14nm 6th Time Over: Intel Readies 10-core "Comet Lake" Die to Preempt "Zen 2" AM4

[Topweasel]

The single core boost speed falls within the 95W TDP spec. Dual core probably as well as the anandtech article actually sees higher performance within the 95W TDP spec for many benchmarks compared to the unleashed PL2 spec.

You keep harping on false advertising for the 9900K but reality is the 5GHz spec is not the issue, it's the all core boost state they advertise nowhere.

The PL2 value, or sustained power delivery, is what amounts to the turbo. This is the maximum sustainable power that the processor can take until we start to hit thermal issues. When a chip goes into a turbo mode, sometimes briefly, this is the part that is relied upon. The value of PL2 can be set by the system manufacturer, however Intel has its own recommended PL2 values.

In this case, for the new 9th Generation Core processors, Intel has set the PL2 value to 210W. This is essentially the power required to hit the peak turbo on all cores, such as 4.7 GHz on the eight-core Core i9-9900K. So users can completely forget the 95W TDP when it comes to cooling. If a user wants those peak frequencies, it’s time to invest in something capable and serious.

That doesn't sound like the manufacturers are running PL2 out of spec. PL2 is recommended to be set basically unlimted and therefore thermally limited by defualt. Setting the PL2 value is meant for OEM's to restrict the the Turbo's for their cooling and board limitations. More then likely boards, as again the idea of the turbo is short burst of extra power usage then falling back down due to cooling limitations. They can and are suggested to limit max time at full power usage as they feel comfortable. This came out with the 8700k as the big kicker that the mobo makers could be blamed for and the feature I was talking about most retail boards come with this non-overclocking naming feature I can't remember but an actual Intel spec feature, that disables time limits on Turbo (please if anyone can think of the name tell me it's still escaping me). PL2 adjustments are for OEM, retail boards are expected to be unlocked unless they have power limitations themselves (not enough VRM's).

So 5GHz max. All core boost. None of this matters if Intel always intended to have the PL2 set to unlimited. At some point the CPU runs at 66% extra power usage and heat out put for extended and if mystery feature left on unlimited time. Now that feature being on might be up to MB manufacturers. But I am guessing not. Otherwise Intel also wouldn't be suggesting a PL2 of 210w. Which circles back on everything. All of this stuff is hidden, not mentioned, not specced out, so they don't have to support these absolutely intended operational setting when it blows up in the face. Just because Intel and more like because Intel doesn't put in their information sheets doesn't absolve them of the confusion that this CPU and to a lesser extent others place on power usage and cooling requirements.

 

[IntelUser2000]

I know exactly what it is. It's not a guarentee for the end user but a spec for cooling manufacturers to build appropriate cooling solutions.

Therefore, it is a guarantee. It's a guarantee that in a significant period of time it won't go above 95W, so a heatsink solution that's rated for 95W will work like it should on a 95W CPU.

And chips like the 9900K are likely disproportionately sold to DIY people. The biggest problem is that Intel is doing this because without misleading they worry they won't have the performance crown over AMD and that'll hurt sales, and thus face the wrath of stockholders.

 

[Topweasel]

I don't have a problem with the 95W TDP and the 130W cooler spec.
I knew before the chip was released that it would be right on the edge of the cliff.

I do. This isn't a 95w CPU and Calling it a 95w CPU angers me. Which the TDP rating is always supposed to be a cooling rating in the first place. Then calling a recommended cooler a spec later just rubs me the wrong way. Specially since even that isn't enough.

I also called it as well I think I put the number at 180w if they got more than 4.5GHz all cores on it. I couldn't believe it when they announced it as a 95w CPU. It's not and there isn't enough backpedaling and loopholing and side speak that makes it ok to call this a 95w cpu. Calling it a 95w CPU to make system builders happy is a crap way screw over entry level enthusiasts.

 

[Spartak]

So 5GHz max. All core boost.

There is no such thing as 5GHz all core boost. When every argument you make is being disproved, I guess making up figures is a last, rather pathetic resort.

 

[LTC8K6]

IIRC, Intel hasn't specified an all core turbo for a while. Bolding is mine.

I refer back to the Anandtech 8700K review:

Turbo Modes

One of the interesting things to come out of our briefings with Intel was the fact that Intel made a very clear change in policy when it comes to press disclosure. When the question was asked about per-core turbo values for each of the CPUs, Intel made a clear statement first, then a secondary one when quizzed further:

“Intel will no longer provide this information”​

"We are only including processor frequencies for base and single-core Turbo in our materials going forward - the reasoning is that turbo frequencies are opportunistic given their dependency on system configuration and workloads"​

This change in policy is somewhat concerning and completely unnecessary. The information itself could be easily obtained by actually having the processors and probing the required P-states (assuming the motherboard manufacturer does not play silly tricks), so this comes across as Intel withholding information for arbitrary reasons.


Nonetheless, we were able to obtain the per-core turbo ratios for each of the new processors for our motherboard. Given Intel's statement above, it seems to suggest that each motherboard might have different values for these, with no Intel guidelines given.

 

[Topweasel]

There is no such thing as 5GHz all core boost. When every argument you make is being disproved, I guess making up figures is a last, rather pathetic resort.

Multiple sentences. I was talking about two different things. Saying neither or anything in between matters what is causing it to to hit the 160w that Anandtech saw if Intel is the one responsible for the default solution for the PL2 being 210w.

Tell me a statement that has been disproved. Please. I have shied away from making blanket statements. But lets see the facts that I laid out.

1. Intel rates this as a 95w CPU. False or not?
2. Intel recommends a PL2 of 210w for this CPU. True or False?
3. I don't think Intel reccomending their own OEM cooler that they don't actually ship with the CPU both because of it's cost and well it not actaulling being enough and part of the reason they stopped offering coolers with their higher end CPU's as a legit spec. Well this is more opinion you are free to disagree.
4. That the stark contrast between rated power usage and actual power usage at load is confusing and approaches if not goes over into deceiving. Another Opinion. But man if feels like a fact.
5. That you need a cooler approaching 200w or more in heat dissapation to keep the CPU from being pinned against it's 100c thermal limit. If statement 1. is True. Is this True or false.
6. That this is what people feared in 2004 when TDP for Intel stopped being Max power usage to somewhat heavy load power usage with them determining on their own what it was. True or false?

Did I leave something out. Or is this about me calling 3.5Ghz the base clock by accident?

 

[Topweasel]

IIRC, Intel hasn't specified an all core turbo for a while. Bolding is mine.

I refer back to the Anandtech 8700K review:

Which was done partially because of the 8700 but also probably in anticipation of this CPU. Something where guys can scream "Well Intel never said what the All core turbo clock is or power usuage at that level. It's not in spec so it's obviously mobo manufacturers fault for "overclocking"". I don't care what combination of boost and core usage makes it run at 160w. If those are the settings that Intel intended on having set as so by default. Then its not a 95w CPU.

 

[jpiniero]

Which was done partially because of the 8700 but also probably in anticipation of this CPU. Something where guys can scream "Well Intel never said what the All core turbo clock is or power usuage at that level. It's not in spec so it's obviously mobo manufacturers fault for "overclocking"". I don't care what combination of boost and core usage makes it run at 160w. If those are the settings that Intel intended on having set as so by default. Then its not a 95w CPU.

What the official specs say and what Intel encourages the board makers to put are two different things I imagine.

 

[Topweasel]

What the official specs say and what Intel encourages the board makers to put are two different things I imagine.

Where is the official specs on the PL2 setting?

 

[jpiniero]

Where is the official specs on the PL2 setting?

119W.

 

[Topweasel]

119W.

Where is this listed?

 

[jpiniero]

Where is this listed?

https://www.intel.com/content/www/u...core/8th-gen-core-family-datasheet-vol-1.html

 

[Markfw]

119W.

Yes Anandtechs REVISED numbers still say 160 watt.

 

[Topweasel]

That isn't a spec. It its a guideline (TDP*1.25) and there is a default set by Hardware. Do we actually know that Mobo manufacturers are setting the PL2 to 201w and not that Intel decided that this CPU needed a higher PL2 as note 15 suggests?

 

[jpiniero]

That isn't a spec. It its a guideline (TDP*1.25) and there is a default set by Hardware. Do we actually know that Mobo manufacturers are setting the PL2 to 201w and not that Intel decided that this CPU needed a higher PL2 as note 15 suggests?

I'm sure Intel is encouraging it but it's up to them as to what the default PL2 and time is.

 

[PotatoWithEarsOnSide]

@Topweasel
1 - the 9900k runs all core at base within 95w, comfortably. It also hits all of it's turbo specs for a finite period of time if the max Pl2 state has been set to a recommended time, and not infinite as the BIOS actually allows.
2 - Intel recommends Pl2 of Pl1*1.25 asper the link that was provided to you, and the 210w reference that you quoted was in fact a third party commentary not a listed Intel spec.
3 - I have no issues with your opinion here.
4 - power usage at load for a correctly configured 9900k is 95w, with load being defined as a 24/7 sustainable load. You get all cores at 4.2GHz for this. Turbo is not intended to be sustainable, hence the name.
5 - you misunderstood how Turbo Boost is designed to work, primarily because you base all of your comments on CPU behaviours caused by poor default settings on almost all motherboards. A heatsi k takes time to heat up to the point that it cannot stand any more input heat (power) that it can dissipate; the power draw during Turbo Boost simply affects the time it takes to reach that limit. Consequently, setting very high Pl2 can cause temperature spikes and trigger thermal throttling, and this completely overrides P-states in any event. Thermal throttling does not equal turboing down to the sustainable Pl1 state.
6 - max load as TDP was always a scientifically inaccurate metric, hence why if listed it must now conform to standards. Besides which, its usefulness as a max power figure did serve a purpose that was otherwise legitimate, which I will not deny; it helped plan for PSU choice.
What Turbo Boost does is utilise the spare heatsink dissipation capacity that running on all cores at a base of 3.6GHz allows. There would be no such Turbo Boost capacity if it didn't run well below 95w at base. The fact that it can run 24/7 at 4.2GHz within 95w tells us that there's loads of headroom. Pl2 of up to 210w is there to allow actual overclocker* to better control their Turbo Boost according to their own choice of cooling. It's not there for the mobo manufacturers to decide for us, which is precisely what they've done. The 9900k is a great CPU that meets its spec within its TDP limits, but its biggest weakness is that it doesn't come at spec by default out of the box due to choices made by mobo manufacturers. It's not worth it's price at the spec limits, but for an actual overclocker then it clearly does come at an acceptable price...for many.

* by this I mean anyone that still wants Turbo Boost functionality but does not go the whole hog of altering their base multiplier. The latter folk would be fully aware of a potentially big increase in overall power draw, and would plan their cooling accordingly. The former are folk that invest in a better cooling solution than recommended, but want to push their turbo a little harder for longer...akin to XFR2 functionality except by manually altering Pl2.

 

[epsilon84]

@Topweasel
1 - the 9900k runs all core at base within 95w, comfortably. It also hits all of it's turbo specs for a finite period of time if the max Pl2 state has been set to a recommended time, and not infinite as the BIOS actually allows.
2 - Intel recommends Pl2 of Pl1*1.25 asper the link that was provided to you, and the 210w reference that you quoted was in fact a third party commentary not a listed Intel spec.
3 - I have no issues with your opinion here.
4 - power usage at load for a correctly configured 9900k is 95w, with load being defined as a 24/7 sustainable load. You get all cores at 4.2GHz for this. Turbo is not intended to be sustainable, hence the name.
5 - you misunderstood how Turbo Boost is designed to work, primarily because you base all of your comments on CPU behaviours caused by poor default settings on almost all motherboards. A heatsi k takes time to heat up to the point that it cannot stand any more input heat (power) that it can dissipate; the power draw during Turbo Boost simply affects the time it takes to reach that limit. Consequently, setting very high Pl2 can cause temperature spikes and trigger thermal throttling, and this completely overrides P-states in any event. Thermal throttling does not equal turboing down to the sustainable Pl1 state.
6 - max load as TDP was always a scientifically inaccurate metric, hence why if listed it must now conform to standards. Besides which, its usefulness as a max power figure did serve a purpose that was otherwise legitimate, which I will not deny; it helped plan for PSU choice.
What Turbo Boost does is utilise the spare heatsink dissipation capacity that running on all cores at a base of 3.6GHz allows. There would be no such Turbo Boost capacity if it didn't run well below 95w at base. The fact that it can run 24/7 at 4.2GHz within 95w tells us that there's loads of headroom. Pl2 of up to 210w is there to allow actual overclocker* to better control their Turbo Boost according to their own choice of cooling. It's not there for the mobo manufacturers to decide for us, which is precisely what they've done. The 9900k is a great CPU that meets its spec within its TDP limits, but its biggest weakness is that it doesn't come at spec by default out of the box due to choices made by mobo manufacturers. It's not worth it's price at the spec limits, but for an actual overclocker then it clearly does come at an acceptable price...for many.

* by this I mean anyone that still wants Turbo Boost functionality but does not go the whole hog of altering their base multiplier. The latter folk would be fully aware of a potentially big increase in overall power draw, and would plan their cooling accordingly. The former are folk that invest in a better cooling solution than recommended, but want to push their turbo a little harder for longer...akin to XFR2 functionality except by manually altering Pl2.

I think Intel is at fault here as well for allowing mobo makers free reign on the PL2 durations. It's clear that they want as much as a lead over the 2700X as possible and that lead looks better at 4.7GHz than 4.2GHz.

I'm a bit confused about why there has been pages of debate surrounding TDP and P states though. As you explained, a properly configured 9900K does in fact fall into the 95W spec as proven by Gamers Nexus (and others).

This is the reason I said in my earlier post that a properly configured 10C Comet Lake chip will be able to classed as a '120W' CPU if power scales like the 9900K.

Sure, you're not going to get 5GHz all core, but 10 cores at 4.2GHz with reasonable power draw makes this a realistic prospect.

Yeah it's still 14nm, old tech blah blah. I get it. But Intel needs a response to Zen 2.

If the new i9 is indeed 10C / 20T, then maybe that paves the way for i7 @ 8C / 16T and i5 @ 6C / 12T as I predicted in another thread.

 

[PotatoWithEarsOnSide]

The debate exists primarily because getting the 9900k to work to spec is simply not an easy task, except with the Maximus XI Hero. I assume that most folk aren't buying it, or still don't know exactly what 2 BIOS settings they need to check.
Anyone that knows what they are getting will have no issues with the CPU.

 

[dacostafilipe]

Anyone that knows what they are getting will have no issues with the CPU.

Anyone that knows != Anyone that buys/builds

 

[LTC8K6]

Has anyone tested the chip with the all core turbo set to 119W?
Or maybe to the 130W cooling number?

 

[Topweasel]

Has anyone tested the chip with the all core turbo set to 119W?
Or maybe to the 130W cooling number?

These would be nice to see. I would also love some actual clarification on the 210w PL2. Because the Intel basically states in its sheet that it's the default target. But if Intel "suggests" or requires that the default be 210w it isn't in violation of their datasheet. What is the PL2 set to when some plugs in a 8700k? If it's ~113PL2, wouldn't this imply that Intel set the default for the 9900K and not every single mobo manufacturer ignoring the data sheet to illegally overclock your CPU for you?

 

[coercitiv]

Or maybe to the 130W cooling number?

Let's put this "130W TDP" cooler thing to rest:

From the Intel 9th gen family datasheet we can see the design target coolers and Tcase requirements for 8c 95W TDP CPUs:

So all 90W+ TDP SKU are required to be kept at 65C Tcase in order to meet specifications. Let's go back to my previous post on the thermal profile of the 2015D cooling solution:

When required to operate with a Tcase of 65C the 2015D cooler has a max TDP of 95W. It would only be able to reach a max TDP of 130W if it were allowed to function with a higher Tcase of 72-73C.

 

[jpiniero]

These would be nice to see. I would also love some actual clarification on the 210w PL2. Because the Intel basically states in its sheet that it's the default target. But if Intel "suggests" or requires that the default be 210w it isn't in violation of their datasheet. What is the PL2 set to when some plugs in a 8700k? If it's ~113PL2, wouldn't this imply that Intel set the default for the 9900K and not every single mobo manufacturer ignoring the data sheet to illegally overclock your CPU for you?

The 8700K's stock FCT is 4.3, so without enabling overclocking you won't draw anywhere near where the 9900K would do, given the 9900K has 8 cores and 4.7 FCT.

 

[Topweasel]

The 8700K's stock FCT is 4.3, so without enabling overclocking you won't draw anywhere near where the 9900K would do, given the 9900K has 8 cores and 4.7 FCT.

It's not about power draw. The assumption is that mobo manufacturers (all of them) are setting the PL2 to 210w just because they want Intel's CPU's to look better (someone actually said this in this thread). I know you are headfast stuck to 119w being the default for the 9900k. So that means motherboard manufacturers are not setting the board to default with the PL2 at 210w but are in fact setting the PL2 to 210w right off the bat. If that's the case boards where you can manually make the change and therefore can see what the PL2 is set to. I wonder how they react to other CPU's like the 8700k. It's not going to use as much power as the 9900k, but is it still going to be at 210w? Because if it's not 210w it would imply that they are picking up the default from the CPU. Which would mean that 210w isn't motherboard manufacturers illegally overclocking your CPU by default as has been suggested, but the actual default settings of the CPU by Intel.

A note for a diagram stating general ratio for the platform while noting that it may be changed to more efficient options by hardware. Does not make a set in stone standard on what the default is.

 

[coercitiv]

BTW, there is no mention of a 210W PL2 in Intel's datasheet.

They only refer to PL2 as being a hardware default of PL1 x 1.25, which in the case of a 95W TDP CPU would be ~120W.

From the same datasheet, the contents of Note 16:

The formula of PL2=PL1*1.25 is the hardware default but may not represent the optimum value for processor performance.

By including the benefits available from power and thermal management features the recommended
value for PL2 found in the PDG/Power Map can be higher.