Only Intel knows.
They went with 6 core because they probably thought it was enough, and/or that was the best they could get out during the short time they had to plan.
For sure all core turbos will be significantly higher than that for retail, wouldn't be surprised at 3.0ghz base though.
You do have a good point though, Intel's ring bus is much more power efficient than AMDs fabric...at low core counts.
Ryzen cores are more power efficient than skylake...making them competitive at moderate core counts despite the infinity fabric power demands.
Intel can't cram 8 skylake cores and clock them past 4ghz without blowing the power budget to pieces, 8 v 8 cores is where AMD pulls ahead in efficiency, even if coffeelake s is certainly going to be faster across the board.
I expect 4.8ghz single core turbo though, 3.0ghz base, 3.8ghz all core turbo...105w.
Do you mean alot better than skylake X or ring bus latency?
I thought mesh had a latency penalty? Perhaps they can clock it faster?
Better than Skylake-X but worse than the ring. I'm convinced that the mesh latency comes from essentially having to decide on what path to take on the mesh. Might be ways of improving that decision time beyond just higher clocking the mesh.
We can test the chip with various multi-core loads and see what frequencies it maintains for those loads, and then we know what the frequencies are with various numbers of cores loaded.Thank you.......finally some one who is willing to hang some numbers along side his "'alias"..........but what actually does all core turbo really mean?.... I just found this:
"Intel’s Coffee Lake Core i7-8700K has launched, earning our top CPU recommendation at its price point. But there’s one issue worth being aware of going forward: Intel has declared that it will no longer officially disclose its per-core Turbo frequencies. When we queried Intel why this was being changed, the company told us the following:"
[W]e’re no longer disclosing this level of detail as its proprietary to Intel. Intel only specifies processor frequencies for base and single-core Turbo in our processor marketing and technical collateral, such as ARK, and not the multi-core Turbo frequencies. We’re aligning communications to be consistent. All Turbo frequencies are opportunistic given their dependency on system configuration and workloads.
given that. I think we can drop the "all core turbo" comparisons
........how convenient is it that they no longer quote all core turbo in going from 4C 14nm+ to 6C 14nm++??..........but we do know they have to quote base frequencies.........and we know that the base frequency on 4C 14NM+ was 4.2GHz and the base frequency on the 6C 14nm++ is 3.7GHz.......How about that??..........and the base frequency of the 2700X 8C is 3.7GHz
Why bother? 6 core was good enough performance for the vast majority of users, costs far less to produce (more profit), and Intel probably had hopes to solve the 10 nm problems by now when they started the 6-core design.so what do you predict for 8 core?....(freq and power).....why did they even bother then with a 6 core if 8 core
The 8700K is 14nm++ and it doesn't have a 4.0 base clock, so I think we can rule that right out for an 8 core 14nm++ chip.Why bother? 6 core was good enough performance for the vast majority of users, costs far less to produce (more profit), and Intel probably had hopes to solve the 10 nm problems by now when they started the 6-core design.
As for 8 core predictions, Intel has three possibilities assuming that an 8-core chip is basically based on the 8700K chip:
1) Increase power requirements.
2) Decrease base clock speeds.
3) Find incremental power savings.
Note: these three items can be done in any combination. Also, to simplify this post, I'll assume that Intel is keeping the power at 95 W.
Low Estimate (decreased base clock):
If Intel cannot find any incremental power savings over the 8700K, then the base clock for an 8-core chip will have to drop. In this case it will be 2.8 GHz. Turbo speeds are arbitrary as long as you allow for thermal throttling, so I'll assume in this case that turbo speeds are the same as the 8700K.
High Estimate (incremental power savings):
Intel once claimed that they can get 52% power savings by going from 14 nm to 14 nm++. https://g.foolcdn.com/editorial/images/438596/perf-enh_large.png Assuming that Intel can meet this VERY lofty goal, then the 8-core chip can run the same speeds as the 4-core 6700K chip. That would mean a 4.0 GHz base.
Likely Estimate (somewhere in between):
The 8700K was disappointing since it was only 25% more power efficient than the 14 nm 6700K chip, not the full 52% that Intel once claimed that 14 nm++ would be. But, the rumored Xeon E-2176G mentioned above https://en.wikichip.org/wiki/intel/xeon_e/e-2176g is 37% more power efficient. That seems like a reasonable compromise, short of their 52% goal but better than what they did in year 2017. That would put the 8-core chip at 3.3 GHz base speeds.
Die layout would matter more. If the 8C variant would be a simple "insert 1 more module" approach starting from the current layout, heat management would arguably get worse, not better.Does the bigger die play any factor with a larger surface area to remove heat?
The 8700K is 14nm++ and it doesn't have a 4.0 base clock, so I think we can rule that right out for an 8 core 14nm++ chip.
The base clock is going to be lower than 3.7, barring something we don't know about Intel and 14nm++.
Does the bigger die play any factor with a larger surface area to remove heat?
Presumably the 2176G is also 14nm++, yet it has a considerably lower TDP than the 8700K, despite all other specs being the same.
So something is different between those two chips. But what? They seem identical save for ECC support.
Is the 2176G a better stepping or something?
Die layout would matter more. If the 8C variant would be a simple "insert 1 more module" approach starting from the current layout, heat management would arguably get worse, not better.
It also isn't unlocked. It does seem to support the same 2666 ram.Maybe because lower ram support, better binned, or aggressive power management(meaning it will throttle more often).
Why bother? 6 core was good enough performance for the vast majority of users, costs far less to produce (more profit), and Intel probably had hopes to solve the 10 nm problems by now when they started the 6-core design.
As for 8 core predictions, Intel has three possibilities assuming that an 8-core chip is basically based on the 8700K chip:
1) Increase power requirements.
2) Decrease base clock speeds.
3) Find incremental power savings.
Note: these three items can be done in any combination. Also, to simplify this post, I'll assume that Intel is keeping the power at 95 W.
Low Estimate (decreased base clock):
If Intel cannot find any incremental power savings over the 8700K, then the base clock for an 8-core chip will have to drop. In this case it will be 2.8 GHz. Turbo speeds are arbitrary as long as you allow for thermal throttling, so I'll assume in this case that turbo speeds are the same as the 8700K.
High Estimate (incremental power savings):
Intel once claimed that they can get 52% power savings by going from 14 nm to 14 nm++. https://g.foolcdn.com/editorial/images/438596/perf-enh_large.png Assuming that Intel can meet this VERY lofty goal, then the 8-core chip can run the same speeds as the 4-core 6700K chip. That would mean a 4.0 GHz base.
Likely Estimate (somewhere in between):
The 8700K was disappointing since it was only 25% more power efficient than the 14 nm 6700K chip, not the full 52% that Intel once claimed that 14 nm++ would be. But, the rumored Xeon E-2176G mentioned above https://en.wikichip.org/wiki/intel/xeon_e/e-2176g is 37% more power efficient. That seems like a reasonable compromise, short of their 52% goal but better than what they did in year 2017. That would put the 8-core chip at 3.3 GHz base speeds.
I suspect that if Intel allowed a greater temp and specified a better heasink or AIO they could run an eight-core chip built on the same process as the 8700k.
You somewhat answered your own post. Is the 8700K truly living up to the 14nm++ potential? I don't think so. If 14 nm++ was really supposed to be 52% more power efficient than 14nm, then why does adding 2 cores require a base clock speed drop? And why can the 2176G do it with so much less TDP? I suspect that Intel was being highly conservative with the base speed to ensure high yields when they were having trouble producing enough 8700K chips to meet expected demand for this rushed chip.The 8700K is 14nm++ and it doesn't have a 4.0 base clock, so I think we can rule that right out for an 8 core 14nm++ chip.
Presumably the 2176G is also 14nm++, yet it has a considerably lower TDP than the 8700K, despite all other specs being the same.
If we lock the 8700K what would it's TDP be? Would it be lower?You somewhat answered your own post. Is the 8700K truly living up to the 14nm++ potential? I don't think so. If 14 nm++ was really supposed to be 52% more power efficient, then why does adding 2 cores require a base clock speed drop? And why can the 2176G do it with so much less TDP? I suspect that Intel was being highly conservative with the base speed to ensure high yields when they were having trouble producing enough 8700K chips to meet demand.
Likely Estimate (somewhere in between):
The 8700K was disappointing since it was only 25% more power efficient than the 14 nm 6700K chip, not the full 52% that Intel once claimed that 14 nm++ would be. But, the rumored Xeon E-2176G mentioned above https://en.wikichip.org/wiki/intel/xeon_e/e-2176g is 37% more power efficient. That seems like a reasonable compromise, short of their 52% goal but better than what they did in year 2017. That would put the 8-core chip at 3.3 GHz base speeds.
Yes they no longer advertise, but we pretty much know what they are, we can't pretend they don't exist.Thank you.......finally some one who is willing to hang some numbers along side his "'alias"..........but what actually does all core turbo really mean?.... I just found this:
"Intel’s Coffee Lake Core i7-8700K has launched, earning our top CPU recommendation at its price point. But there’s one issue worth being aware of going forward: Intel has declared that it will no longer officially disclose its per-core Turbo frequencies. When we queried Intel why this was being changed, the company told us the following:"
[W]e’re no longer disclosing this level of detail as its proprietary to Intel. Intel only specifies processor frequencies for base and single-core Turbo in our processor marketing and technical collateral, such as ARK, and not the multi-core Turbo frequencies. We’re aligning communications to be consistent. All Turbo frequencies are opportunistic given their dependency on system configuration and workloads.
given that. I think we can drop the "all core turbo" comparisons
........how convenient is it that they no longer quote all core turbo in going from 4C 14nm+ to 6C 14nm++??..........but we do know they have to quote base frequencies.........and we know that the base frequency on 4C 14NM+ was 4.2GHz and the base frequency on the 6C 14nm++ is 3.7GHz.......How about that??..........and the base frequency of the 2700X 8C is 3.7GHz
I'm thinking it will be a mesh topology also, skylake X was actually pretty good with higher clocks to the mesh, what effect that will have with power?Better than Skylake-X but worse than the ring. I'm convinced that the mesh latency comes from essentially having to decide on what path to take on the mesh. Might be ways of improving that decision time beyond just higher clocking the mesh.
I don't think just having overclocking feature itself adds any power of note until you use it. The chip will use all the power that it can use given its thermal environment. Advertised TDP is at base clock speed with everything locked down, so your question is a bit confusing.If we lock the 8700K what would it's TDP be? Would it be lower?
How much power does an 8700K use if you leave it stock?
Back in the day we had a thing with CPUs called a "stepping".
Often, a later stepping of a chip performed better.
Today, we don't seem to have different steppings of the same chip.
True, but for a rough estimate, it is close enough.You can't just interpolate linearly though. Lower frequencies require linearly lower power, but also run at a lower voltage, regardless of core count, which scales with the square. So speeds would end up somewhat higher. This is one of the core benefits of multicore processors in the first place.
Intel claimed 52% less power OR 26% more performance, and the 8700k had higher single core and multi core boost clocks and more cores vs its 14nm(non plus) predecessor(6700k) so they chose a mix of more performance and lower power.The 8700K was disappointing since it was only 25% more power efficient than the 14 nm 6700K chip, not the full 52% that Intel once claimed that 14 nm++ would be.
Intel claimed 52% less power OR 26% more performance, and the 8700k had higher single core and multi core boost clocks and more cores vs its 14nm(non plus) predecessor(6700k) so they chose a mix of more performance and lower power.
Also if you look at skylake-X the home agents in each core are massive, upping the clocks in the mesh is likely to be very expensive power wise.The mesh latency comes from the fact that the mesh is clocked substantially lower than the rings on the mainstream CPUs. The lower mesh frequency also means slower L3 cache/higher L3 latency because the L3 clock is tied to the frequency of the mesh. Nothing more.
To answer your second question first, there's no longer a "stock" as we're used to. Mobo makers get a spec sheet from Intel and they implement it as they see fit, within some thresholds. One MB might produce a stock 8700K with no power limit at all and uncore running at 4300Mhz, another may enforce 95W TDP and 3700Mhz uncore. The same has recently become true for AMD as well, so we might say this is the norm nowadays: we can no longer expect "Auto" settings to give us the stock experience and numbers. (clocks, TDP, etc)If we lock the 8700K what would it's TDP be? Would it be lower?
How much power does an 8700K use if you leave it stock?