What's the point of TDP?

[17030644]

I've found out that TDP does not correlate to power consumption or temperatures.

 

Because two CPUs that are completely different such as a 4 and a 10 core CPU from Intel can be rated at the same TDP and heat output depends on how good the cooling is.

 

CPUs or GPUs that draw more power but have better cooling solutions and TIMs run cooler

 

so what is tdp good for?

[stealth80]

No, heat output has nothing to do with cooling. A given CPU consuming the same voltage and running the same clocks and work load will out put the same heat (slight variations binning etc) which would be the TDP. It is the efficiency of the cooling that dictates the actual CPU temp. 

[AlTech]

3 minutes ago, 17030644 said:

I've found out that TDP does not correlate to power consumption or temperatures.

 

Because two CPUs that are completely different such as a 4 and a 10 core CPU from Intel can be rated at the same TDP and heat output depends on how good the cooling is.

 

CPUs or GPUs that draw more power but have better cooling solutions and TIMs run cooler

 

so what is tdp good for?

TDP is meant to show how much thermal output a product may produce and this loosely correlates to power consumption.

[Guest]

4 minutes ago, 17030644 said:

TDP does not correlate to power consumption

 

The point of TDP is to know roughly how much power the cpu/gpu whatever draws, so you known how big of a psu you need.

[tikker]

It gives you a measure of how much heat it will output under load, based on which you can select a suitable cooler capable of dissipating that.

[RollTime]

TDP has nothing to do with power consumption. It describes how many watts of heat the CPU puts out.

[Jurrunio]

TDP is meant to give buyers a rough idea on how much cooling they need for the GPU/GPU whatever to run at factory settings. On graphics cards both AMD and Nvidia use % of TDP to represent power draw, but that doesnt represent the actual heat output.

[Glenwing]

2 minutes ago, RollTime said:

TDP has nothing to do with power consumption. It describes how many watts of heat the CPU puts out.

Those are the same thing. CPU Heat output = CPU power draw. ch heat

[genexis_x]

In CPUs: Guaranteed base clock speeds at TDP limit

[RollTime]

11 minutes ago, Glenwing said:

Those are the same thing. CPU Heat output = CPU power draw. ch heat

Not necessarily. TDP in watts is only the electrons that are stopped in the transistors. Long story short, they can be similar but they are NOT the same thing.

[Glenwing]

7 minutes ago, RollTime said:

Not necessarily. TDP in watts is only the electrons that are stopped in the transistors. Long story short, they can be similar but they are NOT the same thing.

Maybe you can differentiate them at the quantum scale but for all intents and purposes they are the same thing, certainly at the scale we're talking about here.

[Mira Yurizaki]

1 hour ago, 17030644 said:

I've found out that TDP does not correlate to power consumption or temperatures.

 

Because two CPUs that are completely different such as a 4 and a 10 core CPU from Intel can be rated at the same TDP and heat output depends on how good the cooling is.

 

CPUs or GPUs that draw more power but have better cooling solutions and TIMs run cooler

 

so what is tdp good for?

It's just that, a blanket rating. It's probably to ease the design of people who make coolers and if anything, overestimating how much heat you need to dissipate is better than underestimating or even being just at the cusp of it.

 

Also you're correct that TDP does not correlate to temperature, but given the same cooler, a part with a higher TDP will run hotter if the cooler's performance mode doesn't change. Or it has to work harder to achieve the same temperature.

[RollTime]

50 minutes ago, Glenwing said:

Maybe you can differentiate them at the quantum scale but for all intents and purposes they are the same thing, certainly at the scale we're talking about here.

Fair enough.

[17030644]

2 hours ago, stealth80 said:

No, heat output has nothing to do with cooling. A given CPU consuming the same voltage and running the same clocks and work load will out put the same heat (slight variations binning etc) which would be the TDP. It is the efficiency of the cooling that dictates the actual CPU temp. 

And how does it differ to what I just said? I said that depending on cooling temps can be lower or higher one example is intel cpus that despite having a lower TDP than some AMD cpus run hotter because of the poor TIM on them. I also mentioned that two cpus such as ryzen 3 1200 differ in cache size, core count and clocks but still they are given the exact same TDP.

 

2 hours ago, AluminiumTech said:

TDP is meant to show how much thermal output a product may produce and this loosely correlates to power consumption.

 

2 hours ago, Some Random Member said:

The point of TDP is to know roughly how much power the cpu/gpu whatever draws, so you known how big of a psu you need.

 

2 hours ago, RollTime said:

TDP has nothing to do with power consumption. It describes how many watts of heat the CPU puts out.

2 hours ago, tikker said:

It gives you a measure of how much heat it will output under load, based on which you can select a suitable cooler capable of dissipating that.

But two cpus like I said can have very different specs how could they have the same thermal output under load?

 

2 hours ago, Jurrunio said:

TDP is meant to give buyers a rough idea on how much cooling they need for the GPU/GPU whatever to run at factory settings. On graphics cards both AMD and Nvidia use % of TDP to represent power draw, but that doesnt represent the actual heat output.

 

 

2 hours ago, Glenwing said:

Those are the same thing. CPU Heat output = CPU power draw. ch heat

Could you both please explain?

 

2 hours ago, ZM Fong said:

In CPUs: Guaranteed base clock speeds at TDP limit

But rarely do CPUs operate at their base speeds, intel turbo boost or amd turbo core usually makes them run above the rated speeds. And even still let's take the i5 8400 and the i7 8700, they both have a 65W TDP, yet the 8700 is clocked at 3.2 GHz base vs 2.8 on the i5 and has hyperthreading

 

2 hours ago, RollTime said:

Not necessarily. TDP in watts is only the electrons that are stopped in the transistors. Long story short, they can be similar but they are NOT the same thing.

1 hour ago, Glenwing said:

Maybe you can differentiate them at the quantum scale but for all intents and purposes they are the same thing, certainly at the scale we're talking about here.

I would like the long answer please or just redirect me to an article I want to know.

 

1 hour ago, M.Yurizaki said:

It's just that, a blanket rating. It's probably to ease the design of people who make coolers and if anything, overestimating how much heat you need to dissipate is better than underestimating or even being just at the cusp of it.

 

Also you're correct that TDP does not correlate to temperature, but given the same cooler, a part with a higher TDP will run hotter if the cooler's performance mode doesn't change. Or it has to work harder to achieve the same temperature.

That makes more sense, to make it easier for new users to have an idea of what kind of CPU they are dealing with

 

But what if I have a part with a higher TDP that is almost identical, like the 7700K and the 7740X, the 7740X is a tiny bit faster (by 100 mhz I think) and has no iGP yet it has a 112W TDP as opposed to 91W on the 7700K how is it that much hotter or more power hungry according to its TDP?

 

 

[17030644]

Sorry I double posted by mistake

[stealth80]

8 minutes ago, 17030644 said:

And how does it differ to what I just said? 

 

 

2 hours ago, 17030644 said:

Because two CPUs that are completely different such as a 4 and a 10 core CPU from Intel can be rated at the same TDP and heat output depends on how good the cooling is.

Heat output is always the same, the temp of the CPU depends on the efficiency of the cooling applied, that includes TIM, Paste, heatsink/block etc, the heatput is a constant at a given voltage/workload

[17030644]

5 minutes ago, stealth80 said:

 

Heat output is always the same, the temp of the CPU depends on the efficiency of the cooling applied, that includes TIM, Paste, heatsink/block etc, the heatput is a constant at a given voltage/workload

OK I see my mistake thanks I didn't want to mean that, heat output is the same that's right the cooling is what changes

[tikker]

8 minutes ago, 17030644 said:

And how does it differ to what I just said? I said that depending on cooling temps can be lower or higher one example is intel cpus that despite having a lower TDP than some AMD cpus run hotter because of the poor TIM on them. I also mentioned that two cpus such as ryzen 3 1200 differ in cache size, core count and clocks but still they are given the exact same TDP.

 

 

 

But two cpus like I said can have very different specs how could they have the same thermal output under load?

 

 

 

Could you both please explain?

 

But rarely do CPUs operate at their base speeds, intel turbo boost or amd turbo core usually makes them run above the rated speeds. And even still let's take the i5 8400 and the i7 8700, they both have a 65W TDP, yet the 8700 is clocked at 3.2 GHz base vs 2.8 on the i5 and has hyperthreading

 

I would like the long answer please or just redirect me to an article I want to know.

 

That makes more sense, to make it easier for new users to have an idea of what kind of CPU they are dealing with

 

But what if I have a part with a higher TDP that is almost identical, like the 7700K and the 7740X, the 7740X is a tiny bit faster (by 100 mhz I think) and has no iGP yet it has a 112W TDP as opposed to 91W on the 7700K

 

 

It runs 100 MHz faster. It also has a 100 MHz higher all-core turbo. Maybe (even though Intel defines the TDP at base frequency) the voltage (and hence TDP) needs to be higher to guarantee a stable turbo.

[Mira Yurizaki]

21 minutes ago, 17030644 said:

But what if I have a part with a higher TDP that is almost identical, like the 7700K and the 7740X, the 7740X is a tiny bit faster (by 100 mhz I think) and has no iGP yet it has a 112W TDP as opposed to 91W on the 7700K how is it that much hotter or more power hungry according to its TDP?

The 7740X runs 300MHz hotter, but it's not enough to really bump up the TDP rating. It was probably something Intel did just to do.

[17030644]

4 minutes ago, M.Yurizaki said:

The 7740X runs 300MHz hotter, but it's not enough to really bump up the TDP rating. It was probably something Intel did just to do.

I think 7740X is 4.5 ghz across all cores and 7700K is 4.4 so about the same

 

Alright thanks so the conclusion is: TDP doesn't matter at all. It sometimes hits the nail but very rarely

 

[Mira Yurizaki]

Just now, 17030644 said:

I think 7740X is 4.5 ghz across all cores and 7700K is 4.4 so about the same

 

Alright thanks so the conclusion is: TDP doesn't matter at all. It sometimes hits the nail but very rarely

 

I must've looked at the wrong part. Again.

 

Well, TDP matters in that it tells you what to expect heat production wise. A 95W TDP processor is very likely going to run cooler than a 112W TDP processor. Unless you really want to go through the trouble of measuring it yourself, you can at least make a ballpark estimate of how much cooling you need. This is important if you have say a server farm and you want to cut down on your HVAC bills.

[17030644]

Just now, M.Yurizaki said:

I must've looked at the wrong part. Again.

 

Well, TDP matters in that it tells you what to expect heat production wise. A 95W TDP processor is very likely going to run cooler than a 112W TDP processor. Unless you really want to go through the trouble of measuring it yourself, you can at least make a ballpark estimate of how much cooling you need. This is important if you have say a server farm and you want to cut down on your HVAC bills.

What do you employ for measuring power? Isn't that stuff like thousands of dollars? Or are software readings accurate?

[Mira Yurizaki]

2 minutes ago, 17030644 said:

What do you employ for measuring power? Isn't that stuff like thousands of dollars? Or are software readings accurate?

It depends on how accurate you want it. You can use a Kill-A-Watt (which costs like $30) to get system power consumption. Or you can break out an O-Scope and a current probe that is likely to run hundreds of dollars to measure just the CPU itself.

[17030644]

2 minutes ago, M.Yurizaki said:

It depends on how accurate you want it. You can use a Kill-A-Watt (which costs like $30) to get system power consumption. Or you can break out an O-Scope and a current probe that is likely to run hundreds of dollars to measure just the CPU itself.

you mean oscilloscope right? Don't you need more equipement for it or you can run it by itself?

 

and there's the kill-a-watt but that doesn't take into consideration energy losses from the psu or does it?

[Mira Yurizaki]

4 minutes ago, 17030644 said:

you mean oscilloscope right? Don't you need more equipement for it or you can run it by itself?

That's why I mentioned you need a current probe.

4 minutes ago, 17030644 said:

and there's the kill-a-watt but that doesn't take into consideration energy losses from the psu or does it?

It doesn't take into account of energy losses, you plug your PC into it and plug it into a socket.