Intel lying about their CPUs' TDP: who's not surprised?

[cst1992]

I know that, but at least I know in the case of the card how much power it'll consume, so that I plan my build accordingly.
Even PCPartPicker uses the TDP value in their PSU calculations(and yes, you're right - the NVIDIA value is board power draw).
IMO it's just a shitty marketing move to define power consumption to base frequency and some arbitrary lab-only load and use that in spec sheets.
I guess we'll have to turn to reviews for PSU calculations...

 

[RandallFlagg]

I know that, but at least I know in the case of the card how much power it'll consume, so that I plan my build accordingly.
Even PCPartPicker uses the TDP value in their PSU calculations(and yes, you're right - the NVIDIA value is board power draw).
IMO it's just a shitty marketing move to define power consumption to base frequency and some arbitrary lab-only load and use that in spec sheets.
I guess we'll have to turn to reviews for PSU calculations...

Most people do not worry because they do not build their own rig, and if they do they just get a mfr rec spec psu.

I left the lights on in my bathroom this morning. There are 8 x 60W incandescent bulbs in there. By my calc, thats 4hrs x 60W x 8 bulbs = 1920W-Hr I wasted. That is about like the little 17W max load difference between Intel and AMD running Prime95 for 113 hours straight or 14.1 8-hour days.

This is not something sane people worry about.

 

[cst1992]

Most people do not worry because they do not build their own rig, and if they do they just get a mfr rec spec psu.

I left the lights on in my bathroom this morning. There are 8 x 60W incandescent bulbs in there. By my calc, thats 4hrs x 60W x 8 bulbs = 1920W-Hr I wasted. That is about like the little 17W max load difference between Intel and AMD running Prime95 for 113 hours straight or 14.1 8-hour days.

This is not something sane people worry about.

I just don't know what to say to you right now.

 

[Caring1]

This is not something sane people worry about.

And yet you took the time to not only think about it, but also do the math.

 

[ThrashZone]

I know that, but at least I know in the case of the card how much power it'll consume, so that I plan my build accordingly.
Even PCPartPicker uses the TDP value in their PSU calculations(and yes, you're right - the NVIDIA value is board power draw).
IMO it's just a shitty marketing move to define power consumption to base frequency and some arbitrary lab-only load and use that in spec sheets.
I guess we'll have to turn to reviews for PSU calculations...

Hi,
Doubt many would believe it and just say it's all 200w overkill but here's Intel's psu chart

 

[95Viper]

Stop the fanboy BS.
Stay on topic.
And, if you wish to have a personal argument with someone... take it to PMs; and, not in the thread.

Thank You and Have a Great Discussion.

 

[trickson]

Honestly no one really buys a CPU biased on it's TDP ( No one means general public). And It's been said Intel isn't lying nor is AMD.
Honestly I no longer see the point here other than to bash a company (Which is against the rules I Think) .. Maybe Not Maybe I am wrong.

 

[Bill_Bright]

Most people do not worry because they do not build their own rig, and if they do they just get a mfr rec spec psu.

I agree that most people don't build their own and instead buy factory built systems, I disagree that most builders "just get a manufacturer recommended spec PSU". Frankly, I don't even know what that means.

What manufacturer? Motherboard? GPU? CPU? PSU?

In an attempt to heed 95Viper's guidance and get back on track (this thread is about Intel), I looked at two Intel and two AMD processors. No PSU recommendations there. And how could they? They don't know which motherboard or in particular, which graphics solution (plus all the other devices) that PSU will need to support.

I would contend while every self-builder was a first time builder at some point, most have more than one build (or major upgrade) under their belts and have learned a thing or two about picking brands and sizing up PSUs - either by doing their homework with their first build, or if not, learning from their first build mistakes.

I've been helping folks research parts for many years. Very few visit the actual PSU makers sites. So again, not sure what manufacturer you speak of. I rarely visit PSU makers sites when researching PSUs. I visit review sites and retailers. I want the facts, not the marketing fluff from the maker. And if I even look at the TDP, it is to get an idea of my cooling requirements, not PSU size.

If the builder knows enough to pick a PSU brand and use their site to determine a recommended size, that's not a bad thing. But if they know enough to build their own computer, chances are the majority are going to do their homework and seek out advice for their first build or two until they get some experience under their belts.

NO DOUBT there a few builders out there who fail to do their homework and simply buy the cheapest PSU they can find. Typically they simply guess at the size or arbitrarily pull a number out of thin air (650W is pretty popular). But that certainly is not "most" self-builders.

 

[kapone32]

Just leaving some information here...

The only time the Intel rig drew more power was under artificial load like Prime95. Under gaming, single thread load, normal multi-thread load, and idle the 9900K drew less power than the 3700X.

So if your primary use case is running Prime95 AMD is definitely your best bet.


View attachment 187435

What about the 10900K? Why do Z590 boards have VRMs that rival Threadripper? Yesterday on The Full Nerd a question was asked. Would a 1200 Watt PSU be enough to run to 3090s and a 10900K? The answer was non committal. I have run 2 Vega 64s with a 2920X with no concern on my 1200 Watt PSU.

 

[freeagent]

^^ A setup like that TR uses 87% of a 750w PSU for 94% efficiency, on a 1200w PSU that is a 55% load with 94.7% efficiency

With No OC

Only because I am looking at PSU's right now..

 

[kapone32]

^^ A setup like that TR uses 87% of a 750w PSU for 94% efficiency, on a 1200w PSU that is a 55% load with 94.7% efficiency

With No OC

Only because I am looking at PSU's right now..

Well I seriously doubt that a 750 Watt PSU could handle that load on constant as even 750 is just a burst wattage for plenty of PSUs.

 

[trickson]

I agree that most people don't build their own and instead buy factory built systems, I disagree that most builders "just get a manufacturer recommended spec PSU". Frankly, I don't even know what that means.

What manufacturer? Motherboard? GPU? CPU? PSU?

I think the person meant the manufacture of the prebuilt computer.
HP/ Dell and the like all use "there own" PSU units unless the buyer specifies the PSU.
12 pages and still haven't seen a thing about the lie Intel is pimping.

 

[Bill_Bright]

I think the person meant the manufacture of the prebuilt computer.

But they are not going to recommend a PSU. They are just going to include what they have decided is needed. Even if you choose to customize your Dell, for example, you might be able to swap in a different CPU or add more RAM, but you are still going to get the PSU they decide you need.

Pretty much the only time you can truly select your own PSU is if you have a local shop custom build it for you. The bigger the computer brand (Dell, HP, Acer, Lenovo) the few custom options you are offered.

 

[kapone32]

But they are not going to recommend a PSU. They are just going to include what they have decided is needed. Even if you choose to customize your Dell, for example, you might be able to swap in a different CPU or add more RAM, but you are still going to get the PSU they decide you need.

Pretty much the only time you can truly select your own PSU is if you have a local shop custom build it for you. The bigger the computer brand (Dell, HP, Acer, Lenovo) the few custom options you are offered.

The PSU is usually where the most profit lies for the OEM.

 

[Bill_Bright]

The PSU is usually where the most profit lies for the OEM.

Not sure how that applies to my comment - which was about customers having a choice, or rather limited or no choice.

You may be right but pretty sure I am correct to say the reason PSUs might provide their largest profit margin (at least by percentage) would simply be (1) because they can promise the OEM PSU maker to buy a million or two PSUs in the coming year, then demand and get them at super deep volume discounts. And then (2), because they can use that same model PSU in several different model PCs.

For example, 3 different model computers might require 3 different motherboards, 3 different CPUs, and 3 different cases. But the same model PSU could be used in all 3.

 

[kapone32]

Not sure how that applies to my comment - which was about customers having a choice, or rather limited or no choice.

You may be right but pretty sure I am correct to say the reason PSUs might provide their largest profit margin (at least by percentage) would simply be (1) because they can promise the OEM PSU maker to buy a million or two PSUs in the coming year, then demand and get them at super deep volume discounts. And then (2), because they can use that same model PSU in several different model PCs.

For example, 3 different model computers might require 3 different motherboards, 3 different CPUs, and 3 different cases. But the same model PSU could be used in all 3.

I was supporting your point. You expanded on my thought.

 

[trickson]

This thread is like watching CNN or MSNBC.

 

[freeagent]

Well I seriously doubt that a 750 Watt PSU could handle that load on constant as even 750 is just a burst wattage for plenty of PSUs.

That's why I laughed, I was checking out the BeQuiet PSU Calculator. Seemed a bit optimistic..

It just goes to show even the pro's don't know what they are talking about when it comes to power.. a calculator said a 9900K would only be 30w more than my current one.

In the end, no one will tell you how they formulate their equation, and you should just buy the biggest cooler you can because everyone is lying anyways, because 255w is the new 65w.. here's a shitty piece of aluminum and some screws have fun

 

[cst1992]

just buy the biggest cooler you can

Not an easy choice, considering larger coolers are much more expensive than smaller ones. The price increase and cooling capacity increase is not a linear relationship.

 

[Bill_Bright]

Not an easy choice, considering larger coolers are much more expensive than smaller ones.

It is not just about price. Many cases limit the height of the coolers they can support. And I am not talking about "slim" cases either. Some larger coolers are very tall. On some motherboards, larger coolers could interfere with RAM too. And perhaps larger graphics cards.

The price increase and cooling capacity increase is not a linear relationship.

Setting price aside, bigger does not necessarily mean better cooling efficiency. The number, size and shape of the fins matter. As does the material used in the fins and the baseplate where the heatsink makes contact with the die. Then not all fans are created equal either.

 

[newtekie1]

Well I seriously doubt that a 750 Watt PSU could handle that load on constant as even 750 is just a burst wattage for plenty of PSUs.

Not, a good PSU should be able to handle full load for long periods of time. They days of PSU manufacturers listing peak numbers instead of constant is pretty much gone on the good manufacturer's PSUs thanks to the increase in PSU reviews.

Not an easy choice, considering larger coolers are much more expensive than smaller ones. The price increase and cooling capacity increase is not a linear relationship.

I mean, honestly, not really. You can get a 92mm tower cooler that will handle pretty much any Intel processor at stock settings and fit in pretty much any normal width case for $20. The generic extruded aluminum coolers are like $15. So not really a big difference.

 

[Mouth of Sauron]

To clarify it more, what I said is basically that peak increase in TDP isn't important (and it's my opinion for quite a long time, as said).

Typical desktop CPU draw is like 90-100W since, well, forever. More recent 'gimmicks' which allow CPU to draw additional power for small speed gains means little in overall power usage. They are also pretty much pushed to the limit, and therefore undervolting actually became a thing.

 

[Bill_Bright]

and therefore undervolting actually became a thing.

Undervolting, underclocking, and other similar techniques and "gimmicks" to reduce heat production and buildup in personal computers have been around for nearly 20 years that I know of, if not longer. It has only recently become "a thing" because the kids of today finally started to become aware of the problems of excessive heat in their gaming rigs, and because manufacturers have learned how to "market" undervolting as a "feature" in their products.

Early computer enthusiasts, folks who have been around awhile, were some of the first to build custom PC PVRs and HTPCs (home theater PCs). Computers where "silent running" was an absolute must. So totally "passive" (no fan) cooling was essential. Undervolting and underclocking was commonly done to ensure our fanless systems remained properly cooled without any noisy fans or water pumps making a racket in the background.

And it wasn't child's play like it is today where you can simply enter the BIOS Setup Menu or run a little program, change a setting, reboot and be done. Or where if things go wrong, you just change the setting back or run a little recovery app, reboot and be good to go. Back in the day you had to physically modify the motherboard by tracing circuits and chasing voltages, cutting runs and soldering in jumpers (without schematics, BTW) - then say a couple Hail Mary's, cross your fingers and toes, connect power, boot and pray everything doesn't go up in smoke.

They are also pretty much pushed to the limit

No they aren't. If they were pushed to the limit, there would be no "Turbo" modes. TDP would not be stated at "base" levels. There would be no such thing as overclocking because there would be no headroom to allow it.

What is "a thing" these days is manufacturers have made the process of implementing undervolting "a thing". And they have learned how to market that feature. But undervolting, as a method to reduce heat and increase efficiency, has been around almost as long as Ben Franklin and his key and kite.

 

[trickson]

I personally think they are lying us all.
There is not one mention of AMD having "Unlocked" Multipliers anymore it some how just went to this "Boost" which is IMHO the absolute TDP of the Chip.
First off if you want the boost it's fine it works like AMD says.
You are NOT going to be able to keep that OC ( Unless you want a bricked out CPU ). Like if you really want 4.4GHz on a Ryzen CPU you have to add the Vcore and this chip doesn't like that, This chip wants less power and Less MHZ it's like once it reaches that max turbo boost it get's really HOT really fast and the cooling can NOT keep up no matter what Air cooler you have NO way!
I took the advice of others running at 4.4GHz with 1.4Vcore would ultimately KILL this Chip and ANY other Ryzen CPU.
I see this now in the way this CPU acts with Vcore and heat. backing off to absolute stock settings I see radical voltages and MHz as well as temp's it's one of the most erratic CPU lines I have ever seen. No wonder everyone is all caught up and spinning there wheels! You can not OC the chip without radically changing voltages and that changes the temps so radically that the cooler can not even manage that fast of a heat transfer that the chip gets damaged.
I am not risking it on any of the CPU's. Till AMD can officially come and announce that we can OC once again. Like in the FX Black editions.

So yes in conclusion I would have to say that YES Intel and AMD both are lying about a lot of things!

 

[80-watt Hamster]

I think anyone still following this thread, if they haven't already done so, should look at this post by Zach_01 for Intel's own summary of TDP for their chips, as well as watch (as in the whole thing; I know it's long-ish) the GN video on AMD TDP linked a couple posts later.

Things we've learned over the course of this thread (YMMV):

• TDP does not mean, and has never meant, maximum power consumption
  • It can, however, resemble average power consumption at base frequency, particularly with Intel
• Intel and AMD calculate TDP differently, in ways that don't necessarily produce a useful value for an end user
• Modern turbo and boost strategies can push power consumption well past TDP for short periods
  • Certain computational loads can also drive it higher over longer periods.
• Overclocking completely obviates TDP as a useful value.

My takeaway is this: neither manufacturer is lying about TDP, AFAICT. It's more Thermal Design Power not actually meaning what it sounds like it should, and we DIY-ers latching onto it because it's all there is (outside of reviews and such, of course). Something more meaningful would be nice, but I'm not sure there's a compelling reason for either company to provide it. If they do, it's certainly not going to be on behalf of a "handful" of enthusiasts on forums. I mean, if even the cooler manufacturers are unhappy with it, and the chipmakers won't provide something better for them, it's probably a lost cause.