8K6]

I mean, I get that, but how many houses have two different 15A circuits in a single room that also doubles as the room you'd want to use your machine in? Are we talking about running extension cords into an adjacent room? And then remind yourself to never run the vacuum cleaner off either of those circuits, too?

Edit:
I put their whole show case on the same level of getting on stage with liquid nitrogen. Irrelevant for anything but a science experiment.

I think unless your house is old, you should have plenty of circuits.

A hair dryer is commonly 1200-1500W, a microwave 1200W, a toaster oven 1500W, etc.

A dyson vac can be 1400W at full chat.

But the overall point is certainly valid. It's a lot of power for a computer.

 

[HutchinsonJC]

It's a ridiculous amount of electrical power for a computer haha. And then you'll want a UPS on the chiller and one on the computer. Then you need that much more power for Air Conditioning (watts)

Add a GPU (watts).

It's so ridiculous that I can't imagine a real purpose for it except to show off.

Most anyone needing that kind of compute power, would surely be better served with some multi-socket solution or some kind of dual computer setup: One machine on a multi-socket platform, and another with the higher Hz for anything that's single (or a few) thread intensive/demanding like gaming.

Edit: Better served in up front costs, better served in electrical consumption, better served in reliability, etc, etc

 

[maddie]

The chiller is 1100W / 9.6A from the wall.
So, you'd just need to plug the chiller and the PS into different 15A USA house circuits.
You wouldn't need to install a new breaker or wiring.

What about the PC itself? Free vacuum energy at work?

Apologies.