Question Intel vs. AMD CPUs - out of the box POWER DRAW

Kocicak

Senior member
Jan 17, 2019
982
973
136
I was surprised, after I read in the recent review, that if you place Intel CPU in the out of the box motherboard, it will draw over 200W instead of advertised 65W.

I have a fresh experience with three different CPUs from the new AMD 5000 series, out of the box they run at their advertised power draw of 105W. I placed them all in one motherboard so I do not know if this is not specific just to my motherboard.

So is it now a general fact across more motherboard manufacturers, that if you put Intel CPU in OUT OF THE BOX (default BIOS settings not changed by user) motherboard, it will draw much more than advertised power, and if you put AMD processor in out of the box MB, it will run at specified power draw?
 
Last edited:

Gideon

Golden Member
Nov 27, 2007
1,598
3,520
136
I was surprised, after I read in the recent review, that if you place Intel CPU in the out of the box motherboard, it will draw over 200W instead of advertised 65W.

I have a fresh experience with three different CPUs from the new AMD 5000 series, out of the box they run at their advertised power draw of 105W. I placed them all in one motherboard so I do not know if this is not specific just to my motherboard.

So is it now a gereral fact across more motherboard manufacturers, that if you put Intel CPU in OUT OF THE BOX (default BIOS settings not changed by user) motherboard, it will draw much more than advertised power, and if you put AMD processor in out of the box MB, it will run at specified power draw?

Just talked about the subject in this thread.

Overall, yes you are correct.
AMD's power draw @ stock is always capped at TDP * 1.35 which is called "PPT" and this is configurable in the BIOS if needed:

That means that 65W chips max-out at ~88W and 105W chips at ~141W
 

uzzi38

Platinum Member
Oct 16, 2019
2,556
5,531
146

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
Neither are running at rated temp. Base speeds are based on all core power draw at rated power envelope. Both act more like video cards than previous CPU's.

I do think there is some significant issue with how far Intel's power rating is off on their top chips from max. But it's not even the more you have to cool a 10900k that's the issue. But that it spikes up so high that you may not realize you need a better power supply. Specially if partnered with a 3k or 6k video card. Not a fan of these guy having their spikes either but a 30% increase is less bad than a 250% increase in power usage.
 

Kocicak

Senior member
Jan 17, 2019
982
973
136
I remember seeing something like 107W somewhere when running 5900 or 5950 and thinking, that it is almost perfectly the spec.

I am not sure if it was total CPU power draw or just power draw of cores.
 

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
I remember seeing something like 107W somewhere when running 5900 or 5950 and thinking, that it is almost perfectly the spec.

I am not sure if it was total CPU power draw or just power draw of cores.
Its the Cooling envelope. Get extra performance by going higher but it will turbo on single core for some length and so on but will fall back to the 107 when needed.

It will extend even when only having 110w of cooling to ~150w. The question is for how long. Funny thing was on the 3950x the combination on clocks and core usuage maxes out power usage at about 8-10 cores at full load.
 

Kocicak

Senior member
Jan 17, 2019
982
973
136
I had low temps in all core heavy load, so the 107W I saw was not temperature constrained.

At 150W power draw of 5900 when I enabled PBO I started hitting the cooling ceiling of my air cooler, but it stayed more or less the same.

200W of the 5950 with enabled PBO would drop, because the air cooler could not handle that.

Now I am not sure anymore if that 107W was really OOTB, or with altered setting - PBO DISabled. I remember having it disabled for some time.
 

Mopetar

Diamond Member
Jan 31, 2011
7,794
5,897
136
The flip side of this is that you can buy a non-K Intel part and get the same MC performance for around $50 less since that supposed low power part can hit the same TDP as the unlocked part and even though the single core boost isn't as high, the all core boost is either identical or essentially the same.

The AT results with the 10700 showed that anything that's highly parallel or can use up all available threads is a basically a wash in terms of performance. The power draw was slightly worse, but it's to be expected that the K parts are a better bin. Whether that's worth the cost savings is up to you
 
  • Like
Reactions: Tlh97

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
I had low temps in all core heavy load, so the 107W I saw was not temperature constrained.

At 150W power draw of 5900 when I enabled PBO I started hitting the cooling ceiling of my air cooler, but it stayed more or less the same.

200W of the 5950 with enabled PBO would drop, because the air cooler could not handle that.

Now I am not sure anymore if that 107W was really OOTB, or with altered setting - PBO DISabled. I remember having it disabled for some time.
In theory 107w is the thermal requirements. Even with a paltry cooler, under most conditions if CPU usage spiked it will hit a similar limit (that PBO releases, so the 5900 and 5950 draws probably exceed them what you would normally see), and generally step down from those peaks based on several thermal limitations. Also because it takes less power to run from a colder temp to clock to a particular level you will see even with the cooling a drop rated speed as the bios steps up power to maintain performance (with good cooling it means it generally lowers clocks so at the new power level it will match temp limits).

In the 1k and 2k it was a little more traditional with just an extra layer of perf rated if you have the cooling before it.

But the important part is that it isn't just thermally limited but powered limited to within a certain point to keep it manageable. I lost the war on right thermal reporting and I still think Intel is more extreme in its breaking that but its a war that was lost in maximizing potential performance out of the gate. But the unlocked power allowance that started with the 8700k is bad and taken to an extreme on the 10900k that can spike upwards of 300w at times. You get a 600w PSU for a 125w CPU and 300w 3080 or 6800XT. You know having extra power throughput around 20% (and nearing limitations on SFX power supplies). Heck even 700w considering efficiency. So 1 CPU and 1 GPU. Could crash a 700w CPU. Even though their power rating is rated at 425w.
 

dullard

Elite Member
May 21, 2001
24,964
3,299
126
I was surprised, after I read in the recent review, that if you place Intel CPU in the out of the box motherboard, it will draw over 200W instead of advertised 65W.
The key part is the bolded word. What did Intel advertise? We need to be clear about what exactly is advertised. Intel's definition is in articles below (color emphasis mine):
Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload.


What is the maximum power consumption for my processor?
Under a steady workload at published frequency, it is TDP. However, during turbo or certain workload types such as Intel® Advanced Vector Extensions (Intel® AVX) it can exceed the maximum TDP but only for a limited time, or
  • Until the processor hits a thermal throttle temperature, or
  • Until the processor hits a power delivery limit.


TDP is not the maximum power that the processor can dissipate



We as a community too often make the mental shortcut of TDP = maximum power possible. Turbo modes make that shortcut not accurate. Intel exceeds TDP power by quite a lot, for a limited amount of time. AMD chips exceed TDP too, but to a far lesser extent (see Gideon's post above).

We as a community need to come to terms with this. TPD is the minimum thermal design power to design your cooling system for. TDP is not maximum power used.
 
Last edited:
  • Like
Reactions: Zucker2k

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
I was surprised, after I read in the recent review, that if you place Intel CPU in the out of the box motherboard, it will draw over 200W instead of advertised 65W.
Only for the few seconds that the turbo boost lasts, for the 10900 for example it will run at 224W for 28 seconds, so if you do something that takes 10 minutes you will have 224W for 28 sec and 65W for 9:32.
The longer the thing you do lasts the lesser the deviation from the 65W.
 

jpiniero

Lifer
Oct 1, 2010
14,487
5,154
136
Only for the few seconds that the turbo boost lasts, for the 10900 for example it will run at 224W for 28 seconds, so if you do something that takes 10 minutes you will have 224W for 28 sec and 65W for 9:32.
The longer the thing you do lasts the lesser the deviation from the 65W.

The mobo is completely in charge of how long the turbo lasts, as noted in the earlier posts. Any recommendations from Intel are just that.
 

dullard

Elite Member
May 21, 2001
24,964
3,299
126
Only for the few seconds that the turbo boost lasts, for the 10900 for example it will run at 224W for 28 seconds, so if you do something that takes 10 minutes you will have 224W for 28 sec and 65W for 9:32.
The longer the thing you do lasts the lesser the deviation from the 65W.
Exactly. Yes, the Intel chips will draw more than TDP (as described by Intel). But, this is only for a short time period. It is more work for reviewers, but they need to do the work to review both short term bursts AND long-term steady use cases.

In the short-term burst situation, imagine a user doing little and then suddenly clicking a "calculate" button, or entering a formula into Excel, or starting a Word spell check, or loading a program. The Intel processor will use low idle power then jump up to over 200W for a short burst then back to low idle power.

In the long-term steady use case, imagine a long-running digital currency mining program, a long running video edit, a long-running software simulation. In these cases, the 200W is for a few seconds, then it sits at 65W for almost the entire rest of the long-running work.

These two completely different situations are rarely covered in reviews. Instead, a test script might be performed with benchmarks run back-to-back-to-back and totally miss the bursty workload that many people actually experience.

I actually wish reviewers go even further. Show us the performance when the box is in a dusty closed desk like many users have. Or in a hot room. Or in a room that is pretty cold. You'll be surprised at how much that impacts turbo power usage and few reviews even get into this significant performance impact.
 
  • Like
Reactions: MangoX

Gideon

Golden Member
Nov 27, 2007
1,598
3,520
136
  • Like
Reactions: Tlh97

Kocicak

Senior member
Jan 17, 2019
982
973
136
The key part is the bolded word. What did Intel advertise? We need to be clear about what exactly is advertised. Intel's definition is in articles below...

Well, that is the trouble. When you have a single spec number characterising power draw being published, and being reused in the description of the retailers selling this product, I call it being advertised. You expect such number to reflect reality somehow. You do not expect to be required to read any technical articles to understand what this number means.
 
  • Like
Reactions: Tlh97 and scannall

dullard

Elite Member
May 21, 2001
24,964
3,299
126
Not if the power supply, case, cooler, and motherboard is configured to maintain the boost indefinitely.
Corrected that for you. Ian's review gave me quite a laugh. First thing he did was pull off the 65W cooler that would keep the average power to 65W in the long term. Then he complained that with better cooling, the chip would turbo longer, run faster, and use more than 65 W.
 
  • Like
Reactions: Zucker2k and MangoX

dullard

Elite Member
May 21, 2001
24,964
3,299
126
Well, that is the trouble. When you have a single spec number characterising power draw being published, and being reused in the description of the retailers selling this product, I call it being advertised. You expect such number to reflect reality somehow. You do not expect to be required to read any technical articles to understand what this number means.
Counter to Ian's point in that article, I argue that the better solution would be to bury the TDP number. Replace it with a maximum possible power use number. TDP should not be used for very much--basically only for knowing the proper minimum cooling system needed to run as specs. The TDP should thus be buried to only where the few system designers that need to know the TDP will find it.
 
  • Like
Reactions: Zucker2k and MangoX

zir_blazer

Golden Member
Jun 6, 2013
1,158
399
136
So what is the point of the TDP if everyone is going to ignore it? As if a sustained power consumption of 200W+ doesn't require far heavier cooling plus a bigger Power Supply to begin with.

I would love to see benchmarks with these Processors capped to the same TDP than Ryzens, albeit the results are obvious cause they would be completely stomped.
 
  • Like
Reactions: Tlh97

dullard

Elite Member
May 21, 2001
24,964
3,299
126
So what is the point of the TDP if everyone is going to ignore it? As if a sustained power consumption of 200W+ doesn't require far heavier cooling plus a bigger Power Supply to begin with.

I would love to see benchmarks with these Processors capped to the same TDP than Ryzens, albeit the results are obvious cause they would be completely stomped.
Use a cooling system below TDP and you have a major problem on your hand. That is the point. It is the amount of power that at a minimum you need to design your system around. That is why it is called "thermal design power". It is not maximum power. It is not peak power. It is the minimum power to design your system for.

You are correct that in short-term bursty tests, Ryzen would stomp Intel if Intel's turbo mode is castrated.
 

blckgrffn

Diamond Member
May 1, 2003
9,098
2,995
136
www.teamjuchems.com
I was surprised at the lack of overall quality of the article, really. I mean, there was a throw away comment "if an OEM actually adheres to the specified TDP for the chip you'll get lower performance" but then no charts showing the difference?

So if Intel chips have the same PL1 and PL2 set the worse binned chip will use more power! Woah. Shut the front door! :p

I want to know how it works if PL1 is really set to 65W, how long between Turbo bursts? What do we lose if we set the 10700 to a Ninth Gen PL2 (like 125 or 150W)? If we are tuning systems for long life, are we giving up tiny gains to get lower power usage (like how nvidia tuned 3090's out of the gate)?

I wasn't sure what new this added other than giving us Ian's journey to discovering what other sites had already posted.
 

blckgrffn

Diamond Member
May 1, 2003
9,098
2,995
136
www.teamjuchems.com
So what is the point of the TDP if everyone is going to ignore it? As if a sustained power consumption of 200W+ doesn't require far heavier cooling plus a bigger Power Supply to begin with.

I would love to see benchmarks with these Processors capped to the same TDP than Ryzens, albeit the results are obvious cause they would be completely stomped.

But would they really? I set my daughters 9700k PL1 to 88W just for grins, PL2 is 115W and the thing still boosts to 4.8 ghz single core all the time when launching apps, booting, etc. Feels very responsive.

(I know they would win even bigger, but how much bigger? That's what Anandtech should tell us IMO)
 

dullard

Elite Member
May 21, 2001
24,964
3,299
126
I wasn't sure what new this added other than giving us Ian's journey to discovering what other sites had already posted.
Actually, it was Ian's 4th (maybe more?) article on the same subject:



 

Kocicak

Senior member
Jan 17, 2019
982
973
136
Compare it to MPG. You get the (optimistic) figures for different use scenarios which reflect reality. You never get MGP VALID ONLY when the car travels 20 MPH. Because nobody rides that slow. And nobody manually clocks the CPU to run at the advertised base clock.

The car industry have no problem to come up with the numbers to inform customers about the real property of their products. Why should not the CPU industry do the same thing?

AMD 3000 series CPUs had advertised boost clock which told you nothing, because some CPUs never reached them. Advertised boost clocks of the Ryzen 5000 MEAN SOMETHING now, because the CPUs can actually do some useful work at that (or even higher!) frequency.