Core i9-9900K Power & Thermals, Did Linus (and OC3D TV) Get it Wrong?

[asus killer]

13 minutes ago, Stefan Payne said:

Is irrelevant.

And they should be called out for cheating as their "specification" has no real world value or implication.

So basically they can do a 1.4GHz Base 4 Core/8 Thread CPU and call it 10W TDP, when it boosts to 4GHz and consumes 100W at that clockrate.

 

But the CPU is still specified at 10W TDP....

 

The worst thing:
Ever remember the AMD "ACP" thing?? Average CPU Power.

And the outrage over that?

Intel does exactly that right now. Only worse than AMD...

TDP is something subjective depending on how you measure it, you can see this by the discussion here. If they define their TDP values like that, i fell they aren't misleading no one. It's not even in small fine print, it's pretty obviously stated next to the value.

[porina]

17 minutes ago, Stefan Payne said:

People have measured of up to 260W CPU Power Consumption for the 9900K and see for yourself:

All bets are off when intentionally overclocking. I'm not an extreme overclocker yet, but I've seen my Ryzen 1800 report nearly 200W when I tried to get it past 4.0 GHz.

 

When Anandtech's original data was published, before they knew about the mobo excessive voltage (1.47v), that reached 220W. Dropping to 160-ish with a more sensible mobo.

[schwellmo92]

11 minutes ago, leadeater said:

I'm not, I'm pointing out what is actually correct that EVERYONE has been wrong about or fails to understand. I don't specifically blame anyone for not getting it, cTDP values have been messed with this whole time it seems so every review you have ever seen by anyone that has sustained CPU package power draw over the rated TDP has been overriding Intels TDP, which is allowed but is not stock configuration.

 

The TDP has very specific purpose, it's not to tell you what the power draw is. Certainly not at boost either. It's there to design a cooler and unless the cTDP has been increased after 16 seconds (on current CPUs) the CPU will not draw any more than 95W.

 

 

So explain to me exactly how the 95W TDP Intel puts on the 9900K is pointless because this graph shows it has a point.

 

Edit:

Btw you example is totally wrong for the point. Again turbo is NOS, your German car will overheat with infinite NOS being pushed in to the engine. And my cars? Read my location tag, I'm not American and drive a Japanese car anyway so...

The confusing thing for me is that the stock configuration wouldn’t allow for Intel claimed boost speeds, it’s supposed to do 4.7GHz all-core, and it won’t be able to do that with a 120w short budget.

[Carclis]

Just now, schwellmo92 said:

The confusing thing for me is that the stock configuration wouldn’t allow for Intel claimed boost speeds, it’s supposed to do 4.7GHz all-core, and it won’t be able to do that with a 120w short budget.

I think that's the point. The boost speeds were never supposed to be guaranteed. But seemingly we would typically hit them in the past. It looks like 8c/16t may have something to do with it.

[leadeater]

2 minutes ago, schwellmo92 said:

The confusing thing for me is that the stock configuration wouldn’t allow for Intel claimed boost speeds, it’s supposed to do 4.7GHz all-core, and it won’t be able to do that with a 120w short budget.

The 120W boost spec from what I have seen does allow it to hit 4.7Ghz all core, the power required as frequency increases is not linear so if it's a 10W increase from 4.5Ghz to 4.7Ghz it will be more than 10W to go from 4.7Ghz to 4.9Ghz etc.

[smount92]

5 hours ago, MMKing said:

 

But 5GHZ boost is within specifications... but then you're outside TDP specifications. Basically, the 5GHZ boost and the 95W TDP are at odds with one another. However, 5GHZ is NOT guaranteed while 95W is guaranteed at stock 3.6GHZ all core clock.

This is like Turbo 4-cylinder engines. You don't get 35mpg when you're gunning it at every light, and shifting at 5K RPM.

 

You can either get good performance, OR good efficiency. Not both at the same time.

[Stefan Payne]

26 minutes ago, leadeater said:

I'm not, I'm pointing out what is actually correct that EVERYONE has been wrong about or fails to understand.

When everyone is wrong or "fails to understand", than the thing its about, is wrong.

Easy as that.

 

Or would you agree that FX9590 has a TDP of 95W as well??

 

26 minutes ago, leadeater said:

I don't specifically blame anyone for not getting it,

You blame everyone except the perputrator of this thing wich is Intel.

They messed it up, not "everyone else".

Why don't you call out Intel for the shit they do?
Wouldn't you agree that it is kinda cheating?!

Wich is kinda the whole reason for that...

26 minutes ago, leadeater said:

cTDP values have been messed with this whole time it seems so every review you have ever seen by anyone that has sustained CPU package power draw over the rated TDP has been overriding Intels TDP, which is allowed but is not stock configuration.

When I put it together plug everything in and don't touch anything in the BIOS, it is stock configuration.

 

What Intel specifies or not is irrelevant, as we do not know what Intel tells the Mainboard Manufacturers the Default Settings have to be.

 

So pls stop blaming everyone else except Intel.

Oh and before you come back to older CPUs:

they don't Turbo 1GHz, they only do 400MHz at best. 

 

26 minutes ago, leadeater said:

The TDP has very specific purpose, it's not to tell you what the power draw is.

that is what YOU claim but the reality is that it is for system builder to build their Systems.

And it should be a reliable measure - wich it kinda was, until very recently.

Just look at Skylake! (6700K)

4GHz base, 4.2GHz Turbo. Yeah, great. And not really useful. on Kabylake it wasn't much different either for the K versions, the 7700 had a 600MHz Turbo.

Right now the 9900K has a max Turbo of almost 50% of the Core Clock!

(~40% to be more precise). Or 1.4GHz. 

Doesn't look much different than the 8700 (or K).

 

So that is something that really became an issue in the last generation or two. Not before.

And now people are realizing the useless TDP Definition from Intel.

Because they violate it so darn much...

26 minutes ago, leadeater said:

Certainly not at boost either. It's there to design a cooler and unless the cTDP has been increased after 16 seconds (on current CPUs) the CPU will not draw any more than 95W.

That is how you are defending Intel.

And intel tries to weasel its way out and claim a lower TDP than the average power consumption under normal circumstances.

 

Again, it wasn't much an issue before, when Ryzen wasn't available. Now there is Preassure on Intel and they make mistakes - like this one with the TDP!

 

26 minutes ago, leadeater said:

 

 

So explain to me exactly how the 95W TDP Intel puts on the 9900K is pointless because this graph shows it has a point.

Should I do something similar with cTDP of 35W with my A10-7850K to prove that it consumes only 35W??

That's pointless and not out of the box...

 

26 minutes ago, leadeater said:

Edit:

Btw your example is totally wrong for the point. Again turbo is NOS, your German car will overheat with infinite NOS being pushed in to the engine. And my cars? Read my location tag, I'm not American and drive a Japanese car anyway so...

No, its not.

Its normal usage and no Car comes with a NOS System out of the factory.

You can look it up, if you do not believe me but there is a difference in Cooling of the Engine within the same model for Europe and everywhere else because of the German, unlimited "Autobahn", that is the reason the cooling was modified.

 

And you are arguing that the Cooling of an American car is absolutely no issue on a German Autobahn, with an average speed of around 130km/h, wich is possible in certain regions and certain days, with long constant driving at 160km/h (or 180km/h even).

 

The power of the car is the same, lets say 150HP but the European version has bigger Coolers for the Water stuff and sometimes they add an additional oil cooler as well.

 

One example of this being done, that caused problems in colder enviroments is the Mazda RX-8 where the oil doesn't get warm in German gars in Winter.

 

 

So enough, I'm going to watch "that time I got reincarnated as a slime"...

 

[leadeater]

55 minutes ago, Stefan Payne said:

Should I do something similar with cTDP of 35W with my A10-7850K to prove that it consumes only 35W??

That's pointless and not out of the box...

 

95W is the out of the box setting, how many times does one have to tell you this. Gamers Nexus review shows this, that graph is stock. Real stock, not altered by the motherboard like in the past, it leaves the 95W TDP in place. Edit: To reiterate the Asus Maximus XI Hero default/stock configuration leaves the 95W TDP in place, other motherboards do not.

 

You want TDP to be power draw, that's not what the spec is for. You also want the TDP to be max power draw at full boost, even more not what it's for. As you can plainly see if I was designing a cooler that graph shows the continuous power draw to be 95W, excellent information if I was making a cooler for that product because I need to know the sustained value not the boost value.

[leadeater]

46 minutes ago, Stefan Payne said:

And now people are realizing the useless TDP Definition from Intel.

Because they violate it so darn much...

No they do not, it's the motherboard makers setting the cTDP value above the Intel spec.

 

Edit:

And the reason why I care is because I care about correctness of information so we are all more educated. What something actually means is important, what it's for is important, the conditions a test is done under is important.

 

I don't care about someones opinion on a spec very much, so spreading an opinion like it's factual is what I disagree with, a lot.

 

The fact is Intel CPUs have a cTDP and the default value is that which is stated on the Intel spec sheet. If it says 95W then the default package TDP is 95W and sustained power draw will be 95W. If you see higher than the spec then the cTDP has been changed and therefore the CPU is no longer operating in default configuration or in rated spec.

 

You're trying to blame Intel for other 3rd parties, not Intel, for changing the cTDP and then using that to try and show the TDP spec is pointless and does nothing when in fact it does.

 

The issue is, as far as I can see with the information at hand, that motherboards for a while now have been setting cTDP above the Intel spec by default. This means pointing to a review with a sustained power draw above the TDP does nothing to prove that the TDP spec is useless because the CPU is not running under Intel spec.

 

I'm not saying, nor have ever said, not to add a max all core turbo power draw to the spec sheet. TDP however has a set purpose and we should not dilute or change it's meaning, it's a thermal specification not a power specification. If a system designer sets the cTDP above 95W then the cooling solution needs to match it and that has nothing to do with Intel.

 

46 minutes ago, Stefan Payne said:

What Intel specifies or not is irrelevant, as we do not know what Intel tells the Mainboard Manufacturers the Default Settings have to be.

That's a bit hypocritical is it not? You say Intel's spec is irrelevant then also harp on about the spec, if it's irrelevant then why do you care? You bring up this rhetoric about Intel violating the spec but ignore the motherboard makers are the ones actually doing it. How about to criticize the ones actually causing the issue.

 

46 minutes ago, Stefan Payne said:

When everyone is wrong or "fails to understand", than the thing its about, is wrong.

If something is wrong, being collectively wrong about it doesn't make it right, it's still wrong. Everyone thought the Earth was flat until is was proven otherwise, everyone was wrong about that until they were educated otherwise, or are you also part of 'that movement'.

[pas008]

1 minute ago, leadeater said:

95W is the out of the box setting, how many times does one have to tell you this. Gamers Nexus reviews shows this, that graph is stock. Real stock, not altered by the motherboard like in the past, it leaves the 95W TDP in place.

dont think you will get through to him

the hatred flows through him

 

its like how people not understanding mpg on vehicles works

[XenosTech]

1 hour ago, leadeater said:

I'm not American and drive a Japanese car anyway so...

Better be a supra or a r32 >.>

[leadeater]

2 minutes ago, XenosTech said:

Better be a supra or a r32 >.>

370GT

[Curufinwe_wins]

7 hours ago, schwellmo92 said:

 

Ssnip

 

Yet another overblown and wrong video by hardware unboxed. Getting so annoyed with him these days.

 

It isnt a vrm issue. It's a setting issue. The boards he ran is running the processor OUT OF SPEC. Technically speaking his results are wrong, not ltt. Anthony went through all of the stuff he did in great detail to check and double check and confirm the data he measured.

 

Likewise, Anandtech had to redo their results because it turned out the board they used was massively overvolting on load default.

[XenosTech]

4 minutes ago, leadeater said:

370GT

I'll allow you this one sir lol

 

On topic: Seems people still have a hard time differentiating TDP from power draw even though people have been explaining this for months or years probably.

[Stefan Payne]

44 minutes ago, leadeater said:

No they do not, it's the motherboard makers setting the cTDP value above the Intel spec.

And why do they do that?!
Do they do that because Intel asked them to perhaps??

 

44 minutes ago, leadeater said:

And the reason why I care is because I care about correctness of information so we are all more educated. What something actually means is important, what it's for is important, the conditions a test is done under is important.

I agree but that doesn't change the fact that the Intel TDP, the way it is done right now, with the 8 and 9k Generation is useless and doesn't help anyone in any way.

 

Or what would I have lost when Intel would not have mentioned the TDP at all?

But why not also mention a second TDP, like Anandtech (and I) sugested??

 

44 minutes ago, leadeater said:

I don't care about someones opinion on a spec very much, so spreading an opinion like it's factual is what I disagree with, a lot.

Now lets use this Case as an Example:

https://antec.com/product/case/isk110-vesa-u3.php

 

What Intel CPU can I use with that?

And what AMD CPU (or rather APU) is possible?

 

You see the Problem...

 

For you the TDP is not important because you do work with Desktops (and Servers), for me its important to be accurate because I like working with Small Form Factor systems. And for that I need it to be the upper end and not some useless specified thing.

 

44 minutes ago, leadeater said:

The fact is Intel CPUs have a cTDP and the default value is that which is stated on the Intel spec sheet. If it says 95W then the default package TDP is 95W and sustained power draw will be 95W. If you see higher than the spec then the cTDP has been changed and therefore the CPU is no longer operating in default configuration or in rated spec.

That's your interpretation, not how the reality works, as the TDP is violated and Intel shouldn't have specified the 9900K as a 95W TDP CPU. But with AMD having most of their 8cores even at 65W TDP with the exception of the X versions wich are specified to be 95W and the 2700X with 105W, there was some pressure for Intel to specify it lower rather than higher or more honestly.

 

And that is what that is about, that the Intel TDP means that the CPU is heavily throtteling - wich is also bullshit - or that the CPU consumes far more than specified. 

Both are situations that aren't really acceptable and you have to accuse Intel of cheating with the TDP because of that.

 

Or do you disagree that a 125W TDP would have been more honest and better for all people involved??

 

44 minutes ago, leadeater said:

You're trying to blame Intel for other 3rd parties, not Intel, for changing the cTDP and then using that to try and show the TDP spec is pointless and does nothing when in fact it does.

I don't blame, I question. 

And you we do not know, what Intel told the motherboard manufacturers how to handle this and they don't seem to care that much as the option of the hard 95W TDP Limit doesn't seem to be on at default.

 

Hell, even with the 8k Series the "MCE" was on at default on some Boards.

Thing is that we do not have information about the communication between Motherboard Manufacturers and Intel.

 

But we do know that Intel would have had the Power to ask the motherboard manufacturers nicely to value the specification and disable the cTDP only at the press of a button or something like that.

 

But that didn't happen, as the Motherboard Manufacturer did not do that. So the most probable solution is that Intel did not ask the Motherboard manufacturers, didn't care or even wanted them to do it this way and did ask them to let the CPU run freely.

 

Or another point that is possible. The motherboard manufacturer didn't really know because there are some BIOS Parts that come from the CPU Manufacturer that are only implemented...

44 minutes ago, leadeater said:

I'm not saying, nor have ever said, not to add a max all core turbo power draw to the spec sheet. TDP however has a set purpose and we should not dilute or change it's meaning, it's a thermal specification not a power specification. If a system designer sets the cTDP above 95W then the cooling solution needs to match it and that has nothing to do with Intel.

WE aren't the ones who dilute or change the Meaning of the TDP, Intel is doing that. And they should be criticised for that, wich didn't happen when they introduced the changed TDP Spec because it wasn't really an issue because the CPUs didn't boost ~50% of the clockspeed like they do now.

44 minutes ago, leadeater said:

That's a bit hypocritical is it not? You say Intel's spec is irrelevant then also harp on about the spec, if it's irrelevant then why do you care? You bring up this rhetoric about Intel violating the spec but ignore the motherboard makers are the ones actually doing it. How about to criticize the ones actually causing the issue.

Sorry, I misspoke and failed to articulate my thoughts correctly.

 

So what I mean is that the important part is what Intel tells the Motherboard manufacturers to do, not the specification. 

So if Intel tells/asks the Motherboard manufacturers to value the specification or not.

 

On the other side whe have AMD with the Threadripper CPUs that are limited by the TDP out of the box, as you can see in various reviews, that the CPU Package consumption is no more than 180W; regardless of what you do (with the 180W TDP Threadrippers of course). 

You can switch this limit off and let the CPU run freely. 

That is how it is on AMD Hardware.

 

It is not how it is on Intel Hardware. There the CPU runs freely, thus the spec is not relevant, what Intel told the Motherboard manufacturers to do is relevant. As they seem to all ignore the spec and let the CPU run without limits (or set high TDP Limits that are way over the spec).

 

44 minutes ago, leadeater said:

If something is wrong, being collectively wrong about it doesn't make it right, it's still wrong.

Everyone thought the Earth was flat until is was proven otherwise, everyone was wrong about that until they were educated otherwise, or are you also part of 'that movement'.

True but we are talking about a technical specification, done by the manufacturer.

 

If everyone sees the specification or whatever the manufacturer claims different than the manufacturer, than the manufacturer is wrong. And has to rework their specification in a way so that everyone understands it.

Simple as that.

 

 

 

Your example with the Earth is not really fitting in this case, but I can also use that analogy in this case as well.

Because the Flat Earth thing was propagated by large institutions like "the Church", while the "freethinking people" did some tests/calculations that proved the earth to be wrong.

 

We have something similar right now, on the one side a big institution claiming that a thing is one way. And on the other hand "the people" claiming that it has to be another way. And also calling out the Institution for claiming what they think is not accurate or even wrong.

 

The Big Institution in this case is Intel and "the People" use the TDP the way it was introduced and accepted for decades.

Now Intel changed their mind about the TDP they themselves introduced and specified as the "maximum amount of heat generated by under any workload."

And that is how it was used for more than 20 years.

 

And now the Institution (and you) expect people to change the way they "used to know the TDP"?? ÖHm, no. Just no.

 

The TDP was introduced back in the mid 90s, we are talking P54C or P55C here. And redefined either at the end of the 2000s or beginning of the 2010s - 15-20 Years later.

That that might come back to bite Intel, well d'uh, obviously. Because people are used to work with TDP in this one way for more than a decade...

And now its completely different? Sorry, I blame Intel for not listening to Intel and redefining the Intel TDP Specification, wich is hardly more than the ACP they made fun of back in 2011...

 

 

[Stefan Payne]

24 minutes ago, XenosTech said:

Better be a supra or a r32 >.>

Now that you write that, I want my MA70 back...

In beautiful white and automatic transmission...

 

Was a nice Car, though a bit on the loud side...

[XenosTech]

1 minute ago, Stefan Payne said:

Now that you write that, I want my MA70 back...

In beautiful white and automatic transmission...

 

Was a nice Car, though a bit on the loud side...

Ewww automatics

[The Benjamins]

28 minutes ago, XenosTech said:

Ewww automatics

CVT

[leadeater]

2 hours ago, Stefan Payne said:

And why do they do that?!
Do they do that because Intel asked them to perhaps??

Why would Intel ask that, why would they care? Intel sets the TDP, that's what they would ask for and recommend, that's why the spec exists. What else do you think they bother giving a spec for.

 

Intel asks us to over clock our CPUs? I mean it's possible to do so which means Intel is asking us to do so? Now the frequency specs are useless?

 

Really you're going to blame Intel for others running their CPU out of spec just because they allow it? They could change it to a hard fixed limit but I bet you wouldn't actually like that.

 

2 hours ago, Stefan Payne said:

Now lets use this Case as an Example:

https://antec.com/product/case/isk110-vesa-u3.php

 

What Intel CPU can I use with that?

And what AMD CPU (or rather APU) is possible?

 

You see the Problem...

 

For you the TDP is not important because you do work with Desktops (and Servers), for me its important to be accurate because I like working with Small Form Factor systems. And for that I need it to be the upper end and not some useless specified thing.

 

Omg no that is not the point. Intel sets the default TDP for the CPU, you use that for designing the cooling solution. Of course you aren't going to put a 95W CPU in a small PC like that with a 90W configuration, you could but that would make YOU at fault, not Intel, end of story. Period, nothing else.

 

You could use the cTDP to lower the 9900K to 90W TDP if you wanted to though, but you're ignoring that aren't you.

 

Instead of writing a text wall how about you just admit you were wrong about this Intel TDP thing. With the TDP in place the CPU will not long term draw more than that so that is the minimum cooling required, that is what the spec is for.

 

Any setting outside of spec Intel has no blame for, other than allowing it which seems to be your only reasoning for blaming Intel.

 

Edit:

And again unmodified the 9900K will use 95W sustained so the TDP spec is correct and you can use that, one of the most trusted tech reviewers has given you direct proof of that. You're just denying reality if you disagree. There's not much else to call it.

[XenosTech]

1 hour ago, The Benjamins said:

CVT

Even worse *runs into the hills screaming*

[The Benjamins]

26 minutes ago, XenosTech said:

Even worse *runs into the hills screaming*

Honestly I don't like the disconnect of car people vs engineers on their stance on the CVT.

For performance and efficiency the CVT is the best type of transmission.

 

But Manual is the most fun.

[Dylanc1500]

I'm just going to buy florinert and build a Cray like cooling system. Then I'll make a case with AMD, IBM, Intel, and VIA systems in it. I won't ever have to worry about any TDP.

[mr moose]

4 hours ago, Stefan Payne said:

When everyone is wrong or "fails to understand", than the thing its about, is wrong.

Easy as that.

 

 

When are you going to realise that you are the only one who is wrong on this.  It has been explained to you at least 30-40 times on this forum by at least 5 different people now.  Your desire for TDP to be something it is not and your failure to understanding why that is important is what is the problem here, not Intel.

 

Seriously every single Intel thread gets derailed by you lying about TDP nowadays.  

[BuckGup]

According to der8uar thermals were horrible due to thicker silicon used on the dye and lapping with liquid metal dropped temps by 22C. He was used a 360 AIO too and was hitting 90C+ at 1.22v 5ghz

[Curufinwe_wins]

1 hour ago, The Benjamins said:

Honestly I don't like the disconnect of car people vs engineers on their stance on the CVT.

For performance and efficiency the CVT is the best type of transmission.

 

But Manual is the most fun.

Spoiler

 

Car people are fng idiots for 100% directly causing the lack of investment and slow mass adoption of CVT. Properly designed they last longer, perform better, and have better efficiency. Ugh that makes me so pissed off as an engineer (with significant power systems experience).

 

Guess what? People thought CVTs were so boring and non-responsive that multiple manufacturers gave their CVT launches discrete steps to mimic the stupidity that is hardwired into people that lurching gear shifts are somehow more fun and better. Completely avoiding the point.

 

I swear, I've never in my life heard someone complain that a snowmobile was unfun to drive, and basically every snowmobile uses a cvt. CVT direct control and much more rapid feedback responses are way more engaging than slowass manuals (dual clutch manuals are not manuals. They never have been and never will be).

 

Sorry. Angry engineer rant.