Intel's Comet Lake Absence at CES Reportedly Related to Power Consumption Wall

[Raevenlord]

Reports are flooding the web regarding Intel's total lack of reference to their upcoming Comet Lake family of CPUs, which will be branded under the Intel Core 10000 series. As reports would have it, motherboard makers had stock of LGA 1200 motherboards ready to showcase at CES, but were told to pull them in what is equivalent to a logistical "last minute". It seems that both Intel's lack of commitment to Comet Lake on its CES presentation and absence of ecosystem showcase at this year's CES might have something to do with, well, close to shame on Intel's parts.

Comet Lake will increase the maximum core count for their desktop CPUs up to 10 cores and 20 logical threads. But being built on the same 14 nm process as previous Intel generations since Skylake, there isn't much that can be done to offset increased power consumption. This is why industry sources are claiming Intel decided to skip Comet Lake at this CES - a difficulty to rein in the processors' power consumption in time for the event, with power consumption hitting 300 W. And with Intel's Core i9 10900K being configured with a PL2 (Power Level 2) of 250 W, a maximum 300 W under full load seems more than plausible.



View at TechPowerUp Main Site

 

[john_]

I guess Intel have hit the ceiling of what it's 14nm++ can offer.
I only pity those consumers, buying OEM systems with the greatest and latest 10 core CPU from Intel, that will be throttling all the time.

 

[birdie]

total lack of reference

- wut?

As reports would have ti

- wut?

Have you read what you wrote at least once? ;-)

with power consumption hitting 300 W

That's not power consumption, that's PL2 - peak power consumption for a very short period of time. Power consumption is rumored to be 125W and unlike AMD Intel sticks to what it advertises (unless a motherboard vendor has set the limits differently). I know AMD fans have infiltrated all social media and forums but AMD'ism shouldn't infiltrate your reasoning and sanity.

And while we're at it let me remind everyone that AMD rated the Ryzen 7 3700X CPU 65W while in reality its sustained power draw is 91W. No one bats an eye about this obvious marketing lie.

Edit: made bold the statement about motherboard vendors setting their own limits - people who've replied to this message clearly cannot read.

 

[KarymidoN]

- wut?

- wut?

Have you read what you wrote at least once? ;-)



That's not power consumption, that's PL2 - peak power consumption for a very short period of time. Power consumption is rumored to be 125W and unlike AMD Intel sticks to what it advertises (unless a motherboard vendor has set the limits differently). I know AMD fans have infiltrated all social media and forums but AMD'ism shouldn't infiltrate your reasoning and sanity.

And while we're at it let me remind everyone that AMD rated the Ryzen 7 3700X CPU 65W while in reality its sustained power draw is 91W. No one bats an eye about this obvious marketing lie.

The year is 2020 and people still don't understand that the 65W TDP is for THERMALS and not power draw... so sad...

 

[birdie]

Straight from the horse's mouth:

Though both are often measured in watts, it is important to distinguish between thermal and electrical watts. Thermal wattage for processors is conveyed via thermal design power (TDP). TDP is a calculated value that conveys an appropriate thermal solution to achieve the intended operation of a processor. Electrical watts are not a variable in the TDP calculation. By design, electrical watts can vary from workload to workload and may exceed thermal watts. GD-109

I guess you're right but weirdly some of AMD CPUs follow their own TDP figures while others don't which is extremely confusing. Also, if I remember correctly they equalled electrical watts to TDP in the past and then they stopped doing that.

 

[EatingDirt]

That's not power consumption, that's PL2 - peak power consumption for a very short period of time. Power consumption is rumored to be 125W and unlike AMD Intel sticks to what it advertises (unless a motherboard vendor has set the limits differently). I know AMD fans have infiltrated all social media and forums but AMD'ism shouldn't infiltrate your reasoning and sanity.

And while we're at it let me remind everyone that AMD rated the Ryzen 7 3700X CPU 65W while in reality its sustained power draw is 91W. No one bats an eye about this obvious marketing lie.

9700k, 8700k and 9900k on the graph you posted would like to have a word with you.

 

[ppn]

200mm2 5.Ghz is pushing to the limits and is hitting 250 watts with AVX2 only. For the average user that is never needed so we are looking at down to earth 125 watts. Not bad at all. 10nm 8 core tiger lake around 200mm2 will be the same power at the same clocks. Then the 50% smaller die 100mm2 on 7nm which is the equivalent of 5nm TSMC.

 

[londiste]

For real usage - power consumption = amount of heat.
Both manufacturers have their own definitions and variables but despite some good technical reasoning for how they work with TDP, it is clearly not easily understandable for a normal user.

Anandtech has pretty good articles on what TDP is for both AMD and Intel:
Intel: https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21
AMD: https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/19

Edit:

9700k, 8700k and 9900k on the graph you posted would like to have a word with you.

IT would be good to know the methodology of how that is measured. Intel CPUs TDP spec with default values says 125% TDP for 8 seconds. This holds true for non-K CPUs. All bets seem to be off for K CPUs, motherdoards are ignoring spec and impose limits that are far higher than spec or do no impose limits at all. Intel at the same time is the one tolerating these shenanigans.

 

[ZoneDymo]

big oofs, as the kids would say

 

[Berfs1]

- wut?

- wut?

Have you read what you wrote at least once? ;-)



That's not power consumption, that's PL2 - peak power consumption for a very short period of time. Power consumption is rumored to be 125W and unlike AMD Intel sticks to what it advertises (unless a motherboard vendor has set the limits differently). I know AMD fans have infiltrated all social media and forums but AMD'ism shouldn't infiltrate your reasoning and sanity.

And while we're at it let me remind everyone that AMD rated the Ryzen 7 3700X CPU 65W while in reality its sustained power draw is 91W. No one bats an eye about this obvious marketing lie.

Edit: made bold the statement about motherboard vendors setting their own limits - people who've replied to this message clearly cannot read.

It is like people are so f*cking blind, cannot even understand that some things do NOT make sense AT ALL. Want an example? Oh ok, SO PLEASE EXPLAIN WHY THE i9-7920X IS CAPPED AT ABOUT 140W?? MAYBE ITS BECAUSE INTEL HAS A *DIFFERENT* DEFINITION OF TDP FOR X299 WOWOWOWOW ITS LIKE INTEL CHANGED THEIR STORY A LOT OF TIMES ok whatever, but please, i may not have a college degree, but PLEASE explain to me how a 9700K takes less energy than an 8700K. 9700K has more cores.

200mm2 5.Ghz is pushing to the limits and is hitting 250 watts with AVX2 only. For the average user that is never needed so we are looking at down to earth 125 watts. Not bad at all. 10nm 8 core tiger lake around 200mm2 will be the same power at the same clocks. Then the 50% smaller die 100mm2 on 7nm which is the equivalent of 5nm TSMC.

"for the average user" Ok the average user is totally going to buy a 8 core 16 thread processor because they totally need 8 cores and 16 threads (soon to be 10 cores 20 threads) because the average user wants to spend 500$ *on a CPU alone*. The average user goes for i3s and i5s for Intel because they are much cheaper, and because they have less cores, ~60W is a good estimate. Now, the people that buy higher performance parts, usually need much more performance, hence why they bought a CPU with more cores in the first place, therefore the "125W realistic power draw" is nonsense. Who in their right mind pays *five hundred* dollars on an *unlocked* processor that they *won't* overclock, but will buy a high end cooler for it (since Intel doesn't provide coolers with these unlocked parts)? It should be simple, but i guess everyone nowadays needs a 9900K with five hundred gigahurtz and all the speed in the world.

 

[JackCarver]

I got nearly 200W with my 8700K@4700 all core in OCCT AVX2 small dataset benchmark. With a good CPU cooler no problem but VRM got too hot, so it throttled after 10 mins.

 

[ScaLibBDP]

A message to Editor....

>>...desktop CPUs up to 10 cores and 20 logical threads ...

Absolutely incorrect "translation" and it should be corrected to:

...desktop CPUs up to 10 cores and 20 logical processors...
or
...desktop CPUs up to 10 cores and 20 hardware threads ...

 

[Deleted member 158293]

That is burning a lot of power for little gain.

Definitely outside the efficiency curve

 

[Steevo]

A message to Editor....

>>...desktop CPUs up to 10 cores and 20 logical threads ...

Absolutely incorrect "translation" and it should be corrected to:

...desktop CPUs up to 10 cores and 20 logical processors...
or
...desktop CPUs up to 10 cores and 20 hardware threads ...

Nah, cause HT can be turned off, it will have 10 physical cores, each capable of handling 2 threads so 20 logical threads can be active at a time. But still only 10 can execute simultaneously.

 

[Juankato1987]

The year is 2020 and people still don't understand that the 65W TDP is for THERMALS and not power draw... so sad...

A Watt is a Watt, TDP is a market number, to try to compare between INTEL and AMD processors, a number that both calculate from diferent ways.
So is not a valid number, for me at least, for comparision, when actual power ussage is more high than suggested TDP.

There is not a Thermal Watts, or Electrical Watts, just Watts.
Is like say that there are Solid Liters, and Liquid Liters, or a pound of feathers and a pound of rocks etc....

 

[newtekie1]

A Watt is a Watt

Yep, and since there is no moving parts in a processor, those thermal Watts are directly coming from electrical Watts. The laws of physics tells us that a CPU that is outputting 150w in heat is consuming 150w of electricity.

 

[Deleted member 158293]

Here's one of the better summaries of what TDP means, according to Intel.

Along with other dubious stuff from motherboard vendors did a while ago. Lots of better sources for this, but this video is easier to understand.

The formula AMD uses to calculate TDP is different.

Both are correct, in their own way...

 

[Valantar]

Nah, cause HT can be turned off, it will have 10 physical cores, each capable of handling 2 threads so 20 logical threads can be active at a time. But still only 10 can execute simultaneously.

That depends on which part of the execution ports of the core each thread utilizes - the main point of SMT is to run threads simultaneously to utilize resources that would otherwise be idle if the core was running a single thread. That's why you get performance increases from SMT at all - otherwise you'd just be queuing threads at 100% performance. The reason SMT fails to scale to more than 125-150% performance is (put simplistically, but pretty accurate) that there's a limit to how realistic it is to run two simultaneous threads that will never require the same resources.

Both are correct, in their own way...

Indeed, seeing how TDP as a term is defined entirely by the manufacturer it would be rather absurd to claim that the manufacturer's definition of it is wrong. One can argue that the definition is poorly thought out or misleading, but it can never be wrong.

 

[Juankato1987]

Yep, and since there is no moving parts in a processor, those thermal Watts are directly coming from electrical Watts. The laws of physics tells us that a CPU that is outputting 150w in heat is consuming 150w of electricity.

Yes, my point is that TDP is an arbitrary number put there just to try to foolish people about power consumption.
And then (both AMD and INTEL ) try to say that it's about thermal measure, why they don't just put actual power ussage in the box.

 

[Steevo]

That depends on which part of the execution ports of the core each thread utilizes - the main point of SMT is to run threads simultaneously to utilize resources that would otherwise be idle if the core was running a single thread. That's why you get performance increases from SMT at all - otherwise you'd just be queuing threads at 100% performance. The reason SMT fails to scale to more than 125-150% performance is (put simplistically, but pretty accurate) that there's a limit to how realistic it is to run two simultaneous threads that will never require the same resources.


Indeed, seeing how TDP as a term is defined entirely by the manufacturer it would be rather absurd to claim that the manufacturer's definition of it is wrong. One can argue that the definition is poorly thought out or misleading, but it can never be wrong.

It doesn't execute the threads at the same time, it holds one if it it is waiting for data, so it seems simultaneous and that is the reason its called SMT. But yeah, the performance increase is due to the option of processing another thread on that core while the other is in a wait state.

Physical > Logical cores.

 

[poopybangusore]

- wut?

- wut?

Have you read what you wrote at least once? ;-)



That's not power consumption, that's PL2 - peak power consumption for a very short period of time. Power consumption is rumored to be 125W and unlike AMD Intel sticks to what it advertises (unless a motherboard vendor has set the limits differently). I know AMD fans have infiltrated all social media and forums but AMD'ism shouldn't infiltrate your reasoning and sanity.

And while we're at it let me remind everyone that AMD rated the Ryzen 7 3700X CPU 65W while in reality its sustained power draw is 91W. No one bats an eye about this obvious marketing lie.

Edit: made bold the statement about motherboard vendors setting their own limits - people who've replied to this message clearly cannot read.

BRUH.... "Intel sticks to what it advertises" *cries about 65w rated processor sustaining 91w (both laughably low)* *Posts graph showing "95w" 8700k pulling 150w, "95w" 9700k pulling 121w and "95w" 9900k pulling 168w* BAHAHAHAHAH (posting this from my 9900kf system) I just built a system with an r5 2600 and its lit, amd is definitely killing it right now, you're clearly a butthurt fanboy just as bad as amd fanboys. Stop fighting for a company in comments sections, they don't even know you exist. Just buy whatever is best for your use case at the time.

 

[Metroid]

any difference between intel's comet lake and a mini toaster? i see none

 

[Digital Dreams]

any difference between intel's comet lake and a mini toaster? i see none

The mini toaster isn't full of security holes?

 

[timta2]

Reading that first paragraph makes me feel like a non-native English speaker. For example, when someone says that something was pulled out, in my experience, usually that means that it was pulled out to be displayed. If something is "pulled" (without "out"), it's usually hidden from view and not displayed. There are also a number of other issues, but I will leave it at that, in fear of being too "mean".

 

[londiste]

It doesn't execute the threads at the same time, it holds one if it it is waiting for data, so it seems simultaneous and that is the reason its called SMT. But yeah, the performance increase is due to the option of processing another thread on that core while the other is in a wait state.

This is the classical definition of SMT. With superscalar processors different threads do execute instructions simultaneously if there are free resources to do so. Of course, this does not always happen and there are possible bottlenecks all over the place but simultaneous execution does happen.