What will happen when Intel brings core counts greater than four to mainstream?

[cbn]

What will happen when Intel brings core counts greater than four to mainstream?

Here is one thing I think could happen:

Microsoft adds limited multi-user capability to certain home versions of Windows 11 (Maybe a hint of this to come is the virtual desktops feature in Windows 10). This begins to process of desktops becoming a server hub for various devices used by the family and home.

 

[master_shake_]

unless amd gets their ass in gear we won't see that for a while.

hypothetically i would say that the HEDT would be around 12 cores.

 

[happy medium]

I'll buy one.

 

[tenks]

There will be no Windows "11". MS announced Windows 10 is the last version of windows.

 

[coercitiv]

There will be no Windows "11". MS announced Windows 10 is the last version of windows.

They also said Start Button will be gone starting with Win 8. I might have believed them if they had named it just "Windows".

 

[frozentundra123456]

Hell will freeze over.

 

[DigDog]

There will be no Windows "11". MS announced Windows 10 is the last version of windows.

are they moving to a unix based system like macos ?

 

[Dave2150]

Hell will freeze over.

Pretty much this.

It's extremely profitable for Intel to be able to sell tiny 22nm, 14nm 4c dies, as they are absolutely tiny, compared to the 6core+ dies.

It would also mean they'd be forced to reduce the price of their xeon server offerings - so no chance of that happening.

The only way it would ever happen is if Zen actually gets produced and if it actually has good performance. Two things that are extremely unlikely.

 

[scannall]

They also said Start Button will be gone starting with Win 8. I might have believed them if they had named it just "Windows".

Since Windows isn't for the most part distributed on disks anymore, then the assorted 'Model Names' don't really make a lot of sense anyway. Further, they are trying to get away from having to support many different versions.

So it'll just be 'Windows' in the future.

 

[cbn]

It's extremely profitable for Intel to be able to sell tiny 22nm, 14nm 4c dies, as they are absolutely tiny, compared to the 6core+ dies.

Xeon-D octocore is only 160mm2 and that includes the two 10GbE LAN and the usb 3.0.

The i7-5775C quad may be larger due to iGPU (though I am waiting to see how Intel arranged the die and whether or not the core size is the same as Broadwell U).

 

[escrow4]

Intel should kill all dual cores and set:

- Base i5 quad for $100 ish

- Base i7 hexa for $200 ish

- Base i9 octa for $300 ish

And split those further according to clock speed and cache. I'd expect a dual core in a phone not a desktop anymore. Also have the base models split one with no iGPU one with iGPU. Make everything overclockable. That would kill off AMD in 48hrs.

 

[cbn]

Intel should kill all dual cores and set:

- Base i5 quad for $100 ish

- Base i7 hexa for $200 ish

- Base i9 octa for $300 ish

And split those further according to clock speed and cache. I'd expect a dual core in a phone not a desktop anymore. Also have the base models split one with no iGPU one with iGPU. Make everything overclockable. That would kill off AMD in 48hrs.

So Escrow has brought up a good question: What specs do you consider mainstream for hexcore and beyond? What TDP? What price?

At this point I would be happy enough with a 65 watt locked hexcore (with a good 1C,2C,4C turbo) and 80-95 watt watt locked eight core (with a good 1C,2C,4C turbo) as a starting point. This based on the Xeon-D type die with various server features disabled.

Of course, add unlocked multipliers and I am even more happy.

In any event, I would like to see Intel get more proactive with this area of consumer electronics. (re: cores > iGPU for the new era of desktop/home server)

 

[scannall]

Intel should kill all dual cores and set:

- Base i5 quad for $100 ish

- Base i7 hexa for $200 ish

- Base i9 octa for $300 ish

And split those further according to clock speed and cache. I'd expect a dual core in a phone not a desktop anymore. Also have the base models split one with no iGPU one with iGPU. Make everything overclockable. That would kill off AMD in 48hrs.

Why in the world would they adopt your suggested pricing? It makes no business sense at all. They don't care about AMD. Any sales AMD gets are a rounding error on their balance sheet, and makes a great anti-trust foil.

Never leave money on the table.

 

[ninaholic37]

What happened when Bulldozer went mainstream? Kinda nothing really...

 

[jpiniero]

I don't see more than 4 cores happening any time soon on the mainstream line. The main thing I see happening on the mainstream line in the near future is the PCH ondie and don't-call-it-HSA CPU-GPU integration features. 'Course they will continue to lower the TDP and power consumption. Maybe HBM or something similar although I know Intel isn't so hot on fabbing memory.

 

[TeknoBug]

A six core i7 with hyperthreading (or should that be i9) would be sweet.

 

[cbn]

A six core i7 with hyperthreading (or should that be i9) would be sweet.

What clockspeed?

For a locked 65 watt hexcore (based on a Xeon-D type die) I would be willing to take 2.7 Ghz base clock with 2C turbo @ 3.6 Ghz.

But it really depends on how much Intel charges for it.

And, of course, I also want the unlocked SKUs too (assuming Xeon-D's process tech does not cripple maximum clocks, see this post here).

 

[cbn]

Crossing my fingers Xeon-D's (or lets say a Skylake-D based on a Skylake Xeon-D) process tech scales well on frequency because imagine what we could do with a board like the one below:

As we can see LGA 2011-3 is pretty much overkill for the amount of PCB real estate available on this Mini-ITX (re: only two of the four memory channels can be used, only 16 of either 28 or 40 PCIe 3.0 lanes can be used due only one expansion slot)

But with a hexcore or octocore Xeon-D based consumer processor everything falls into place. (re: We can remove the X99 chipset and the SATA express and replace with one or two M.2 ultras to fully make use of Xeon-D's 24 PCIe 3.0 lanes. Furthermore, I'm pretty sure a Xeon-D based SoC even with the on package PCH is much smaller than that LGA 2011-3 socket area, freeing up even more space.)

 

[TeknoBug]

What clockspeed?

For a locked 65 watt hexcore (based on a Xeon-D type die) I would be willing to take 2.7 Ghz base clock with 2C turbo @ 3.6 Ghz.

But it really depends on how much Intel charges for it.

And, of course, I also want the unlocked SKUs too (assuming Xeon-D's process tech does not cripple maximum clocks, see this post here).

Going off what the i7 4790T does, probably what you said at 65W.

 

[BonzaiDuck]

Whatever it gives extra, I'm just not in a hurry.

I've got "build-it fever," with an annual budget and even a plan, not much need for mobile devices except for phone calls on shopping trips, the frequency of which I would like to reduce.

I don't even care anymore if I can watch CNN on my i-phone during those regular moments in a warm and quiet place every morning.

I'm even toying with the idea of upgrading to a 4790K system instead. But I keep my eye on threads like this one, and . . . we'll see . . .

 

[Yuriman]

Intel should kill all dual cores and set:

- Base i5 quad for $100 ish

- Base i7 hexa for $200 ish

- Base i9 octa for $300 ish

And split those further according to clock speed and cache. I'd expect a dual core in a phone not a desktop anymore. Also have the base models split one with no iGPU one with iGPU. Make everything overclockable. That would kill off AMD in 48hrs.

"Hey, you know that $300 CPU we have that sells so well? Let's drop the price by $100... just because."

 

[escrow4]

"Hey, you know that $300 CPU we have that sells so well? Let's drop the price by $100... just because."

Imagine how many would buy it. You make a loss that doesn't matter due to sheer volume.

 

[Yuriman]

Imagine how many would buy it. You make a loss that doesn't matter due to sheer volume.

And where would those sales come from? Intel would sell fewer $100 i3's in exchange for selling more $100 i5's? What about next gen, once this new pricing is established?

Personally, if Intel started selling 4 core CPUs at $100, I'd probably still get the dual core + HT at $50 for most of my builds. Quads are overkill for a vast majority of the tasks. I wouldn't put a quad in my dad's or sister's web-browsing PCs. My wife and I play games together, but honestly, her Haswell i3 runs all of the games we play almost exactly as well as the i5 in my machine. I wouldn't need a quad in my laptop, but rather, would prefer a lower power part to extend battery life within the same performance envelope.

I mean, yeah, it'd be lovely if Intel decided to sell me 50% more CPU at the same price, but I wouldn't suddenly buy twice as many PCs. I'd upgrade once and then it'd be back to square one.

 

[swilli89]

Nothing would happen. Current quads with HT are more than enough for gaming and multimedia. Also, Look at a current die shot of haswell; the iGPU takes up way more space than actual cpu cores. Intel profit margin would dip like 1 to 2%, would be negligible to make 6 cores standard.

Most importantly, yes we need AMD to force this into a reality by delivering a competitive solution.

 

[tenks]

Could a dual-channel MC feed 6 cores? Maybe thats a reason why Intel hasn't done it..adding more channels is more complex and expensive..That and because of the cadence, the HEDT line would become useless, or would have to move to 8c/10c or 8c/12c to be relevant. If skylake came out with a mainstream hex, no one would buy Broadwell-E.

Either way I'd love an Intel mainstream Hex. But it ain't gonna happen. Same boat as an Apple desktop (non pro) built on desktop parts .