Skylake (Mainstream) CORE count Speculation Poll

Skylake (Mainstream) CORE count Speculation Poll

  • 4 Cores (8 threads)

  • 6 Cores (12 threads)

  • 8 Cores (16 threads)

  • Something else (ideally, please explain, below)

  • SKY top, LAKE below, therefore CORES = Middle = 0.17265 recurring


Results are only viewable after voting.

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
Poll question is, What do you think the maximum number of cores in the upcoming Skylake (Intel) mainstream processor line, going to be ?

Since some source(s) seem to say that mainstream Skylake (Intel) will start at Quad core, and Haswell-E is going to be up to 8 core, will we see Skylake produce the first (mainstream, non-extreme) hex (or even 8) core desktop cpus ?

(Yes, I know extreme versions will go above quad core already, but I want them to be the latest architecture, and NOT cost excessive amounts of money for the cpu, motherboard and potentially other bits).

(Also, I know that AMD had the (now eventually out-going) FX 8 cores, and Atom goes to 8 cores, but I want/mean something which can give 50% (approx) more performance than a haswell 4771 (BEFORE any overclocking), IF the software being run is suitably multi-threaded).

http://en.wikipedia.org/wiki/Skylake_(microarchitecture)
 
Last edited:

jpiniero

Lifer
Oct 1, 2010
15,660
6,135
136
It's going to be 4. The earliest for more than 4 would be after Cannonlake, but even that seems extremely unlikely.

but I want/mean something which can give 50% (approx) more performance than a haswell 4771

$999
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
It's going to be 4. The earliest for more than 4 would be after Cannonlake, but even that seems extremely unlikely.



$999

Thanks for the reply.
Given that it is probably going to be another 15 .. 18(+) months before Skylake is first available (to buy), and the ever increasingly long delays between new Intel Architectures and/or size reductions (e.g. 14 nm has already been delayed, probably due to challenges getting it that small), we could be talking about many, many years to get to "after Cannonlake".
Which still may only be quad core.

tl:dr; I DON'T want to wait (approx) 5 years for Intel to exceed quad core in their mainstream cpus, or even to find (in 5 years time) that they are STILL quads.

"$999"
Approximately three times the cost (4771), to get the "OLD" generation hardware, missing the latest new instruction set features, and (presumed) IPC (instruction per clock) speed gains, and built on the "OLD" larger feature size (22 nm) hence worsening power consumption and potential clock speed improvements. Even a basic IGP is missing, so workstation/serious use PCs have to have extra expense/power consumption of an additional graphics card.

As time goes on, games can/will be increasingly able to usefully utilize more than 4 cores, especially as modern day games consoles are already at about the 6 core (usable directly for games, 8 in total) level.
 
Last edited:
Aug 11, 2008
10,451
642
126
Personally, I would rather intel already had made a mainstream hex core instead of devoting so much die space to the igp, but obviously the market sees it differently. The problem is you usually have to sacrifice some clockspeed to keep the TDP down when adding more cores, and there are still not enough well threaded apps to take advantage of more than 4 cores. So I dont really see a mainstream hex core in intel's future. Maybe if AMD suddenly came out with an 8 core cpu with per core performance close to intel, intel would move agressively, but I dont see that happening either.
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
Or the lake is part of the sky because of clouds, so Skylake will evaporate. D:

But because Skylake (might) only have 4 cores, it is NOT suitable for cloud (computing centres).

Also if Skylake evaporates, it may h'ARM' the environment, (h)aswell as the PC industry as a whole.
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
Personally, I would rather intel already had made a mainstream hex core instead of devoting so much die space to the igp, but obviously the market sees it differently. The problem is you usually have to sacrifice some clockspeed to keep the TDP down when adding more cores, and there are still not enough well threaded apps to take advantage of more than 4 cores. So I dont really see a mainstream hex core in intel's future. Maybe if AMD suddenly came out with an 8 core cpu with per core performance close to intel, intel would move agressively, but I dont see that happening either.

I guess one positive way of looking at Skylake (and beyond), is that even if it is only quad core (mainstream), the combination of the improved Ipc (hopefully), and the possible higher efficiency hyper-threading (i.e. HT makes the cpu act as if it is more like a >4 core cpu, as the % speed benefits of HT increase, if the greater availability of on-chip processing resources increases *theoretical speculation here*).

I would speculate that another significant issue is that if mainstream Intel processors were "too powerful", e.g. 6/8 cored at $300, then that would badly hit the highly profitable Xeon (Server) market.

That is, why pay a huge $ price for a high core count Xeon (Skylake) part, when a cheap mainstream 6/8 core Skylake cpu at e.g. $300 could replace it.
 
Last edited:

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
They should go Volvo and offer 5 core units.

I'd really like that (assuming they were hex/octo cored Xeon parts), and the faulty core(s) are disabled (fused off). As long as the price was suitably reduced to make it a mainstream part.

(Ideally leave out the other Volvo eccentricities).

I want the "straight six", core version!.
(But the V8 cored version would be even better).
 
Last edited:

know of fence

Senior member
May 28, 2009
555
2
71
It will be -say- 192 graphics cores and 4/8 CPU cores. So 200 Cores total?
- 4K Resolution requires both lots of graphics processing and memory bandwidth, yet all these things share space on the die, thus the core count of x86 cores won't increase at the expense of graphics.
- There are massive advantages to be had from heterogeneous memory access (basically we'll see Intel's version of HSA and huma), thus an even bigger focus on integrated graphics.
- For general application there is a balance to be had between CPU and GPU compute, parallel tasks like any kind of encoding are better served with infinitely scalable graphics cores rather than more big CPU cores. I'm assuming graphics compute is more power efficient, due to the lower clocks of GPU cores.
- Increasing the number of cores is a good second or third step after process miniaturisation comes to an end or Intel hits a massive road block in their node development, there is no reason to expect it from their regular cadence.

Is there even room between GPU compute and general 8 thread computing for some kind of special sufficiently complicated tasks? It would probably have to be something that is "real time".
 
Last edited:

beginner99

Diamond Member
Jun 2, 2009
5,285
1,710
136
Personally, I would rather intel already had made a mainstream hex core instead of devoting so much die space to the igp, but obviously the market sees it differently.

Same here. And going lga 2011 is often not worth it due to much higher cost and usually getting and older and almost outdated platform.
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
It will be 192 graphics cores and 4/8 CPU cores. So 200 Cores total?
- 4K Resolution require both lots of graphics processing and memory bandwidth, yet all these things share space on the die, thus the core count of x86 cores won't increase at the expense of graphics

Unfortunately graphics processing (used as compute engines, e.g. OpenCL) is only suitable for some of the computing tasks that we currently may want to do.
It also can need an enoughmous amount of software development, to efficiently implement existing software on a graphics card/Igp.

So for some stuff (I agree), a suitably powerful graphics card(s) and/or Igp is the way to go.

But for the rest of the possible general software tasks that we want to do, we still need the fastest cores (Ipc), with as many cores on-chip as we can afford.

It might be possible that eventually in the future, gpu compute cores become increasingly flexible, and improved software techniques/compilers come along, which increase the percentage of applications which can run best on a high end graphics card and/or Igp.

I'd prefer Intel to have at least two different CPU dies, one with say a quad CPU and a greatish Igp, for mainstream non-serious gamers, and another, with a much more basic Igp, using the freed up die space for more cores.

The latter CPU would be the favourite for serious non-gaming users AND serious gamers (who would be using very powerful gaming graphics cards anyway, until eventually Igp become competitive with even the high end graphics cards, if that time ever comes).

Keep at 4 cores, get the TDP low enough to fit into a tablet :D

The Intel (Israel) team who I believe are doing Skylake, previously did core2duo (I believe), which I think was one of the best ever architectural CPU improvements done by Intel (in my opinion). Reportedly it was SO good, because they were VERY keen on the upcoming mobile market(s), which was why core2duo was so amazingly power efficient, and yet a great (Ipc) and (over)clocker as well.

So if we end up with another core2duo (like improvement), that will NOT be a bad thing, in my opinion.

4 cores but they'll probably up it after the next tock after Skylake whatever what that will be.

But by then I'll have to use one of my walking sticks to push the keyboard, the other to push the mouse, and I'll have forgotten what a computer is. Maybe the care home nurse will let me use her 512 core Arm mobile phone instead ?
 
Last edited:

sm625

Diamond Member
May 6, 2011
8,172
137
106
Given the fact that intel could easily squeeze four i5 lynnfields (16 cores) into the same space as one of these 14nm chips, they deserve another year of $4+ billion losses in their mobile division if they do not make it at least 8 cores. Greed knows no bounds....
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
Given the fact that intel could easily squeeze four i5 lynnfields (16 cores) into the same space as one of these 14nm chips, they deserve another year of $4+ billion losses in their mobile division if they do not make it at least 8 cores. Greed knows no bounds....

The thing is, sooner or later, we may see competitors, such as Arm variants, with 8, 12 or even 16 cores, at (hopefully) reasonable prices. Such as Amds upcoming new range of Arm based server chips (prices NOT publicly known yet, as far as I am aware), with at up to at least 8 cores initially, with 16 being possible in the future (I guess).
With suitably optimised operating systems, and other software (in time), could end up being a significant competitor, as the desktop/server machines could be VERY economically priced (in theory).

E.g The raspberry PI computer which is Arm based (although very slow compared to desktop Intel processors), has seen a lot of applications, because of its tiny price, almost non-existent power consumption, ready availability of lots of software for it and tiny size, complete with total silence due to it being Fanless (unless the user adds their own fans).

If the raspberry PI MK ??? or something similar had 8 powerful later Arm cores (e.g. arm Cortex A57), at a sensible price and power consumption, I would expect it to sell extremely well.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Given the fact that intel could easily squeeze four i5 lynnfields (16 cores) into the same space as one of these 14nm chips, they deserve another year of $4+ billion losses in their mobile division if they do not make it at least 8 cores. Greed knows no bounds....

You confuse greed and your own desires. And you can soon buy 8 core Haswell-E.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
The thing is, sooner or later, we may see competitors, such as Arm variants, with 8, 12 or even 16 cores, at (hopefully) reasonable prices. Such as Amds upcoming new range of Arm based server chips (prices NOT publicly known yet, as far as I am aware), with at up to at least 8 cores initially, with 16 being possible in the future (I guess).
With suitably optimised operating systems, and other software (in time), could end up being a significant competitor, as the desktop/server machines could be VERY economically priced (in theory).

E.g The raspberry PI computer which is Arm based (although very slow compared to desktop Intel processors), has seen a lot of applications, because of its tiny price, almost non-existent power consumption, ready availability of lots of software for it and tiny size, complete with total silence due to it being Fanless (unless the user adds their own fans).

If the raspberry PI MK ??? or something similar had 8 powerful later Arm cores (e.g. arm Cortex A57), at a sensible price and power consumption, I would expect it to sell extremely well.

ARM chips are still horrible slow. And they dont scale well.

Its actually going the other way around, x86 entering ARM home field.
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
ARM chips are still horrible slow.

Well I've spent all today (when computing), EXCLUSIVELY using arm based computer(s), specifically Apple, and it has accomplished everything I wanted to do, just great. (Ios/IPad, somewhat recent version, but NOT the latest one).

I've been watching music videos, gaming (simpler games than consoles, but still very addictive), and web surfing.

E.g. All my forum activities today have been done using Apple arm device(s).
Including this reply.

Although I'm NOT trying to say that arm is currently faster or more powerful than Intel's offerings, it has performed entirely satisfactorily.

And they dont scale well.

Its actually going the other way around, x86 entering ARM home field.

True.

But the situation is that year on year (in my opinion), arm stuff seem to be getting better and better, and over time seem to be gradually catching up.
 

Insomniator

Diamond Member
Oct 23, 2002
6,294
171
106
I don't see why'd they go more than 4 for mainstream. Unless someone catches up rapidly, or some obvious need that will sell people on it comes up.

They aren't going to release an 8 core i5 chip just because a bunch of nerds on tech forums want them. The more likely thing to me would be similar performance with much lower power. They can still sell 'green' and battery life, but I don't think anyone is desperate for performance with modern dual and quad core chips.
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
I don't see why'd they go more than 4 for mainstream. Unless someone catches up rapidly, or some obvious need that will sell people on it comes up.

They aren't going to release an 8 core i5 chip just because a bunch of nerds on tech forums want them. The more likely thing to me would be similar performance with much lower power. They can still sell 'green' and battery life, but I don't think anyone is desperate for performance with modern dual and quad core chips.

The "Mainstream" users I was thinking of, was the sort of user who buys an I7-4771 (currently), so the higher end of mainstream. Either because they can easily afford a "good" computer, or they are somewhat serious about gaming, and/or they do a fair quantity of stuff such as video editing/conversion, or other similarly heavy computing tasks, which relatively normal users can need to perform.

But many of the above users are relatively happy to pay the extra couple of hundred dollars (approx) to go for a decent I7-4771 based computer, or similar.
But who would flatly rule out spending another 500 or 1000 dollars (approx, the rest of the computer costs more as well as the CPU, hence $500 to $1000 more estimate given), to upgrade to the hex cored extreme chips (later to be 8 cored).

It seems to be a sort of chicken and egg situation here. The sooner mainstream CPUs have more than 4 cores in them (e.g. 6), the sooner operating systems, and the bulk of the software we use, can exploit the extra potential computing power.

Really we should have had reasonably affordable hex cored CPUs (from Intel) by now (which Amd actually released what seems like a long time ago now, e.g. The 1090 hex cored phenom series).

(Higher end Xeons seem to be talking about huge numbers, I'm not sure about the latest announcements, but it was something like 18 cores in a CPU, but at a crazy high price for most non-business users, I'm sure).
 

witeken

Diamond Member
Dec 25, 2013
3,899
193
106
Well I've spent all today (when computing), EXCLUSIVELY using arm based computer(s), specifically Apple, and it has accomplished everything I wanted to do, just great. (Ios/IPad, somewhat recent version, but NOT the latest one).

I've been watching music videos, gaming (simpler games than consoles, but still very addictive), and web surfing.

E.g. All my forum activities today have been done using Apple arm device(s).
Including this reply.

Although I'm NOT trying to say that arm is currently faster or more powerful than Intel's offerings, it has performed entirely satisfactorily.
That's nice, but subjective feelings aren't useful when talking about absolute, measurable facts. Maybe a phone might still be very slow but software has been optimized and/or simplified to make the experience satisfactorily.



True.

But the situation is that year on year (in my opinion), arm stuff seem to be getting better and better, and over time seem to be gradually catching up.
The smartphone market is a fast growing market where a lot of money can be earned. Since the CPU is one of the distinguishing features of a phone and faster CPUs were necessary, there have been a lot of investments in making mobile CPUs faster. Before that, both volume and ASPs were lower, resulting in inferior CPUs.

Today, every effort is made by Qualcomm, Nvidia, Apple and Intel to squeeze the most performance out of a given TDP. So in the end, the CPU with the best transistors will win, independent from ISA.