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

[PotatoWithEarsOnSide]

It isn't a switch that can be set in any of the various BIOSes, except the Maximus XI Hero.

 

[coercitiv]

It isn't a switch that can be set in any of the various BIOSes, except the Maximus XI Hero.

It's a switch - that can be set on any decent Z board. (well, technically, it's a number of switches)

How else do you reckon overclocking would work? There must be a way to increase power limits, current limits.

 

[John Carmack]

Going by official TDP and not knowing about the motherboard shenanigans, even enthusiast users can get tossed a nasty surprise if they were trying to use something like a PicoPSU.

 

[PotatoWithEarsOnSide]

Yes, but a number of switches does not equate to easy to configure for those that don't know what they are doing.
Clearly, such people do get sold to in an inappropriate manner; they would, and should, expect that what they buy will operate to spec as a default.

 

[TheGiant]

Yup, as I said: this comes back and bites Intel where only infrared shines.

They should have handled this TDP conflict the same way they handled XMP memory: make it a switch in BIOS and advertise the higher performance numbers for a win-win scenario. I'm actually surprised AMD marketing people knew to present this better in the form of XFR, which if you think about it, is exactly what Intel wanted: TDP violation cheered by consumer.

The consumers want this: more performance for more power is a trade-off many are willing to make - as long as it's done transparently. The problem was never the power consumption per se, but rather lying about power consumption, and leaving the press to explain it to consumers, which they did poorly as they were blindsided and left in the dark. Not the brightest PR strategy imho.

Personally I'll patiently wait for the Zen 3000 and the approval of the community for whatever XFR brings to the table when pushing the CPU beyond stock. I bit my tongue this year and openly agreed that power consumption is not relevant for enthusiasts (at least less relevant than performance), I hope next year doesn't surprise us with a 180 degree spin on CPU power usage and the importance of stock TDP ratings in reviewing competing products on equal terms.

In fact, we have a good examples of power vs TDP (i5-8400, R1700 no way 65W) and the sorry I must say allround praised 2700x (which consumes with the chipset more than its advertised TDP in watts power, lol).
I simply don't get it limiting the 9900K to 95W and not the 2700X to 95W, its 105W with 9900K to the same, and next increasing the limit.
What tells me the comparison of a top desktop cpu which can do 5GHz with 8C (with 200W) at 95W and the others are limited by....nothing.

Either I don't get it and my cable from my ass to my brain is too long or this is pure politics

 

[TheGiant]

In fact, we have a good examples of power vs TDP (i5-8400, R1700 no way 65W) and the sorry I must say allround praised 2700x (which consumes with the chipset more than its advertised TDP105W comparing to 9900K+chipset in watts power, lol).
I simply don't get it limiting the 9900K to 95W and not the 2700X to 95W, its 105W with 9900K to the same, and next increasing the limit.
What tells me the comparison of a top desktop cpu which can do 5GHz with 8C (with 200W) at 95W and the others are limited by....nothing.

Either I don't get it and my cable from my ass to my brain is too long or this is pure politics

 

[jpiniero]

As it relates to the 10 core, I imagine the base clock will still be 3.6, assuming some improvement in transistor quality. Some amount less if not. With PL2 being 250 or more.

 

[PotatoWithEarsOnSide]

The problem is that people rely on listed specs in order to adequately cool their systems. With CPUs defaulting to out of spec, this can cause huge problems.
For clarity, no-one is moaning about the 9900k's performance, or its power usage when at peak performance. The problem is the lack of choice over what should be expected behaviour.

 

[LTC8K6]

The problem is that people rely on listed specs in order to adequately cool their systems. With CPUs defaulting to out of spec, this can cause huge problems.
For clarity, no-one is moaning about the 9900k's performance, or its power usage when at peak performance. The problem is the lack of choice over what should be expected behaviour.

The CPU is not defaulting to "out of spec" though. The motherboard is the boss.
The cooling spec is 130W, which is in line with the CPU specs as far as I can see.

 

[LTC8K6]

DP

 

[Topweasel]

The CPU is not defaulting to "out of spec" though. The motherboard is the boss.
The cooling spec is 130W, which is in line with the CPU specs as far as I can see.

Power usage is specced at 90w. All that Intel marks as 130w, is what OEM Reference cooler they recommend. But that isn't clear. Even the 130 W AIO that intel recommends wouldn't have let Anandtech meet top clocks.

 

[Spartak]

Going by official TDP and not knowing about the motherboard shenanigans, even enthusiast users can get tossed a nasty surprise if they were trying to use something like a PicoPSU.

Exactly. People keep focussing on heatsinks not able to handle the turbo output power, which certainly is a PITA without much needed "TDP2" guidance from Intel. If there is an official turbo TDP people could buy a heatsink that can handle such output. Intel needs to fix this.

But for SFF builders like myself, we also have to consider the PSU. F.i., my current power supply is rated at 150W with the 12V rail around 96W sustained. The 9900K would cause serious wreckage if I dont change the PL2 state to 95W.

However, the fact you can set PL1 and PL2 TDP, tau and even max. speeds for 1 to n cores seperataly, is a tweakers heaven that AMD simply does not offer for Ryzen.

I'm buying a 8600k these coming weeks and will be tuning it to ~65W power state in a per core way I couldnt possibly dream for with a Ryzen.

So yes, Intel screwed up the marketing with turbo boost 2.0 / coffee lake v2, but the way you can tweak it to your own spec is a godsend.

 

[Topweasel]

The advertised base clock speed is 3.6.

No Stated base clocks is 3.6 (sorry said 3.5 earlier). Advertised Clock speed is 5GHz. Web address says 5GHz. First posted clock speed is 5 GHz (base clocks stated shortly after). Advertising slides from intel on Newegg does it the same way.

I am not saying Intel isn't being truthful. It's not like most of the information isn't out there somewhere in Intel's site and in some of the promotional info. But you have to dig around and the only thing out there that actually tells a semi truthful cooling power recommendation exists only on ARK and again not as general cooler ratio, just Intel telling you that if they packaged a cooler with it, this Intel OEM cooler would be the one they would have gone with. Which I think they went with 130 not because it covers the Turbo spec (because it doesn't), but because they know their crappy over rated OEM coolers at 95w wouldn't even generally allow the processor to run at base spec. Also I don't know if they have an intermediary between 90w (which is lower than the CPU is rated) and the 130w AIO. So they have to list the AIO at well above the 95w rating.

As for PL2 and XFR2. While they may come off as a factory overclock because of how close to silicon or thermal limitations. Neither is overclocking. 5GHz on this and the I believe 4.3 on a 2700x are part of the spec. Part of the spec that requires extra cooling outside TDP. The 2700x just barely and this one by a country mile. But AMD makes it clear the XFR2 if left enabled is extra for people with additional cooling room and you have a good idea what that is and it isn't much over the CPU's rating. That information has gotten much much harder to get from Intel specially when they stopped listing anything but the top single core Turbo rating and base rating.

 

[IntelUser2000]

slippery slope people saw coming back when Intel stopped measuring TDP as max power usage with the Pentium 4.

"History doesn't repeat, but it rhymes".

They are doing the same thing they did nearly 20 years ago for nearly the same reason. They knew Netburst was a dead end architecture but with no alternative in sight they had to do something.

Gamers Nexus (and others; I credit them because they were the first to highlight the functionality properly).
The issue that you are encountering, which is effectively ignoring Turbo Boost in favour of being automatically overclocked above TDP, is not a function of Intel's marketing. It is down to mobo makers ignoring Intel spec.

Want to say 2 things:
1. Kudos to GamerNexus for standing out and being a proper journalistic site rather than bowing down to companies(Intel and Nvidia) because they provide the fat cash.
2. If it's clocked like an overclocked chip, and it uses power like an overclock chip, then it probably is a (factory)overclocked chip.

I am not saying Intel isn't being truthful. It's not like most of the information isn't out there somewhere in Intel's site and in some of the promotional info.

You don't have to act so nice. Just tell like it is. In legal terms, sure they are not "lying". But its quite obvious within that limit they are doing everything to not tell you how it really is. This is also why there's such hate towards lawyers.

 

[Topweasel]

The actual text on Ark: "Intel Reference Heat Sink specification for proper operation of this SKU"

But they don't actually list a heatsink specification do they? No they list the closest Intel OEM Heatsink. Which is still well below the power usage Anandtech saw at load. It's not a spec if its just listing the next closest Heatsink Intel pretends to offer.

 

[Hans Gruber]

If you are going to spend $500-600. Do you think those people care much about power usage?

 

[Topweasel]

"History doesn't repeat, but it rhymes".

They are doing the same thing they did nearly 20 years ago for nearly the same reason. They knew Netburst was a dead end architecture but with no alternative in sight they had to do something.



Want to say 2 things:
1. Kudos to GamerNexus for standing out and being a proper journalistic site rather than bowing down to companies because they provide the fat cash.
2. If it's clocked like an overclocked chip, and it uses power like an overclock chip, then it probably is a (factory)overclocked chip.



You don't have to act so nice. Just tell like it is. In legal terms, sure they are not "lying". But its quite obvious within that limit they are doing everything to not tell you how it really is.

I don't really understand how it's on the Mobo manufacturers that are at fault. Outside maybe using a specific non-overclocking sounding Intel option (God why can't I remember the name) to allow it to run at full speed indefinitely. Which Intel wouldn't have allowed that to be an option if they didn't want all the reviewers to review the CPU at full speed all the time. 5GHz is still part of the spec and a selling point of the CPU.

 

[Topweasel]

If you are going to spend $500-600. Do you think those people care much about power usage?

That's missing 90% of the conversation. It's not power usage as a cost (though people do care). It's not even about cooling cost. Its about the difference in rating and actual power usage as it pertains to cooling and the lack of clarity on the proper solution (even if we recognize the proper one).

 

[scannall]

If you are going to spend $500-600. Do you think those people care much about power usage?

Why yes, yes I would. If reality, rather than the marketing dictates expensive liquid cooling or very loud air then I'm not buying it. I value quiet very much. And running liquid in my computer case creeps me out.

 

[Topweasel]

You don't have to act so nice. Just tell like it is. In legal terms, sure they are not "lying". But its quite obvious within that limit they are doing everything to not tell you how it really is. This is also why there's such hate towards lawyers.

Fine. I have been avoiding it since it's an Intel thread and I do have a bias a bit towards AMD in general. But yeah if not outright lying. They are trying their hardest to hide the important stuff so they can put a big fat 5GHz sticker on the side of the box. The CPU is great and shouldn't need this crap to sell. Offer it as a 4.4 4.5 GHZ 110W CPU and I think it still sells through the roof. But man this is shameful.

 

[jpiniero]

I don't really understand how it's on the Mobo manufacturers that are at fault.

Because the board is in charge as to what PL2 is and how long the PL2 lasts.

No Stated base clocks is 3.6 (sorry said 3.5 earlier). Advertised Clock speed is 5GHz. Web address says 5GHz. First posted clock speed is 5 GHz (base clocks stated shortly after). Advertising slides from intel on Newegg does it the same way.

Which is true. If you go look at AT's review it does hit 5 Ghz at 95W just fine on one core.

 

[Spartak]

Fine. I have been avoiding it since it's an Intel thread and I do have a bias a bit towards AMD in general. But yeah if not outright lying. They are trying their hardest to hide the important stuff so they can put a big fat 5GHz sticker on the side of the box. The CPU is great and shouldn't need this crap to sell. Offer it as a 4.4 4.5 GHZ 110W CPU and I think it still sells through the roof. But man this is shameful.

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.

 

[IntelUser2000]

The single core boost speed falls within the 95W TDP spec.

The review for the 9900k mentions that the 6700k might run 20w higher but their review for that doesn't show it and I believe that it was mostly an AVX issue.

Spartak, let me remind you what TDP meant. Now I do remember, it was Kabylake(7700K for you enthusiasts) that started this debacle. It wasn't a problem, because AVX scenarios were rare, except those running LinX benches.

You could, run nearly every application on it and it won't go above 95W. Sure it would have for seconds to minutes, but never on sustained load. Even with AVX. With AVX it would just run at lower clocks.

So TDP was a guarantee. Your free Intel HSF would adequately handle it because the CPU didn't deviate from that. Yea, with the stock HSF it might run at 75C at load but that's acceptable. You'll also notice motherboard manufacturers coincidentally(or not so coincidentally) didn't violate the specs. Of course you could always go in and change the settings, but at stock TDP = guarantee.

Now they want you to have 2 fan Noctua HSFs that's as heavy as some ITX graphics or water cooling just to run it at stock, if you want the performance they advertise from many review sites that are too afraid to call Intel out.

 

[Spartak]

Spartak, let me remind you what TDP meant.

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. Lack of TDP Turbo boost clarity is the main issue Intel needs to address.

However, Topweasel takes issue with their 5GHz claim and this is patently false. Set PL2 to 95W and you will see it will still operate within TDP spec and hit those speeds.

You confuse false advertising of speeds within TDP spec (=topweasels claim) with out of the box settings that violate that spec. Those are two different things.

 

[LTC8K6]

Power usage is specced at 90w. All that Intel marks as 130w, is what OEM Reference cooler they recommend. But that isn't clear. Even the 130 W AIO that intel recommends wouldn't have let Anandtech meet top clocks.

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.